A mud pump performance analysis method based on numerical simulation and model test

Abstract

This invention relates to the technical field of mud pump performance analysis and discloses a mud pump performance analysis method based on numerical simulation and model testing. The method includes: planning the analysis process, determining the numerical simulation (simulation of the mud pump's internal flow field) and model testing content (measured performance data for different operating conditions); determining the operating condition type based on design parameters and operating conditions; generating a set of performance analysis processes and determining the execution order; generating a list of simulation parameter settings and test plans for each operating condition; and correlating data to generate an overall plan. The method also involves determining the operating condition analysis boundaries, subdividing the operating condition types, dynamically adjusting the process, optimizing test rules, and building a database. Through multi-dimensional data integration and dynamic mechanisms, this method achieves comprehensive and accurate analysis of mud pump performance, improving analysis efficiency and reliability, and is suitable for mud pump performance evaluation and optimization.

Description

Technical Field

[0001] The present invention relates to the technical field of mud pump performance analysis, in particular to a mud pump performance analysis method based on numerical simulation and model test. Background Art

[0002] As a key fluid conveying equipment in mining, metallurgy, dredging and other fields, mud pump performance directly affects the system's operating efficiency, energy consumption and reliability. As industrial production develops towards large-scale and intelligent production, higher requirements are placed on the accurate performance analysis of mud pumps under different working conditions. Traditional mud pump performance analysis methods often have the following limitations:

[0003] On the one hand, analytical models that rely solely on numerical simulation or model testing are unable to fully reflect the true performance of mud pumps. Although numerical simulation alone can visualize the internal flow field of mud pumps, the simulation results are significantly affected by factors such as boundary condition settings and meshing accuracy, and cannot fully reproduce the complex physical phenomena in actual operation. Relying solely on model testing faces the problems of high cost, long cycle time, and limited operating conditions. This is especially true for performance testing of mud pumps under extreme conditions or with new structures, which significantly increases the difficulty and risk of testing.

[0004] On the other hand, existing analysis methods lack scientific validity in terms of operating condition classification and performance evaluation. Traditional operating condition classification is often based on a single flow rate or head metric, failing to comprehensively consider multiple dimensions of the slurry pump, including efficiency, energy consumption, and operating environment. This results in analysis results that fail to accurately reflect the slurry pump's performance under complex, actual operating conditions. Furthermore, the lack of a dynamic adjustment mechanism makes it difficult to adaptively optimize the analysis process and operating condition types based on real-time operating data, limiting the timeliness and practicality of the analysis results.

[0005] Furthermore, traditional methods lack a systematic database construction and maintenance mechanism for data management and long-term performance tracking. Consequently, a large amount of simulation parameters, test data, and performance analysis results are not effectively integrated and utilized, making it difficult to track and compare mud pump performance over the long term. This hinders the discovery of potential patterns in mud pump performance changes and optimization directions.

[0006] With the rapid development of computer and sensor technologies, the accuracy of numerical simulation algorithms has continued to improve, and the automation and data acquisition capabilities of model tests have also been increasingly enhanced, providing technical support for the innovation of mud pump performance analysis methods. How to organically combine numerical simulation and model testing to establish a scientific, efficient, and dynamic mud pump performance analysis method to achieve a comprehensive and accurate assessment of mud pump performance has become a technical problem that needs to be solved in this field. Summary of the Invention

[0007] The purpose of the present invention is to provide a mud pump performance analysis method based on numerical simulation and model test to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a mud pump performance analysis method based on numerical simulation and model test, the method comprising:

[0009] Planning the performance analysis process of the mud pump based on its design parameters and operating conditions, and determining the numerical simulation content and model test content required for the analysis. The numerical simulation content includes simulating the internal flow field of the mud pump, and the model test content includes obtaining measured performance data of the mud pump under different operating conditions.

[0010] Determine different working conditions of the mud pump according to the design parameters and working conditions of the mud pump;

[0011] Generate a mud pump performance analysis process set according to different working condition types, numerical simulation contents and model test contents of the mud pump, and determine the execution order of the performance analysis process set based on the optimal matching principle;

[0012] Generate a simulation parameter setting list for each working condition according to a preset analysis period, different working condition types of the dredge pump, and performance index requirements of the dredge pump under each working condition, and generate a test plan list for each working condition according to preset test rules, different working condition types of the dredge pump, and performance index requirements of the dredge pump under each working condition;

[0013] Based on the execution order of the performance analysis process set, the simulation parameter setting list, the test plan list and the mud pump performance analysis content of each working condition are associated with each other to generate an overall plan for the mud pump performance analysis.

[0014] Preferably, the method plans the performance analysis process of the mud pump based on the design parameters and working conditions of the mud pump, and determines the numerical simulation and model test content required for the analysis, including:

[0015] For a current operating condition, determining a performance analysis boundary for the current operating condition based on historical performance data of the current operating condition, historical performance data of adjacent operating conditions, and a preset operating condition division threshold;

[0016] Extracting a performance analysis data subset of the current working condition according to the performance analysis boundary and a preset time window;

[0017] The content of the numerical simulation and model test related data in the performance analysis data subset is determined as the mud pump performance analysis content of the current working condition.

[0018] Preferably, the method determines different working condition types of the mud pump according to the design parameters and working conditions of the mud pump, including:

[0019] For the current working condition, when the flow rate of the current working condition is greater than a preset flow rate threshold, determining that the current working condition is a high flow rate working condition;

[0020] When the flow rate of the current working condition is not greater than the preset flow rate threshold, classifying the head and efficiency data of the current working condition based on a first preset classification algorithm to obtain performance category data, and determining a central feature of the performance category data;

[0021] determining a performance potential level of the current operating condition based on the working conditions of the current operating condition;

[0022] When the matching degree between the central feature of each performance category data and the preset performance category is greater than a preset matching threshold, determining that the current operating condition is a specific performance operating condition;

[0023] When the matching degree is not greater than the preset matching threshold, the current operating condition is determined to be a normal performance operating condition.

[0024] Preferably, the mud pump performance analysis content also includes mud pump structure data and operating environment data. The different operating conditions of the mud pump include high-efficiency operating conditions, low-efficiency operating conditions and critical operating conditions. According to the different operating conditions of the mud pump, the numerical simulation content and the model test content, a mud pump performance analysis process set is generated, including:

[0025] For the current operating condition, classify the operating environment data of the current operating condition based on a second preset classification algorithm to obtain various operating environment category data; for the current operating environment category data, filter the current operating environment category data based on the influence degree and change frequency of each environmental factor in the current operating environment category data to obtain a key environment subset of the current operating environment category data; and determine an influence label for the current operating environment category data based on the influence characteristics of the key environment subset and the mean value of the mud pump performance index;

[0026] According to whether the working condition type of the current working condition is a high-efficiency working condition;

[0027] Preferably, after determining whether the operating condition type of the current operating condition is a high-efficiency operating condition, the method further includes:

[0028] If the current working condition is a high-efficiency working condition, the impact labels of the operating environment category data are matched with the mud pump performance indicators to generate a targeted optimization strategy;

[0029] If the current working condition is an inefficient working condition, an improvement optimization strategy is generated based on preset improvement rules and performance improvement goals;

[0030] If the operating condition type of the current operating condition is a critical operating condition, a performance stabilization strategy is generated according to historical performance fluctuations and performance stabilization rules.

