A multi-field coupling hydraulic performance prediction method of a high-temperature liquid metal circulating pump

Abstract

The application belongs to the technical field of hydraulic performance prediction, and discloses a multi-field coupling hydraulic performance prediction method for a high-temperature liquid metal circulating pump; the method comprises the following steps: training a hydraulic performance prediction model based on a feature fusion data set to obtain a final performance prediction model, correcting based on the feature fusion data set and the final performance prediction model to obtain a corrected performance prediction model and a corrected prediction data set, predicting the hydraulic performance of equipment based on the corrected performance prediction model and the corrected prediction data set to obtain an equipment performance report, analyzing based on the equipment performance report to obtain a maintenance prediction report and a user decision suggestion report, and providing a visual interface for processing; in general, the application has the advantages of high reliability of basic data, strong self-adaptive ability of the performance prediction model, and good effect of deep mining of prediction data.

Description

Technical Field

[0001] This invention relates to the field of hydraulic performance prediction technology, and more specifically, to a multi-field coupled hydraulic performance prediction method for a high-temperature liquid metal circulating pump. Background Technology

[0002] High-temperature liquid metal circulating pumps are core equipment in advanced energy systems such as advanced metal smelting. Their hydraulic performance (such as efficiency, head, and stability) directly affects the safety and economy of the entire system. Because the equipment operates in an environment of extreme high temperature, strong corrosion, and complex flow, the flow of liquid metal inside the pump is a typical multi-physics coupling process, involving strong interactions between the flow field, temperature field, and structural field (thermal stress and vibration). Therefore, multi-field coupling hydraulic performance prediction of high-temperature liquid metal circulating pumps, that is, accurate calculation and prediction of pump performance parameters under the above multi-field coupling effects, is crucial for optimized design, safe operation, and predictive maintenance.

[0003] However, traditional multi-field coupled hydraulic performance prediction methods for high-temperature liquid metal circulating pumps suffer from several shortcomings in practice. First, these methods largely rely on offline or limited sensor data, which suffers from slow update speeds, incomplete data dimensions, and significant noise interference. This leads to substantial deviations in the prediction results of the system's performance prediction model, making it difficult to provide truly reliable prediction capabilities. Second, traditional methods are mostly based on purely physical mechanism models. While these models have general applicability, they suffer from computational complexity and time consumption. Furthermore, once the model parameters are set, they remain fixed, causing the prediction accuracy to decline under long-term operating conditions. The rapid decline makes it difficult to provide data support for operation and maintenance. Thirdly, most traditional methods stop at performance prediction, and their prediction results are not linked to enterprise operations. Users cannot directly transform the prediction results into deeper-level maintenance calculations, cost analysis, and resource allocation decisions, resulting in the inability to directly convert prediction data into economic benefits, causing a waste of data resources, and thus affecting the practicality of the system. In general, how to effectively solve the problems of poor reliability of basic data, rigid performance prediction models, and insufficient ability to deeply mine prediction data in traditional methods has become the problem that current multi-field coupled hydraulic performance prediction methods for high-temperature liquid metal circulating pumps need to face and solve.

[0004] In view of this, the present invention proposes a multi-field coupled hydraulic performance prediction method for high-temperature liquid metal circulating pumps. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution, including:

[0006] The multi-source data acquisition module is used to collect raw data from the database based on API call commands to obtain the raw dataset.

[0007] Furthermore, the steps for collecting raw data from the database based on API call instructions include:

[0008] S1.1 The system generates API call instructions and retrieves real-time raw data from the database to obtain the first raw dataset, which includes temperature field, pressure field, flow velocity field and vibration field.