[0031] Preferably, after generating the mud pump performance analysis process set, the method further includes:

[0032] Updating the mud pump performance analysis content of each working condition based on the real-time mud pump operation data;

[0033] Recalculate the working condition types of each working condition according to the updated mud pump performance analysis content;

[0034] The execution order of the performance analysis process set is dynamically adjusted according to the recalculated working condition type.

[0035] Preferably, the preset test rules include:

[0036] Dynamically adjust the sample size and number of repetitions of the model test based on the confidence interval of the historical mud pump test data;

[0037] When the deviation between the measured performance data and the numerical simulation results exceeds the preset error threshold, the supplementary test module is automatically triggered.

[0038] Preferably, the preset test rules further include:

[0039] Establish a calibration and compensation mechanism for test equipment, and calibrate the test equipment for error compensation based on the benchmark values ​​of standard flow meters and pressure gauges before each test.

[0040] Preferably, after generating the overall solution for the mud pump performance analysis, the method further comprises:

[0041] Establishing a mud pump performance analysis database, storing data such as a simulation parameter setting list, a test plan list, mud pump performance analysis content, and various working condition types in the overall plan into the database;

[0042] Regularly clean and maintain the data in the database to remove invalid and duplicate data;

[0043] Based on this database, the mud pump performance analysis results are tracked and compared over a long period of time to evaluate the changing trend of mud pump performance over time.

[0044] Preferably, the second preset classification algorithm is a hierarchical clustering algorithm, and the specific steps of classifying the operating environment data of the current working condition include:

[0045] Calculate the similarity between samples in the operating environment data;

[0046] The samples are gradually merged into different categories according to their similarity to form a clustering tree;

[0047] According to the preset number of clustering layers or clustering distance threshold, appropriate categories are intercepted from the clustering tree as the category data of each operating environment.

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] At the analytical process planning level, by combining mud pump design parameters with operating conditions, we systematically plan the content of numerical simulations and model tests, clarifying the analysis focus for different operating conditions. For example, for the current operating condition, we determine the performance analysis boundaries based on historical performance data, adjacent operating condition data, and preset thresholds. We then extract targeted data subsets to ensure that the analysis focuses on key issues, avoid redundant calculations and ineffective testing, and effectively improve analysis efficiency. Furthermore, we generate a set of performance analysis processes and determine their execution order based on the principle of optimal matching, enabling scientific scheduling of simulation and testing processes and reducing process conflicts and resource waste.

[0050] In terms of working condition classification and performance evaluation, the innovative working condition classification method breaks through the limitations of traditional single indicator classification. The flow threshold is used to preliminarily distinguish large flow conditions, and then the head and efficiency data are classified using the first preset classification algorithm. Combined with the performance potential level and matching judgment, the working conditions are subdivided into specific types such as high efficiency, low efficiency, and critical. This multi-dimensional classification method can more accurately reveal the performance characteristics of mud pumps under different operating conditions. For example, for high-efficiency working conditions, by matching the operating environment impact labels with performance indicators to generate optimization strategies, the potential of the equipment can be further explored; for inefficient working conditions, improvement strategies are formulated based on preset rules and improvement goals to solve performance shortcomings in a targeted manner; for critical working conditions, stable strategies are generated based on historical fluctuations and stability rules to effectively prevent the risk of performance mutations.

[0051] In terms of dynamic data management and process optimization, a dynamic adjustment mechanism driven by real-time data has been established. By updating analysis content, recalculating operating condition types, and adjusting process execution sequences using real-time operating data, the analysis method can adapt to changes in the mud pump's operating status. For example, when the deviation between measured data and simulation results exceeds a threshold, a supplementary test module is automatically triggered to ensure data reliability. Dynamic adjustments to sample size and repetition times are made based on the confidence intervals of historical test data, ensuring test accuracy while reducing costs. Furthermore, a test equipment calibration compensation mechanism improves test data accuracy by calibrating errors using standard instrument benchmark values.

[0052] In terms of data storage and long-term tracking, a comprehensive performance analysis database has been constructed, enabling systematic storage and management of data such as simulation parameters, test plans, analysis content, and operating condition types. Regular cleaning of invalid and duplicate data ensures database efficiency and availability. Long-term tracking and comparative analysis based on the database clearly reveals performance trends over time for mud pumps, providing a scientific basis for equipment maintenance and upgrades. For example, by comparing performance data from different periods, signs of equipment degradation can be detected in advance, allowing preventive maintenance plans to be developed and reducing the risk of downtime.

[0053] In terms of technology integration and application expansion, the advantages of numerical simulation in flow field analysis are combined with the reliability of measured data from model testing to form a complementary analysis system. Numerical simulation can predict the internal flow characteristics of mud pumps under different operating conditions in advance, providing guidance for test design; model testing provides verification data for simulation results, corrects simulation parameters, and improves simulation accuracy. This integrated model is not only applicable to the performance analysis of traditional mud pumps, but also provides an efficient analytical method for the development of new mud pumps, significantly shortening the development cycle and reducing R&D costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a working principle diagram of the mud pump performance analysis method based on numerical simulation and model test according to the present invention;

[0055] Figure 2 Design drawings for determining the type of working condition;

[0056] Figure 3 Design diagram for classification of operating environment data;

[0057] Figure 4 Design diagrams generated for the operating condition optimization strategy;

[0058] Figure 5 This is the design diagram of the mud pump performance analysis database. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] See also Figure 1-Figure 5 The present invention relates to a mud pump performance analysis method based on numerical simulation and model test, and the specific implementation steps are as follows:

[0061] Based on the mud pump's design parameters (such as impeller diameter, speed, rated flow rate, etc.) and operating conditions (such as medium density, delivery pressure, operating ambient temperature, etc.), the mud pump's performance analysis process is systematically planned. First, the required numerical simulation content and model test content are clearly analyzed. The numerical simulation content focuses on the simulation of the mud pump's internal flow field. The fluid flow state in key components such as the mud pump impeller and volute is modeled using computational fluid dynamics (CFD) software, and parameters such as flow velocity distribution, pressure field, and turbulence characteristics are analyzed. The model test content includes obtaining the measured performance data of the mud pump under different operating conditions, such as flow rate, head, efficiency, power consumption, etc., by building a physical test bench and configuring sensors (such as electromagnetic flowmeters, pressure transmitters, power meters, etc.) to collect data in real time.

[0062] Identify and categorize the different operating conditions of the dredge pump based on its design parameters and operating conditions. For example, determine typical operating condition categories for different flow ranges, media characteristics, or operating loads, providing clear guidance for subsequent simulations and tests.

[0063] A performance analysis process set for mud pumps is generated by combining different operating conditions, numerical simulation content, and model testing. This process set covers the specific simulation and testing steps, data collection nodes, and analysis methods for each operating condition. Furthermore, the execution order of the performance analysis process set is determined based on the optimal matching principle (e.g., ordering the operating conditions from low to high complexity and prioritizing data dependencies) to improve analysis efficiency and data relevance.

[0064] Based on the preset analysis cycle (e.g., quarterly, annual, or specific project cycle), different mud pump operating conditions, and mud pump performance indicators (e.g., industry standards or internal control indicators), a list of corresponding simulation parameter settings is generated for each operating condition. For example, under high-flow conditions, the grid density, turbulence model parameters, and boundary conditions (inlet flow rate, outlet pressure) for CFD simulation are set. At the same time, based on the preset test rules (e.g., test sample size calculation method, repetition test requirements, etc.), operating condition type, and performance indicator requirements, a list of test plans for each operating condition is generated, including test equipment configuration, test point distribution, data collection frequency, etc.

[0065] Based on the execution sequence of the performance analysis process set, the simulation parameter setting list for each operating condition, the test plan list, and the mud pump performance analysis content (such as flow field analysis objectives and performance data comparison dimensions) are linked and integrated to form an overall mud pump performance analysis plan. This plan is presented in the form of a flow chart or Gantt chart, clearly defining the time nodes, responsible parties, and data transmission paths for each link to ensure the standardization and traceability of the analysis process.

[0066] Example 1:

[0067] For the current operating condition, the performance analysis boundaries must be determined based on historical performance data, adjacent operating condition data, and preset operating condition classification thresholds. Historical performance data includes various parameter records of the current operating condition during past operating cycles, such as flow rate, head, efficiency, power, medium temperature, and vibration frequency. This data is typically stored in the mud pump operation monitoring system or historical database and can be retrieved in real time or imported in batches through a data interface. Historical performance data for adjacent operating conditions refers to other operating condition data that is consistent with the current operating condition in operating condition parameters (such as flow rate and pressure). For example, if the current operating condition is the rated flow condition, the adjacent operating conditions can be set as low flow conditions 10% below the rated flow and medium-to-high flow conditions 10% above the rated flow. The data range covers all operating records with parameter differences from the current operating condition within the preset range. The preset operating condition classification threshold is a pre-set allowable range of parameter fluctuations based on mud pump design standards and industry specifications. For example, the flow threshold can be set to ±10% of the rated flow, and the pressure threshold can be set to ±15% of the rated pressure. The specific values ​​need to be determined based on factors such as mud pump type and application scenario.

[0068] When determining the performance analysis boundaries, first conduct a statistical analysis of the historical performance data for the current operating condition. Descriptive statistics such as the mean, standard deviation, maximum, and minimum values ​​of key parameters (such as flow rate and head) are calculated to identify data distribution characteristics. Then, the historical performance data of adjacent operating conditions is compared with the current operating condition data to analyze parameter change trends. For example, if the current operating condition is a high-flow condition, it is necessary to determine whether the flow rate is continuously above the preset flow threshold or whether it occasionally reaches this threshold within a specific time period. By comparing the parameter differences between the current operating condition and the adjacent operating condition and combining them with the preset operating condition classification threshold, the performance analysis boundary for the current operating condition is determined. Specifically, if the mean flow rate of the current operating condition is Q0 and the preset flow threshold is ±ΔQ, the performance analysis boundary can be defined as the flow interval [Q0-ΔQ, Q0+ΔQ]. All historical data within this interval is included in the analysis. Data outside this interval is considered to belong to adjacent operating conditions or abnormal data and requires separate processing.

[0069] Next, a subset of performance analysis data for the current operating condition is extracted based on the performance analysis boundary and the preset time window. The preset time window ensures the timeliness and relevance of the analysis data and can be set based on the mud pump's operating cycle, maintenance cycle, or project requirements, such as the last three months, the last six months, or the last 100 operating hours. The time window setting must balance the richness and timeliness of the data, avoiding excessive data inclusion due to a long time window or insufficient data due to a short time window.

[0070] When extracting data subsets, historical data that meets the operating parameter range is first filtered according to the performance analysis boundary. Then, further filtering is performed according to the preset time window, retaining only the data within the time window. For example, if the performance analysis boundary is the flow rate interval [Q0-ΔQ, Q0+ΔQ] and the preset time window is from January 1, 2025, to March 31, 2025, the data subset consists of all operating records within this time period with flow rates within the range [Q0-ΔQ, Q0+ΔQ]. The data subset includes both numerical simulation data and model test data. The numerical simulation data includes flow field parameters calculated using CFD software, such as velocity, pressure, turbulence intensity, and vorticity at each computational node, and is typically stored as grid data files, data tables, or visualization graphics (such as velocity vector diagrams and pressure contour maps). The model test data includes measured data collected from the physical test bench, such as flow rate, head, efficiency, and power consumption at different test points. These data are collected in real time by sensors and stored as time series data or test reports.

[0071] When extracting data subsets, attention must be paid to data integrity and consistency. Missing data should be supplemented using interpolation methods (such as linear interpolation and polynomial interpolation) or data filling algorithms. Anomalous data (such as jump values ​​caused by sensor failure) should be removed or corrected using data cleaning algorithms (such as sliding average filtering and median filtering). Furthermore, the timestamps of numerical simulation data and model test data must be aligned to ensure temporal correspondence between simulation data and test data under the same operating conditions, facilitating subsequent comparative analysis and correlation modeling.

[0072] After extracting the data subset, the content of the numerical simulation and model test related data within the performance analysis data subset needs to be determined as the mud pump performance analysis content for the current operating conditions. Specifically, the numerical simulation analysis content includes: the overall flow characteristics of the mud pump's internal flow field, such as whether the velocity distribution at the impeller inlet and outlet is uniform and whether there are vortices or backflow phenomena in the volute; the flow parameters of key components, such as the pressure distribution on the blade surface, boundary layer thickness, and prediction of the cavitation occurrence area; the correlation analysis between flow field parameters and mud pump performance indicators, such as the impact of flow velocity distribution on head and the impact of turbulence intensity on efficiency loss. The model test analysis content includes: measured performance curves under different test conditions (such as flow-head curves and flow-efficiency curves) to verify the accuracy of the numerical simulation results; the fluctuation range and stability of various performance indicators, such as the maximum deviation of head and the standard deviation of efficiency; comparative analysis of test data and design parameters to identify the causes of performance differences, such as manufacturing errors, installation deviations, or changes in medium characteristics.

[0073] When determining the analysis content, it's necessary to prioritize the design objectives and actual application requirements of the mud pump, clarifying the analysis focus. For example, if the mud pump frequently experiences reduced efficiency during operation, the analysis should focus on energy loss areas in the flow field and efficiency-influencing factors in the test data. If excessive vibration is present, the pulsating pressure distribution in the flow field and the coupling relationship between the vibration frequency in the test data and the characteristic frequency of the flow field should be analyzed. Furthermore, a mechanism for prioritizing analysis content should be established. Based on the severity and scope of the issue, key performance indicators (such as head and efficiency) should be analyzed first, followed by secondary indicators (such as vibration and noise) to ensure an efficient and targeted analysis process.

[0074] To ensure traceability and repeatability of the analysis process, detailed records of the performance analysis boundary determination process, data subset extraction methods, and analysis content selection criteria must be maintained to form an analysis log. This log should include: data source (e.g., database name, test report number), parameter settings (e.g., preset operating condition thresholds, time window start and end times), data processing methods (e.g., interpolation and filtering algorithms), and analysis content lists. This analysis log serves as a reference for subsequent performance analysis and can also be used to review and verify the rationality and accuracy of the analysis process.

[0075] In summary, by defining the performance analysis boundaries for the current operating conditions, extracting targeted data subsets, and determining the specific analysis content, we provide a solid data foundation and clear analysis direction for mud pump performance analysis. This ensures that subsequent numerical simulations and model tests can be closely centered around the actual needs of the current operating conditions, improving the accuracy and practicality of the analysis results. This process fully considers the temporal and spatial characteristics and physical significance of the data, and through scientific methods of data screening and processing, avoids interference from irrelevant data, thereby improving analysis efficiency and quality.

[0076] Example 2:

[0077] When the current operating flow exceeds a preset flow threshold (e.g., 120% of the rated flow), the pump is directly classified as operating at high flow. This threshold is a pre-set critical value based on the pump's design characteristics and safe operating range, and can be determined by consulting a pump design manual or industry standards. If the current flow does not exceed this threshold, the pump head and efficiency data are classified using a first preset classification algorithm.