[0009] S1.2. Process the first original dataset based on the time series to obtain the second original dataset;

[0010] S1.3. Convert the second original dataset into a preset format to obtain the original dataset, and store the original dataset in the database;

[0011] S1.4 Output the raw dataset to the data preprocessing module;

[0012] The data preprocessing module is used to preprocess the original dataset to obtain a cleaned dataset;

[0013] Furthermore, the preprocessing steps for the original dataset include:

[0014] S2.1 Based on the original dataset, and using a moving average filter to denoise all data items of all data types in the original dataset, a denoised dataset is obtained. The specific calculation formula for denoising is as follows:

[0015] ;

[0016] Get the time point denoised output value ,in, To adjust the sliding window size, The radius is the size of the sliding window. For time points A data item of a certain data type in the original dataset;

[0017] S2.2 Based on the denoised dataset, the Z-score method is used to detect outliers. When the detection result is an outlier, the data items of the outlier result are removed and replaced with the median of the corresponding data type to obtain the outlier detection dataset.

[0018] S2.3. Based on the anomaly detection dataset, missing values ​​are filled using linear interpolation to obtain the missing values ​​filled dataset.

[0019] S2.4. Use the min-max normalization method to scale all data items in the missing data imputation dataset to the range of [0, 1] to obtain the cleaned dataset;

[0020] S2.5. Output the cleaned dataset to the multi-source data fusion and extraction module;

[0021] The multi-source data fusion and extraction module is used to fuse and extract features from a cleaned dataset to obtain a feature-fused dataset.

[0022] Furthermore, the steps for fusion and feature extraction based on the cleaned dataset include:

[0023] S3.1. Based on the temperature type data in the cleaning dataset, temperature gradient features are extracted to obtain a temperature gradient feature vector. The specific calculation formula for the extraction is as follows:

[0024] ;

[0025] Get the time point Temperature gradient eigenvector ,in, For time points Clean the temperature data in the dataset. For time points Clean the temperature data in the dataset. For time intervals;

[0026] S3.2. Based on the pressure type data in the cleaning dataset, pressure fluctuation features are extracted to obtain a pressure fluctuation feature vector. The specific calculation formula for the extraction is as follows:

[0027]

[0028] Obtain the pressure fluctuation feature vector ,in, To reduce the amount of stress data in the dataset, For time points Clean the stress data in the dataset. To clean the mean of the pressure data in the dataset;

[0029] S3.3 Calculate the average value of the flow velocity type data in the cleaning dataset to obtain the average flow velocity characteristic data;

[0030] S3.4. Based on the vibration type data in the cleaning dataset, select the maximum value to obtain the vibration peak characteristic data;

[0031] S3.5, package temperature gradient feature vector, pressure fluctuation feature vector, average flow velocity feature data and vibration peak feature data, to obtain feature fusion dataset;

[0032] S3.6 Output the feature fusion dataset to the hydraulic performance prediction model training module;

[0033] The hydraulic performance prediction model training module is used to train the hydraulic performance prediction model based on the feature fusion dataset to obtain the final performance prediction model.

[0034] Furthermore, the steps for training the hydraulic performance prediction model based on the feature fusion dataset include:

[0035] S4.1 Divide the feature fusion dataset into a 70% training set and a 30% test set;

[0036] S4.2 Construct a hydraulic performance prediction model. The hydraulic performance prediction model is used to predict hydraulic performance. The hydraulic performance prediction model includes an input layer, a hidden layer, and an output layer.

[0037] S4.3. Train the hydraulic performance prediction model using the training set;

[0038] S4.4 Calculate the evaluation index of the hydraulic performance prediction model using the test set to obtain the model evaluation value. When the model evaluation value is greater than or equal to the evaluation threshold, the final performance prediction model is obtained, and the process proceeds to step S4.5. When the model evaluation value is less than the evaluation threshold, the process returns to step S4.1 to retrain the model. The specific formula for calculating the model evaluation value is as follows:

[0039] ;

[0040] Obtain model evaluation values ,in, The number of samples in the test set. For the first A true performance value, For the first One predicted performance value, This represents the average of the actual performance values.

[0041] S4.5 Store the final performance prediction model in the database and output it to the dynamic model calibration module;

[0042] The dynamic model calibration module is used to calibrate based on the feature fusion dataset and the final performance prediction model to obtain the calibrated performance prediction model and the calibrated prediction dataset.