[0078] The first preset classification algorithm uses the K-means clustering algorithm, which iteratively divides data points into clusters, maximizing the similarity of data points within a cluster and minimizing the similarity between clusters. Before applying this algorithm, the historical head and efficiency data for the current operating conditions must be preprocessed. First, the head and efficiency parameters are extracted from the historical database, and obvious outliers (such as extreme values ​​caused by sensor failure) are removed. Then, the data is normalized to eliminate dimensionality effects. The formula is:

[0079]

[0080] Among them, x is the original data, and x′ is the normalized data.

[0081] Initialize the parameters of the K-means algorithm, including the number of clusters k (usually set based on experience or business needs, such as 3-5 clusters), the maximum number of iterations (such as 100), and the convergence threshold (such as 0.001). Randomly select k data points as the initial cluster centers and calculate the Euclidean distance from each data point to each cluster center:

[0082]

[0083] Where x is the data point, c j is the jth cluster center, c ji is the value of the i-th dimension representing the j-th cluster center, and n is the data dimension (here it is 2, i.e., head and efficiency).

[0084] Each data point is assigned to the cluster with the closest cluster center, and the centroid of each cluster is recalculated as the new cluster center. This process is repeated until the change in cluster center is less than the convergence threshold or the maximum number of iterations is reached. Ultimately, k performance categories are obtained, such as high-lift high-efficiency, medium-lift medium-efficiency, and low-lift low-efficiency.

[0085] Calculate the central characteristics of each performance category data, including statistics such as mean, variance, and covariance. The mean vector μ reflects the central position of the data in this category, and the calculation formula is:

[0086]

[0087] Where N is the number of data points of this type, x i is the i-th data point.

[0088] The variance matrix ∑ describes the degree of discreteness of the data distribution, and the calculation formula is:

[0089]

[0090] The performance potential level of the current operating conditions is determined based on the performance potential evaluation model in the dredge pump design manual, taking into account the current operating conditions. These operating conditions include factors such as medium characteristics (such as viscosity and sand content), operating environment (such as temperature and pressure), and pump speed. The performance potential evaluation model, based on fluid mechanics principles and empirical formulas, comprehensively considers the impact of these factors on dredge pump performance. For example, for high-viscosity media, the model reduces the pump's theoretical efficiency potential based on a viscosity correction factor. For high-temperature environments, the model also considers the limitations of the liquid vapor pressure on the pump's cavitation performance.

[0091] The current operating condition parameters are input into the performance potential assessment model, and a performance potential score is calculated through a multi-dimensional nonlinear mapping relationship. This score is normalized to a range of 0-100, and the performance potential level of the current operating condition is determined based on preset grading thresholds (e.g., 0-30 points for level 1 potential, 31-60 points for level 2 potential, and 61-100 points for level 3 potential).

[0092] Subsequently, the matching degree of the central features of each performance category data and the standard features of the preset performance category is calculated. The standard features of the preset performance category are derived from mud pump design specifications and industry standards. For example, the standard features of high-efficiency working conditions are a head fluctuation range of ±5% and an efficiency greater than 90% of the rated value. The matching degree calculation uses the cosine similarity algorithm:

[0093]

[0094] Among them, A and B are the performance category data center feature vector and the preset standard feature vector respectively.

[0095] If the matching degree exceeds a preset matching threshold (e.g., 0.8), the current operating condition is determined to be a specific performance condition, such as a high-efficiency condition, a low-efficiency condition, or a critical condition. For example, if the cosine similarity between the central eigenvector of a certain type of data and the standard eigenvector of a high-efficiency condition is greater than 0.8, the operating condition corresponding to that type of data can be determined to be a high-efficiency condition. If the matching degree is not greater than the threshold, the operating condition is determined to be a normal performance condition.

[0096] For large flow conditions, further analysis of its stability is required. Time series data of parameters such as head and power under large flow conditions are collected to calculate the fluctuation rate of the parameters:

[0097]

[0098] Where x is a parameter sequence.

[0099] If the fluctuation rate exceeds the preset stability threshold (e.g., head fluctuation rate > 8%), the high flow rate operating condition is determined to be an unstable condition, and subsequent analysis requires focus on the pressure pulsation characteristics of the flow field and the force conditions of the impeller.

[0100] After determining the operating condition type, establish a mapping relationship between the operating condition type and the analysis focus. For high-efficiency operating conditions, the analysis focuses on key factors that maintain current performance, such as the fluid dynamic characteristics of the impeller blades and the flow guidance effect of the volute. For low-efficiency operating conditions, the focus is on analyzing the main sources of energy loss, such as the velocity circulation distribution at the impeller outlet and secondary flow phenomena within the pump. For critical operating conditions, the degree of proximity to the performance boundary needs to be assessed, using parameters such as NPSH and surge margin.

[0101] To ensure the reliability of the classification results, cross-validation was used to evaluate the K-means clustering results. The historical data was randomly divided into a training set and a test set. The clustering algorithm was executed on the training set, and the stability of the classification results was verified using the test set. Consistency metrics for the clustering results under different partitions were calculated, such as the Rand Index and the Adjusted Rand Index. Values ​​closer to 1 indicate more reliable classification results.

[0102] At the same time, a dynamic operating condition update mechanism is established. As the mud pump's operating time increases, new performance data is collected in real time. Cluster analysis and matching calculations are re-performed regularly (e.g., monthly) to adapt to operating condition drift caused by factors such as pump wear and changes in fluid properties. During the update process, the timestamps of historical classification results are retained, forming a record of the evolution of operating condition types, providing a basis for equipment maintenance and performance prediction.

[0103] In addition, a visualization tool was developed to assist in the process of determining the working condition type. This tool uses a graphical interface to display clustering results, the matching calculation process, and the final working condition type. It allows users to interactively adjust classification parameters (such as the number of clusters and matching threshold) and displays the adjusted classification results in real time. This visualization method improves the transparency and explainability of the working condition type determination process.

[0104] In practical applications, the above classification algorithm can be packaged into a standardized module and integrated into a mud pump monitoring system. The system collects operating data in real time, automatically calls the classification module to determine the current operating condition type, and generates a report containing operating condition characteristics, performance potential, and analysis recommendations to provide decision support for operators.

[0105] In summary, this embodiment achieves precise classification of mud pump operating conditions by combining flow threshold determination, cluster analysis, performance potential assessment, and feature matching. This process fully considers the physical and data characteristics of mud pump operation, establishing a complete mapping from raw data to operating condition types, providing a clear goal-oriented approach for subsequent performance analysis and optimization. The introduction of a dynamic update mechanism and visualization tools further enhances the practicality and adaptability of the classification method, ensuring accurate identification of mud pump operating conditions across different operating stages and conditions.

[0106] Example 3:

[0107] Based on the current operating conditions, the operating environment data is classified based on the second preset classification algorithm (hierarchical clustering algorithm). Operating environment data includes multidimensional parameters such as ambient temperature, humidity, vibration intensity, medium viscosity, sand content, and inlet pressure. These parameters are collected in real time by sensors and stored as time series data. Hierarchical clustering algorithms are divided into agglomerative (bottom-up) and divisive (top-down) types. The agglomerative algorithm is used here. The specific steps are as follows:

[0108] Similarity calculation: Calculate the similarity between each sample in the operating environment data (each sample is a combination of multi-dimensional environmental parameters at a certain moment). Considering the difference in parameter dimensions, the data is first normalized using the Z-score normalization method:

[0109]

[0110] Where z j is the value of the jth sample after Z-score standardization, x j is the original data value of the jth sample, μ j is the mean value of the parameter, σ j is the standard deviation of the parameter. After normalization, the Euclidean distance between sample i and sample k is calculated as the similarity measure:

[0111]

[0112] Where d(i, k) is the Euclidean distance between sample i and sample k, M is the dimension of environmental parameters (e.g., M=6 corresponds to parameters such as temperature, humidity, and vibration intensity), z i,m and z k,m are the mth normalized parameter values ​​of samples i and k respectively.

[0113] Sample merging and cluster tree construction: Initially, each sample is a separate class. Then, the two closest classes are merged each time, and the distance between the new class and the other classes is calculated (using the class average method, which takes the average of the distances between all samples in the two classes as the inter-class distance). This process continues until all samples are merged into a single large class, forming a tree-like clustering tree. The nodes of the clustering tree represent classes, the leaf nodes represent the original samples, and the branch lengths represent the inter-class distances.

[0114] Category extraction: Based on the preset number of clustering levels (e.g., L = 3) or the cluster distance threshold T (e.g., T = 2.5), appropriate categories are extracted from the cluster tree as the operating environment category data. For example, if the preset number of clustering levels is 3, the cluster tree is divided into three major categories. If a distance threshold is used, nodes with an inter-class distance less than T are merged into a single category. Each extracted category corresponds to a group of operating data with similar environmental parameter characteristics, such as "high temperature and high humidity category" or "low vibration and stability category."

[0115] For each operating environment category data, the impact degree and change frequency of each environmental factor are analyzed to screen the key environmental subset. The impact degree is determined through sensitivity analysis, and the local regression method (LOESS) is used to calculate the absolute value of the correlation coefficient | r between each environmental parameter and the mud pump performance indicators (such as head and efficiency). m ∣, the correlation coefficient calculation formula is:

[0116]

[0117] Where z i，m is the mth standardized environmental parameter value of the ith sample, y i is the performance index value (such as efficiency) of the i-th sample, and are the mean of parameter m and performance index, and the number of samples in N category. m Parameters with ∣≥0.3 are considered as environmental factors with significant influence.

[0118] Frequency of change is determined by counting the number of fluctuations in a parameter per unit time. For continuous parameters (such as temperature), a sliding window method is used to count the number of times a parameter crosses a preset fluctuation threshold (e.g., ±5% of the mean). For discrete parameters (such as medium type), the number of changes within a time window is counted. Parameters with a frequency of change above the average are selected and intersected with parameters with significant impact to form a critical environment subset. For example, if the temperature has a correlation coefficient |r| = 0.4 and a frequency of change of 2 times / hour (higher than the average of 1 time / hour), it is included in the critical environment subset.

[0119] Based on the impact characteristics of the key environment subset and the mean of the mud pump performance indicators, an impact label is assigned to each operating environment category data. The impact characteristics include the direction of the effect of parameter changes on performance (such as temperature increase leading to decreased efficiency) and the degree of effect (such as efficiency decreases by 0.5% for every 1°C increase). The mean performance indicator is the average performance indicator of all samples in the category, such as the average head H avg , average efficiency η avg The impact label adopts the format of "parameter characteristics + impact description", such as "temperature sensitive type (negative correlation with efficiency)" and "vibration stable type (head fluctuation <3%)".

[0120] Then, different strategy generation logics are executed according to the current working condition type (high-efficiency working condition, low-efficiency working condition, critical working condition):

[0121] For high-efficiency operating conditions: The impact labels of each operating environment category are matched with the mud pump's performance indicators to analyze the potential impact of environmental factors on high-efficiency operation. For example, if a category is labeled "Sensitive to Sand Content (Accelerated Head Decrease Rate)," it is necessary to assess whether increased sand content leads to increased impeller wear, thereby affecting head. By establishing a regression model between environmental parameters and performance indicators, key influencing factors can be identified and targeted optimization strategies can be generated, such as adding a media pretreatment step to reduce sand content or selecting wear-resistant materials for the impeller.

[0122] If it is an inefficient operating condition: an improvement optimization strategy is generated based on preset improvement rules and performance improvement targets. The preset improvement rules include adjustment of impeller geometric parameters (such as increasing the blade outlet angle by 5°), modification of the volute flow channel (such as reducing the curvature radius of the diffuser section), optimization of the sealing structure (such as using a labyrinth seal instead of a packing seal), etc. These rules are pre-set based on fluid mechanics theory and engineering experience. Performance improvement targets are determined according to design requirements or industry standards, such as increasing efficiency to more than 85% and reducing the head fluctuation rate to within 5%. By comparing the performance data of the current operating conditions with the target values, a suitable combination of improvement rules is selected to form an optimization plan. For example, if the cause of inefficiency is insufficient velocity circulation at the impeller outlet, the rule of increasing the blade outlet angle and widening the volute flow channel can be adopted.

[0123] If it is a critical operating condition: a performance stabilization strategy is generated based on historical performance fluctuation data and performance stabilization rules. Historical performance fluctuation data includes the parameter fluctuation amplitude, frequency and corresponding environmental conditions within a certain period of time in the past (such as 3 months), and the fluctuation pattern is identified through time series analysis (such as autoregressive moving average model ARMA). Performance stabilization rules include safety protection threshold settings (such as triggering an alarm when the inlet pressure is lower than 0.1MPa), control parameter adjustment logic (such as maintaining flow stability through variable frequency speed regulation), etc. For example, if the critical operating condition is close to the minimum NPSH (Positive Suction Head) value, based on the flow and pressure characteristics when cavitation occurs in the historical data, the return valve is set to automatically open when the NPSH is lower than 1.2 times the design value, increasing the pump inlet flow to avoid cavitation.

[0124] When generating a performance analysis process set, the aforementioned classification, screening, label assignment, and strategy generation processes must be converted into standardized operational steps, with clear input data, processing methods, and output results for each step. For example, the input for the hierarchical clustering analysis step is standardized operating environment data, and the output is cluster category labels and central features; the input for the critical environment subset screening step is category data and performance indicators, and the output is a list of key parameters and impact labels.

[0125] To ensure the scalability of the process, we developed an automated tool to integrate the algorithms for each step. This tool features data import, parameter settings (such as the number of clustering levels and correlation coefficient thresholds), and result visualization (such as cluster tree diagrams and impact tag cloud diagrams). It allows users to adjust analysis parameters based on their needs and automatically generates reports that include process steps, data charts, and strategic recommendations.

[0126] In addition, a process verification mechanism was established to evaluate process effectiveness by comparing historical analysis results with actual performance under different operating conditions. For example, for high-efficiency operating conditions labeled "temperature-sensitive," the actual fluctuation range of performance indicators when the temperature changes was calculated to verify the accuracy of the impact label. For the generated improved optimization strategy, numerical simulation was used to preview its impact on the flow field and performance to ensure the rationality of the strategy.

[0127] In summary, this embodiment uses a hierarchical clustering algorithm to classify operating environment data, screening key factors based on impact level and frequency of change, and developing analysis processes and strategies tailored to the operating condition type. This process transforms multidimensional environmental data into physically meaningful impact labels, enabling correlation analysis between environmental factors and mud pump performance, and providing differentiated optimization paths for different operating conditions. The introduction of automated tools and verification mechanisms ensures the standardization of the process and the reliability of the strategy, laying the foundation for dynamic analysis and continuous optimization of mud pump performance.