[0043] Furthermore, the steps for calibration based on the feature fusion dataset and the final performance prediction model include:

[0044] S5.1 Retrieve and run the traditional physical model based on the NS equation from the database, input the feature fusion dataset into the traditional physical model, and output the first prediction performance value;

[0045] S5.2 Run the final performance prediction model, input the feature fusion dataset, and output the second predicted performance value;

[0046] S5.3. The first and second prediction performance values ​​are calculated using the weighted average method to obtain the corrected prediction value. The specific formula for the calculation is as follows:

[0047] ;

[0048] Obtain the corrected predicted value ,in, As a weighting factor, This is the first performance prediction value. This is the second performance prediction value;

[0049] S5.4 Adjust the model parameters of the final performance prediction model based on the corrected prediction values ​​to obtain the corrected performance prediction model;

[0050] S5.5. Pack the corrected predicted values ​​and the feature fusion dataset to obtain the corrected predicted dataset;

[0051] S5.6 Output the corrected performance prediction model and the corrected prediction dataset to the hydraulic performance prediction simulation module;

[0052] The hydraulic performance prediction and simulation module is used to predict the hydraulic performance of equipment based on the corrected performance prediction model and the corrected prediction dataset, and to obtain the equipment performance report.

[0053] Furthermore, the steps for predicting the hydraulic performance of the equipment based on the corrected performance prediction model and the corrected prediction dataset include:

[0054] S6.1 Run the correction performance prediction model, retrieve the feature fusion dataset within the preset time window from the database, obtain the historical feature fusion dataset, input the historical feature fusion dataset into the correction performance prediction module, and output the key performance dataset, which includes head data values, correction prediction performance values ​​and life loss values.

[0055] S6.2 Calculate 95% confidence intervals based on all data items in the key performance dataset to obtain a set of confidence intervals;

[0056] S6.3. Perform time series analysis based on key performance datasets to obtain a performance decline trend report. The analysis includes calculating the moving average and rate of change.

[0057] S6.4. Based on the confidence interval set of the key performance dataset and the performance decline trend report, compile them into chart formats and package them to obtain the device performance report;

[0058] S6.5 Output the equipment performance report to the decision report visualization module;

[0059] The decision report visualization module is used to analyze equipment performance reports to obtain maintenance prediction reports and user decision suggestion reports, and provides a visual interface for processing.

[0060] Furthermore, the steps of analyzing equipment performance reports to obtain maintenance prediction reports and user decision-making suggestion reports, and providing a visual interface for processing, include:

[0061] S7.1 Based on the equipment performance report, when the corrected predicted performance value is less than the efficiency threshold or the life loss value is greater than the life loss threshold, a maintenance alarm report is generated.

[0062] The maintenance alarm report includes a description of poor performance or insufficient remaining life of the current circulating pump, requesting staff to contact maintenance personnel as soon as possible for inspection and maintenance;

[0063] S7.2 Calculate the most recent maintenance time based on the life loss value in the equipment performance report. The specific formula for the calculation is as follows:

[0064] ;

[0065] Get the latest maintenance time ,in, For the current time, This is the lifespan loss threshold. This represents the current lifespan loss value. This represents the rate of change in lifetime loss.

[0066] S7.3. The estimated maintenance cost is obtained by summing the products of parts cost and labor cost multiplied by the estimated downtime.

[0067] S7.4. Retrieve the equipment maintenance plan from the database and calculate the optimal inventory level. The specific formula for the calculation is as follows:

[0068] ;

[0069] To obtain the optimal inventory level ,in, Annual demand For ordering costs, Inventory holding costs;

[0070] S7.5. The failure risk assessment value is obtained by calculating the product of the failure probability and the cost of generating loss.

[0071] S7.6 Display maintenance alarm reports, recent maintenance time, estimated maintenance cost, optimal inventory level, failure risk assessment value and equipment performance report based on the visualization panel;

[0072] Furthermore, a multi-field coupled hydraulic performance prediction method for a high-temperature liquid metal circulating pump includes:

[0073] S1. Collect raw data from the database based on API call instructions to obtain the raw dataset;

[0074] S2. Preprocess the original dataset to obtain a cleaned dataset;

[0075] S3. Based on the cleaned dataset, perform fusion and feature extraction to obtain a feature fusion dataset;

[0076] S4. Train the hydraulic performance prediction model based on the feature fusion dataset to obtain the final performance prediction model;

[0077] S5. Based on the feature fusion dataset and the final performance prediction model, correction is performed to obtain the corrected performance prediction model and the corrected prediction dataset;

[0078] S6. Based on the corrected performance prediction model and the corrected prediction dataset, predict the hydraulic performance of the equipment and obtain the equipment performance report;

[0079] S7. Based on the equipment performance report, analyze the data to obtain a maintenance prediction report and a user decision suggestion report, and provide a visual interface for processing.

[0080] The present invention has the following technical effects:

[0081] This invention provides comprehensive, reliable, and multi-dimensional data to the system through the establishment of a source data acquisition module, a data preprocessing module, and a multi-source data fusion and extraction module, fundamentally ensuring the accuracy of performance prediction. Furthermore, through the collaborative operation of a hydraulic performance prediction model training module, a dynamic model calibration module, and a hydraulic performance prediction simulation module, the performance prediction model is no longer a rigid calculation formula, but rather an intelligent agent with self-optimization capabilities that grow with usage time, significantly improving prediction accuracy, prediction speed, and adaptability. Finally, through a decision report visualization module, the prediction results are effectively transformed into maintenance calculations, cost analysis, and resource allocation decisions, thereby significantly improving equipment reliability and data utilization. Overall, this invention has the advantages of high reliability of basic data, strong adaptability of the performance prediction model, and good results in in-depth mining of prediction data. Attached Figure Description

[0082] Figure 1 This is a schematic diagram of a multi-field coupled hydraulic performance prediction system for a high-temperature liquid metal circulating pump according to the present invention;

[0083] Figure 2 This is a schematic diagram of a multi-field coupled hydraulic performance prediction method for a high-temperature liquid metal circulating pump according to the present invention. Detailed Implementation

[0084] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0086] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0087] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0088] In practice, the server-side equipment deployed in the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps may consist of one or more devices. This system can be implemented as a business instance, a virtual machine, or a hardware device. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node to provide multi-field coupled hydraulic performance prediction services to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage each user terminal. Or, it can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more devices configured to provide multi-field coupled hydraulic performance prediction services to various user terminals.

[0089] In terms of implementation, the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps and the user terminal are mutually compatible. That is, if the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps is implemented as an application installed on a cloud service platform, then the user terminal is implemented as a client that establishes a communication connection with the application; or if the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps is implemented as a website, then the user terminal is implemented as a webpage; or if the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0090] like Figure 1 The figure shown is a system architecture diagram of a multi-field coupled hydraulic performance prediction system for a high-temperature liquid metal circulating pump provided in an embodiment of the present invention.

[0091] The multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps described in this invention can be hosted on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server or server cluster), or it can be developed into a website. Depending on the functions implemented, the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps may include a multi-source data acquisition module, a data preprocessing module, a multi-source data fusion and extraction module, a hydraulic performance prediction model training module, a dynamic model calibration module, a hydraulic performance prediction simulation module, and a decision report visualization module. The module described in this invention can also be referred to as a unit, which is a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0092] In this embodiment of the invention, in the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps, each of the above-mentioned modules can be implemented independently and can call other modules. Here, "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. For example, the decision report visualization module can call the same information acquisition module to obtain information collected by that module. Based on the above characteristics, in the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps provided in this embodiment of the invention, the applicable scope of the system architecture can be adjusted by adding modules and directly calling them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the multi-field coupled hydraulic performance prediction system for high-temperature liquid metal circulating pumps. In practical applications, the above-mentioned modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in a cloud server. Example 1

[0093] Please see Figure 1 As shown in this embodiment, a multi-field coupled hydraulic performance prediction system for a high-temperature liquid metal circulating pump includes,

[0094] The multi-source data acquisition module is used to acquire raw data from the database based on API call instructions to obtain the raw dataset;

[0095] The steps for collecting raw data from a database based on API call commands include:

[0096] S1.1 The system generates API call instructions and retrieves real-time raw data from the database to obtain the first raw dataset, which includes temperature field, pressure field, flow velocity field and vibration field.