[0128] Example 4:

[0129] Taking the performance analysis of a certain model mud pump (rated flow rate 1000m³ / h, rated head 50m) as an example, the preset test rules run through the entire process of the model test, covering key links such as dynamic adjustment of sample size, triggering of supplementary tests, and equipment calibration compensation. The specific applications are as follows:

[0130] 1. Dynamic adjustment of sample size and number of repetitions:

[0131] In the model test of a high-flow condition (1200 m³ / h), the sample size was first determined based on historical test data. Assuming a standard deviation of 2.5% for efficiency under this condition in the historical data and a ±1.5% error margin at a 95% confidence level, the required sample size was calculated using a statistical formula: at least 11 sets of valid data were required to meet the accuracy requirements. Initially, the test was conducted using this sample size. In the first three test sets, the measured efficiency values ​​were 78%, 80%, and 77%, respectively, with a calculated mean of 78.3% and a standard deviation of 1.58%, all within the predefined error margin. However, the fourth set of data suddenly increased to 72%, exceeding the historical standard deviation range and thus identified as anomalous data. This automatically triggered a repeat test. This condition was repeated three times, yielding results of 79%, 78%, and 79%, respectively. After combining the results with the first three sets of data and recalculating the data, the standard deviation dropped to 1.1%, meeting the stability requirements. Ultimately, six sets of valid data were retained (the original three sets of normal data plus the three repeated test sets). The remaining anomalous data were marked as invalid and discarded.

[0132] In tests under normal performance conditions (flow rate 800 m³ / h), if the head fluctuations in the first five test data sets are all less than 3%, the data is considered stable and no additional repetitions are required. If the head deviation between two consecutive data sets exceeds 5%, two additional repetitions are automatically performed until the deviation is less than 3% for three consecutive data sets. This mechanism avoids data distortion caused by accidental interference and ensures the reliability of test results.

[0133] 2. Supplementary test trigger mechanism:

[0134] In a test under critical operating conditions (near the cavitation boundary, with a flow rate of 900 m³ / h and an inlet pressure of 0.08 MPa), the measured head was 42 m, while the numerical simulation result was 45 m, a deviation of 6.7% (exceeding the preset error threshold of ±5%). The system automatically triggered a supplementary test module. The test equipment was first verified: the electromagnetic flowmeter (0.5 accuracy) indicated stable flow, and the pressure transmitter (0.25 accuracy) was within its calibration period. However, slight vibration was detected in the inlet piping. Suspecting that vibration might cause pressure measurement errors, a vibration-damping bracket was installed in the inlet piping and a high-precision pressure sensor (0.1 accuracy) was replaced for retesting. In the second test, the measured head was 44 m, with the deviation from the simulation result reduced to 2.2%, meeting the error requirement. During the supplementary test, the boundary conditions of the CFD model were adjusted, and the inlet turbulence intensity was adjusted from the preset 5% to 3% to match the actual flow conditions and avoid errors caused by mismatches between model assumptions and actual operating conditions.

[0135] If the deviation still exceeds the threshold after the additional test, the multi-physics coupling analysis process is initiated: the pressure distribution in the flow field simulation is compared with the pressure pulsation data measured in the test to check for any unaccounted physical factors (such as medium compressibility and impeller elastic deformation). The model parameters are then corrected or the test plan is adjusted accordingly. For example, if it is found that the simulation does not account for the impact of the medium's sand content on turbulent characteristics, the sand content measurement can be added to the test and the discrete phase model (DPM) can be enabled in the CFD model for recalculation.

[0136] 3. Equipment calibration compensation mechanism:

[0137] Before each test, key equipment such as electromagnetic flowmeters, pressure transmitters, and power meters are calibrated and compensated. Take the calibration of pressure transmitters as an example: install a standard pressure gauge (accuracy level 0.1, range 0-1MPa) in parallel with the test pressure transmitter on the mud pump outlet pipeline, introduce clean water and adjust the pump speed to stabilize the outlet pressure at 10%, 50%, and 90% of the range points (i.e. 0.1MPa, 0.5MPa, and 0.9MPa) in turn. Record the difference in readings between the standard gauge and the test gauge. For example, at 0.5MPa, the standard gauge reads 0.502MPa, the test gauge reads 0.498MPa, and the deviation is -0.004MPa. The calibration compensation coefficient is calculated as:

[0138]

[0139] The original data of the test table is multiplied by the compensation coefficient to make corrections. The calibration process of other equipment (such as electromagnetic flowmeter) is similar. It is compared with a standard flowmeter (such as a volumetric flowmeter) to establish a linear calibration equation:

[0140]

[0141] Where a is the slope correction coefficient and b is the intercept correction coefficient, which are determined by regression analysis of at least three calibration points.

[0142] During the calibration process, if a device's deviation is found to exceed the allowable range of its accuracy level (e.g., a pressure transmitter's deviation > ±0.25% FS), the device will be deemed to require repair or replacement, and testing will be suspended until the device passes calibration. For example, if a power meter's full-scale deviation reaches 1.5% during calibration (its accuracy level is 0.5), the device will be immediately deactivated and a spare power meter activated for recalibration to ensure the accuracy of the test data.

[0143] 4. Cross-condition correlation design of test scheme:

[0144] When generating test plans for each operating condition, emphasis was placed on inter-condition relevance and data reuse. For example, in tests for high-flow conditions (1200 m³ / h) and high-efficiency conditions (1000 m³ / h, efficiency >85%), some test equipment and installation procedures were shared, but the distribution of test points was adjusted accordingly: the high-flow condition focused on monitoring pressure pulsation at the impeller inlet (using five high-frequency pressure sensors), while the high-efficiency condition focused on flow velocity uniformity at the volute outlet (using particle image velocimetry (PIV) to measure the flow field). Furthermore, expansion interfaces were reserved in the test plan to allow for the reuse of some calibration data and installation structures for subsequent conditions (such as those with varying sand content in the medium), reducing duplication of effort.

[0145] 5. Data tracing and process recording:

[0146] During the test, the equipment calibration parameters, sample size adjustment reasons, supplementary test trigger conditions and other information are recorded in real time to form a complete test log. The log content includes:

[0147] Calibration time, calibration equipment model, standard instrument number, calibration deviation and compensation parameters;

[0148] The initial and adjusted sample size values ​​for each operating condition, along with the basis for adjustment (e.g., standard deviation exceeding limits, identification of abnormal data);

[0149] Trigger time, cause analysis, adjustment measures and retest results of the supplementary test;

[0150] Auxiliary information such as test personnel, equipment operating status (such as pump speed fluctuation range, medium temperature changes), etc.

[0151] For example, in one test, the log showed that due to abnormal data in the fourth group under high-flow conditions, the test was repeated three times. The second test was invalid due to loose sensor wiring. After rewiring, the test was successful, and the data from the first and third tests were ultimately used. This detailed record provides a basis for subsequent data review and problem tracing, ensuring the reproducibility of the test process.

[0152] 6. Multidisciplinary Collaborative Verification

[0153] The model test plan and numerical simulation form a closed-loop verification mechanism. Before the test, the test point locations and data collection frequency are preset based on the simulation results. During the test, the measured data is fed back to the simulation team in real time to correct the boundary conditions or mesh division. After the test, the accuracy of the model is evaluated by comparing the flow field cloud map with the PIV measurement results, and the performance curve with the simulation curve. For example, in the low-efficiency working condition test, the measured efficiency was 4% lower than the simulated value. After analysis, it was found that the simulation did not take into account the gap leakage between the impeller and the volute. Therefore, a gap flow module was added to the model. After recalculation, the error between the simulation result and the test was reduced to 1.5%.