[0097] It needs to be explained that the temperature field, pressure field, flow velocity field, and structural vibration field in the original dataset refer to the fact that all three-dimensional spatial points inside the circulating pump have corresponding data values ​​at a certain moment.

[0098] S1.2. Process the first original dataset based on the time series to obtain the second original dataset;

[0099] It needs to be explained that the purpose of this step is to arrange all the data items in the first raw dataset according to the data type and the data collection timestamp. That is, each data item corresponds to a timestamp, and the data types include temperature, pressure, flow rate and vibration.

[0100] S1.3. Convert the second original dataset into a preset format to obtain the original dataset, and store the original dataset in the database;

[0101] It should be explained that the default format is manually set, for example, the default format is JSON;

[0102] S1.4 Output the raw dataset to the data preprocessing module;

[0103] The data preprocessing module is used to preprocess the original dataset to obtain a cleaned dataset;

[0104] The steps for preprocessing the original dataset include:

[0105] S2.1 Based on the original dataset, and using a moving average filter to denoise all data items of all data types in the original dataset, a denoised dataset is obtained. The specific calculation formula for denoising is as follows:

[0106] ;

[0107] Get the time point denoised output value ,in, To adjust the sliding window size, The radius is the size of the sliding window. For time points A data item of a certain data type in the original dataset;

[0108] It needs to be explained that the radius of the sliding window size refers to, for example, when When the value is 2, the sliding window size is 5;

[0109] S2.2 Based on the denoised dataset, the Z-score method is used to detect outliers. When the detection result is an outlier, the data items of the outlier result are removed and replaced with the median of the corresponding data type to obtain the outlier detection dataset.

[0110] It should be explained that replacing with the median of the corresponding data type means, for example, if the outlier is a data item of the temperature data type, then the median of the data item of the temperature type will be used for replacement.

[0111] S2.3. Based on the anomaly detection dataset, missing values ​​are filled using linear interpolation to obtain the missing values ​​filled dataset.

[0112] S2.4. Use the min-max normalization method to scale all data items in the missing data imputation dataset to the range of [0, 1] to obtain the cleaned dataset;

[0113] S2.5. Output the cleaned dataset to the multi-source data fusion and extraction module;

[0114] The multi-source data fusion and extraction module is used to perform fusion and feature extraction based on the cleaned dataset to obtain a feature fusion dataset;

[0115] The steps for fusion and feature extraction based on a cleaned dataset include:

[0116] S3.1. Based on the temperature type data in the cleaning dataset, temperature gradient features are extracted to obtain a temperature gradient feature vector. The specific calculation formula for the extraction is as follows:

[0117] ;

[0118] Get the time point Temperature gradient eigenvector ,in, For time points Clean the temperature data in the dataset. For time points Clean the temperature data in the dataset. For time intervals;

[0119] S3.2. Based on the pressure type data in the cleaning dataset, pressure fluctuation features are extracted to obtain a pressure fluctuation feature vector. The specific calculation formula for the extraction is as follows:

[0120] ;

[0121] Obtain the pressure fluctuation feature vector ,in, To reduce the amount of stress data in the dataset, For time points Clean the stress data in the dataset. To clean the mean of the pressure data in the dataset;

[0122] S3.3 Calculate the average value of the flow velocity type data in the cleaning dataset to obtain the average flow velocity characteristic data;

[0123] S3.4. Based on the vibration type data in the cleaning dataset, select the maximum value to obtain the vibration peak characteristic data;

[0124] S3.5, package temperature gradient feature vector, pressure fluctuation feature vector, average flow velocity feature data and vibration peak feature data, to obtain feature fusion dataset;

[0125] S3.6 Output the feature fusion dataset to the hydraulic performance prediction model training module;

[0126] The hydraulic performance prediction model training module is used to train the hydraulic performance prediction model based on the feature fusion dataset to obtain the final performance prediction model.