[0154] By implementing these pre-set test rules, the mud pump performance analysis testing phase achieved the goals of controllable data quality, traceable processes, and pinpointable issues. Dynamic sample size adjustment avoided redundant or insufficient data, supplementary testing mechanisms ensured systematic troubleshooting of deviation issues, equipment calibration and compensation guaranteed data accuracy from the source, and cross-condition correlation design and multidisciplinary collaboration improved overall analysis efficiency. The combined application of these rules enabled model tests to accurately reflect the actual performance of the mud pump, providing a reliable basis for field testing and optimization of numerical simulations and the development of performance improvement strategies.

[0155] Example 5:

[0156] The following is a detailed description of a mud pump used in an oil field (model NB-1200, rated power 800kW):

[0157] 1. Database architecture design and data storage:

[0158] The mud pump performance analysis database adopts a layered architecture, consisting of a raw data layer, a cleaning layer, an analysis layer, and an application layer. The raw data layer stores unprocessed numerical simulation results (such as CFD calculation grid files and pressure distribution data), raw model test records (such as CSV-formatted flow-time series collected by sensors), design parameter documents (such as CAD drawings and technical specifications), and process data for operating condition classification (such as intermediate cluster analysis results). Taking a high-flow condition (1200 m³ / h) as an example, the raw data layer stores the iteration logs of 20 CFD simulations under this condition, raw waveform files (sampling frequency 100 Hz) from five model tests, and the initial center matrix of the K-means clustering.

[0159] The cleaning layer processes the raw data through an ETL (Extract-Transform-Load) process: it removes outliers caused by sensor failures in model tests (such as data points where the head suddenly changes to a negative value in a certain set of tests), standardizes the units of numerical simulation results (for example, standardizing pressure units to MPa), and adds timestamps and operating condition labels to all data (such as "2025-03-15_High-Efficiency Operating Condition"). For example, when processing operating data from April 2025, the cleaning layer automatically identified and corrected a dimensional error in the medium temperature parameter (incorrectly recording °C as Kelvin) and supplemented missing flow data caused by communication interruptions during a certain period using linear interpolation.

[0160] The analysis layer stores structured data generated by the performance analysis process, including a list of simulation parameter settings for each operating condition (such as the turbulence model selection and grid size for the CFD model), a list of test plans (such as test point coordinates and sensor layout diagrams), performance analysis reports (such as flow field vortex intensity distribution analysis and head efficiency comparison curves), and records of the evolution of operating condition types (such as the time point and reason for reclassifying a certain operating condition from "normal performance" to "inefficient operating condition"). For example, the "Critical Operating Condition Stability Analysis" folder in the analysis layer stores records of eight performance stabilization strategy adjustments for this operating condition between January and May 2025, including the return valve opening threshold, trigger conditions, and corresponding historical fluctuation data fragments for each adjustment.

[0161] The application layer provides functional interfaces for data query and visualization analysis. Operators can use the web interface to search for data by time range (e.g., 2025Q2), operating condition type (e.g., "High-Efficiency Condition"), or analysis context (e.g., "Vibration Analysis"), and view charts comparing simulation parameters and test results for a specific operating condition. For example, when querying data for the high-efficiency operating condition in May 2025, the application layer automatically generates a flow field pressure cloud map for that condition and overlays it with the measured pressure curve, visually demonstrating the agreement between simulation and test results.

[0162] 2. Data cleaning and maintenance mechanism:

[0163] The regular cleanup and maintenance process automatically initiates on the 10th of each month, initially identifying duplicate data using a hash value comparison algorithm. For example, during the May 2025 cleanup, it was discovered that three sets of model test data had been uploaded repeatedly due to an inadvertent system trigger. The redundant records were located and deleted through hash value matching. For invalid data, a Z-score algorithm is used to detect outliers: a threshold of ±3σ is set, and data points outside this range (e.g., records with negative power consumption under a certain operating condition) are automatically flagged and deleted.

[0164] During the maintenance process, cross-version data compatibility must also be addressed. For example, when upgrading CFD software from Ansys 2024 to Ansys 2025, the mesh file format of the older version cannot be directly read. It must be converted to the new format using the database's built-in data conversion tools, and a conversion log must be recorded for traceability. For design parameter changes (such as changing the impeller material from cast iron to stainless steel), the system automatically generates a data version branch, retaining the data from the previous version while enabling the parameters of the new version to ensure the continuity of analysis results.

[0165] 3. Long-term tracking and comparative analysis process:

[0166] When conducting database-based performance trend analysis on mud pumps, we first generate a quarterly time series report on key performance indicators. Taking lift as an example, the report includes statistics such as the quarterly average lift, maximum lift fluctuation, and the correlation coefficient between lift and sand content. During the Q1-Q3 2025 tracking period, we found that the average lift in Q3 decreased by 2.3 meters compared to Q1. During the same period, the average sand content in the medium increased from 5 kg / m³ to 8 kg / m³. We initially speculate that the decrease in lift is related to impeller wear.

[0167] The comparative analysis module supports cross-comparison of data across multiple operating conditions and time periods. For example, comparing high-efficiency operating data from 2025 with the same period in 2024 revealed a 15% increase in average turbulence intensity at the impeller outlet, while efficiency decreased by 4%. Combined with flow field simulation results, this increased turbulence intensity led to increased energy loss. This comparison informs equipment maintenance strategies: During the October 2025 overhaul, the impeller will be treated with a wear-resistant coating, and the volute flow path surface roughness will be optimized to reduce turbulence intensity.

[0168] Long-term tracking also includes evaluating the performance analysis process itself. For example, by comparing the classification results of working condition types in different periods, the stability of the clustering algorithm was verified. In May 2025, the system automatically detected that the deviation of the cluster center of a certain working condition exceeded the preset threshold (10%), prompting the need to retrain the clustering model to avoid classification bias caused by changes in data distribution.

[0169] 4. Data Security and Rights Management

[0170] The database uses role-based access control (RBAC), which categorizes the roles into three types: "Administrator," "Analyst," and "Operator." Administrators have the highest permissions, including data deletion and schema modification. Analysts can create analysis tasks and call algorithm interfaces. Operators can only view real-time data relevant to their current work (such as test results for the current operating condition). For example, operators can only access real-time flow and pressure data collected in the monitoring interface and cannot view detailed simulation parameter settings for historical conditions.

[0171] Data transmission and storage are encrypted: Data is transmitted via the HTTPS protocol, and sensitive information (such as design parameters and optimization strategies) is encrypted at the storage layer using the AES-256 encryption algorithm. Data backups are performed regularly, and backup files are stored on off-site servers to ensure that in the event of system failure or data loss, the most recent valid state can be restored (e.g., a complete data snapshot as of April 30, 2025).

[0172] 5. Examples of actual application scenarios:

[0173] In a shale gas extraction project, mud pumps are required to operate continuously in a high-sand medium (10-15 kg / m³). Database tracking revealed that under these operating conditions, mud pump efficiency decreased at a rate of 0.8% per month, and the depth of erosion wear at the impeller inlet increased by 0.2 mm per quarter. Based on this data, the engineering team adjusted the maintenance cycle: impeller inspection frequency was reduced from every six months to every three months, and wear-resistant impeller spare parts were stockpiled in advance. Furthermore, by comparing flow field simulation results under different sand content conditions, the suction line's diversion structure was optimized to reduce the media's impact angle on the impeller, thereby slowing the wear process.