[0127] The steps for training a hydraulic performance prediction model based on a feature fusion dataset include:

[0128] S4.1 Divide the feature fusion dataset into a 70% training set and a 30% test set;

[0129] S4.2 Construct a hydraulic performance prediction model. The hydraulic performance prediction model is used to predict hydraulic performance. The hydraulic performance prediction model includes an input layer, a hidden layer, and an output layer.

[0130] S4.3. Train the hydraulic performance prediction model using the training set;

[0131] S4.4 Calculate the evaluation index of the hydraulic performance prediction model using the test set to obtain the model evaluation value. When the model evaluation value is greater than or equal to the evaluation threshold, the final performance prediction model is obtained, and the process proceeds to step S4.5. When the model evaluation value is less than the evaluation threshold, the process returns to step S4.1 to retrain the model. The specific formula for calculating the model evaluation value is as follows:

[0132] ;

[0133] Obtain model evaluation values ,in, The number of samples in the test set. For the first A true performance value, For the first One predicted performance value, This represents the average of the actual performance values.

[0134] S4.5 Store the final performance prediction model in the database and output it to the dynamic model calibration module;

[0135] The dynamic model correction module is used to correct the model based on the feature fusion dataset and the final performance prediction model to obtain the corrected performance prediction model and the corrected prediction dataset.

[0136] The steps for calibration based on the feature fusion dataset and the final performance prediction model include:

[0137] S5.1 Retrieve and run the traditional physical model based on the NS equation from the database, input the feature fusion dataset into the traditional physical model, and output the first prediction performance value;

[0138] It should be explained that the Navier-Stokes equations refer to the set of partial differential equations in fluid mechanics that describe viscous Newtonian fluids.

[0139] S5.2 Run the final performance prediction model, input the feature fusion dataset, and output the second predicted performance value;

[0140] S5.3. The first and second prediction performance values ​​are calculated using the weighted average method to obtain the corrected prediction value. The specific formula for the calculation is as follows:

[0141] ;

[0142] Obtain the corrected predicted value ,in, As a weighting factor, This is the first performance prediction value. This is the second performance prediction value;

[0143] S5.4 Adjust the model parameters of the final performance prediction model based on the corrected prediction values ​​to obtain the corrected performance prediction model;

[0144] S5.5. Pack the corrected predicted values ​​and the feature fusion dataset to obtain the corrected predicted dataset;

[0145] S5.6 Output the corrected performance prediction model and the corrected prediction dataset to the hydraulic performance prediction simulation module;

[0146] The hydraulic performance prediction simulation module is used to predict the hydraulic performance of the equipment based on the corrected performance prediction model and the corrected prediction dataset, and to obtain an equipment performance report.

[0147] The steps for predicting the hydraulic performance of equipment based on a corrected performance prediction model and a corrected prediction dataset include:

[0148] S6.1 Run the correction performance prediction model, retrieve the feature fusion dataset within the preset time window from the database, obtain the historical feature fusion dataset, input the historical feature fusion dataset into the correction performance prediction module, and output the key performance dataset, which includes head data values, correction prediction performance values ​​and life loss values.

[0149] It should be explained that the historical feature fusion dataset is based on a database and is obtained by packaging the feature fusion dataset retrieved according to a preset time window;

[0150] S6.2 Calculate 95% confidence intervals based on all data items in the key performance dataset to obtain a set of confidence intervals;

[0151] S6.3. Perform time series analysis based on key performance datasets to obtain a performance decline trend report. The analysis includes calculating the moving average and rate of change.

[0152] S6.4. Based on the confidence interval set of the key performance dataset and the performance decline trend report, compile them into chart formats and package them to obtain the device performance report;

[0153] S6.5 Output the equipment performance report to the decision report visualization module;

[0154] The decision report visualization module is used to analyze equipment performance reports to obtain maintenance prediction reports and user decision suggestion reports, and provides a visualization interface for processing.

[0155] The steps involved in analyzing equipment performance reports to obtain maintenance prediction reports and user decision-making suggestion reports, and providing a visual interface for processing, include:

[0156] S7.1 Based on the equipment performance report, when the corrected predicted performance value is less than the efficiency threshold or the life loss value is greater than the life loss threshold, a maintenance alarm report is generated.