[0174] In another scenario, a mud pump frequently issued excessive vibration alarms in high summer temperatures. A database search revealed a positive correlation between vibration severity and ambient temperature (correlation coefficient 0.78), with the operating environment during high-temperature periods categorized as "temperature-sensitive (increased vibration)." Based on this, the team added a temperature-controlled fan start / stop threshold adjustment function to the cooling system. When the ambient temperature exceeded 35°C, auxiliary cooling was automatically activated, keeping the bearing temperature below 70°C and reducing the vibration severity below the alarm threshold.

[0175] 6. Data-driven decision support:

[0176] The database supports the full life cycle management of mud pumps: during the design phase, the number of impeller blades is optimized by referring to flow field analysis data from historical operating conditions (for example, increasing it from 5 to 6 to improve efficiency under low-flow conditions); during the manufacturing phase, quality control standards for key components are established based on test data (for example, the impeller casting tolerance is ±0.05mm); during the operation and maintenance phase, the remaining life of the equipment is predicted through performance trend analysis (for example, the impeller replacement cycle is calculated to be 18 months based on the wear rate).

[0177] For example, database analysis from June 2025 revealed that a particular mud pump experienced a 30% increase in critical operating conditions compared to the previous year, and the number of performance stabilization strategy triggers increased by 50%. An assessment determined that the pump had entered its aging phase, and recommended that it be included in the annual upgrade plan. This included replacing the impeller with a high-efficiency, energy-saving model and installing intelligent monitoring sensors to improve operational stability.

[0178] This implementation achieves full lifecycle management of mud pump performance data by building a structured database, establishing standardized cleanup processes, and conducting multi-dimensional tracking and analysis. This mechanism not only provides reliable data support for current performance analysis but also, through in-depth mining of historical data, reveals the underlying patterns of mud pump performance evolution, providing a scientific basis for equipment maintenance, design optimization, and operational decision-making. Data security and permission management ensure the appropriate use of information, while cross-scenario application cases demonstrate the effectiveness and practicality of this mechanism in real-world projects.

[0179] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0180] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A mud pump performance analysis method based on numerical simulation and model test, characterized in that: The method comprises: The performance analysis process of the mud pump is planned based on the design parameters and working conditions of the mud pump, and the numerical simulation content and model test content required for the analysis are determined. The numerical simulation content includes the simulation of the internal flow field of the mud pump, and the model test content includes obtaining the measured performance data of the mud pump under different working conditions. The mud pump structure data and operating environment data are also included. The different working conditions of the mud pump include high-efficiency working conditions, low-efficiency working conditions and critical working conditions. Based on the different working conditions of the mud pump, the numerical simulation content and the model test content, a mud pump performance analysis process set is generated, including: For the current operating condition, classify the operating environment data of the current operating condition based on a second preset classification algorithm to obtain various operating environment category data; for the current operating environment category data, filter the current operating environment category data based on the influence degree and change frequency of each environmental factor in the current operating environment category data to obtain a key environment subset of the current operating environment category data; and determine an impact label for the current operating environment category data based on the impact characteristics of the key environment subset and the mean value of the mud pump performance index; According to whether the working condition type of the current working condition is a high-efficiency working condition; If the current working condition is a high-efficiency working condition, the impact labels of the operating environment category data are matched with the mud pump performance indicators to generate a targeted optimization strategy; If the current working condition is an inefficient working condition, an improvement optimization strategy is generated based on preset improvement rules and performance improvement goals; If the current operating condition is a critical operating condition, generating a performance stabilization strategy based on historical performance fluctuations and performance stabilization rules; Determine different working conditions of the mud pump according to the design parameters and working conditions of the mud pump; Generate a mud pump performance analysis process set according to different working condition types, numerical simulation contents and model test contents of the mud pump, and determine the execution order of the performance analysis process set based on the optimal matching principle; Generate a simulation parameter setting list for each working condition according to a preset analysis period, different working condition types of the dredge pump, and performance index requirements of the dredge pump under each working condition, and generate a test plan list for each working condition according to preset test rules, different working condition types of the dredge pump, and performance index requirements of the dredge pump under each working condition; Based on the execution order of the performance analysis process set, the simulation parameter setting list, the test plan list and the mud pump performance analysis content of each working condition are associated with each other to generate an overall plan for the mud pump performance analysis.

2. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: Plan the performance analysis process of the dredge pump based on its design parameters and operating conditions, and determine the numerical simulation and model test content required for the analysis, including: For a current operating condition, determining a performance analysis boundary for the current operating condition based on historical performance data of the current operating condition, historical performance data of adjacent operating conditions, and a preset operating condition division threshold; Extracting a performance analysis data subset of the current working condition according to the performance analysis boundary and a preset time window; The content of the numerical simulation and model test related data in the performance analysis data subset is determined as the mud pump performance analysis content of the current working condition.

3. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: The different working conditions of the mud pump are determined according to the design parameters and working conditions of the mud pump, including: For the current working condition, when the flow rate of the current working condition is greater than a preset flow rate threshold, determining that the current working condition is a high flow rate working condition; When the flow rate of the current working condition is not greater than the preset flow rate threshold, classifying the head and efficiency data of the current working condition based on a first preset classification algorithm to obtain performance category data, and determining a central feature of the performance category data; determining a performance potential level of the current operating condition based on the working conditions of the current operating condition; When the matching degree between the central feature of each performance category data and the preset performance category is greater than a preset matching threshold, determining that the current operating condition is a specific performance operating condition; When the matching degree is not greater than the preset matching threshold, the current operating condition is determined to be a normal performance operating condition.

4. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: After generating the mud pump performance analysis process set, the following steps are also included: Updating the mud pump performance analysis content of each working condition based on the real-time mud pump operation data; Recalculate the working condition types of each working condition according to the updated mud pump performance analysis content; The execution order of the performance analysis process set is dynamically adjusted according to the recalculated working condition type.

5. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: The preset test rules include: Dynamically adjust the sample size and number of repetitions of the model test based on the confidence interval of the historical mud pump test data; When the deviation between the measured performance data and the numerical simulation results exceeds the preset error threshold, the supplementary test module is automatically triggered.

6. The mud pump performance analysis method based on numerical simulation and model test according to claim 5 is characterized in that: The preset test rules also include: Establish a calibration and compensation mechanism for test equipment, and calibrate the test equipment for error compensation based on the benchmark values ​​of standard flow meters and pressure gauges before each test.

7. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: After generating the overall solution for the mud pump performance analysis, it also includes: Establishing a mud pump performance analysis database, storing the simulation parameter setting list, test plan list, mud pump performance analysis content and data of each working condition type in the overall plan into the database; Regularly clean and maintain the data in the database to remove invalid and duplicate data; Based on this database, the mud pump performance analysis results are tracked and compared over a long period of time to evaluate the changing trend of mud pump performance over time.

8. The mud pump performance analysis method based on numerical simulation and model test according to claim 1 is characterized in that: The second preset classification algorithm is a hierarchical clustering algorithm, and the specific steps of classifying the operating environment data of the current working condition include: Calculate the similarity between samples in the operating environment data; The samples are gradually merged into different categories according to their similarity to form a clustering tree; According to the preset number of clustering layers or clustering distance threshold, appropriate categories are intercepted from the clustering tree as the category data of each operating environment.

Images