[0157] It should be explained that the efficiency threshold and lifespan loss threshold are manually set and input into the system, for example, the efficiency threshold is 0.8 and the lifespan loss threshold is 0.7;

[0158] The maintenance alarm report includes a description of poor performance or insufficient remaining life of the current circulating pump, requesting staff to contact maintenance personnel as soon as possible for inspection and maintenance;

[0159] S7.2 Calculate the most recent maintenance time based on the life loss value in the equipment performance report. The specific formula for the calculation is as follows:

[0160] ;

[0161] Get the latest maintenance time ,in, For the current time, This is the lifespan loss threshold. This represents the current lifespan loss value. This represents the rate of change in lifetime loss.

[0162] S7.3. The estimated maintenance cost is obtained by summing the products of parts cost and labor cost multiplied by the estimated downtime.

[0163] S7.4. Retrieve the equipment maintenance plan from the database and calculate the optimal inventory level. The specific formula for the calculation is as follows:

[0164] ;

[0165] To obtain the optimal inventory level ,in, Annual demand For ordering costs, Inventory holding costs;

[0166] S7.5. The failure risk assessment value is obtained by calculating the product of the failure probability and the cost of generating loss.

[0167] S7.6 Display maintenance alarm reports, recent maintenance time, estimated maintenance cost, optimal inventory level, failure risk assessment value and equipment performance report based on the visualization panel;

[0168] This embodiment provides the system with comprehensive, reliable, and multi-dimensional data through the establishment of a source data acquisition module, a data preprocessing module, and a multi-source data fusion and extraction module, fundamentally ensuring the accuracy of performance prediction. Furthermore, through the collaborative operation of a hydraulic performance prediction model training module, a dynamic model calibration module, and a hydraulic performance prediction simulation module, the performance prediction model is no longer a rigid calculation formula, but rather an intelligent agent with self-optimization capabilities that grow with usage time, significantly improving prediction accuracy, prediction speed, and adaptability. Finally, through a decision report visualization module, the prediction results are effectively transformed into maintenance calculations, cost analysis, and resource allocation decisions, thereby significantly improving equipment reliability and data utilization. Overall, this invention has the advantages of high reliability of basic data, strong adaptability of the performance prediction model, and good deep mining effect of prediction data. Example 2

[0169] Please see Figure 2As shown, the parts not described in detail in this embodiment are described in Embodiment 1. A method for predicting the multi-field coupled hydraulic performance of a high-temperature liquid metal circulating pump is provided. The method includes: S1, collecting raw data from the database based on API call instructions to obtain the raw dataset;

[0170] S2. Preprocess the original dataset to obtain a cleaned dataset;

[0171] S3. Based on the cleaned dataset, perform fusion and feature extraction to obtain a feature fusion dataset;

[0172] S4. Train the hydraulic performance prediction model based on the feature fusion dataset to obtain the final performance prediction model;

[0173] S5. Based on the feature fusion dataset and the final performance prediction model, correction is performed to obtain the corrected performance prediction model and the corrected prediction dataset;

[0174] S6. Based on the corrected performance prediction model and the corrected prediction dataset, predict the hydraulic performance of the equipment and obtain the equipment performance report;

[0175] S7. Based on the equipment performance report, analyze the data to obtain a maintenance prediction report and a user decision suggestion report, and provide a visual interface for processing.

[0176] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the present invention.

Claims

1. A method for predicting the multi-field coupled hydraulic performance of a high-temperature liquid metal circulating pump, characterized in that, The method includes, S1. Collect raw data from the database based on API call instructions to obtain the raw dataset; The steps for collecting raw data from a database based on API call commands include: S1.1 The system generates API call instructions and retrieves real-time raw data from the database to obtain the first raw dataset, which includes temperature field, pressure field, flow velocity field and vibration field. S1.

2. Process the first original dataset based on the time series to obtain the second original dataset; S1.

3. Convert the second original dataset into a preset format to obtain the original dataset, and store the original dataset in the database; S1.4 Output the raw dataset to the data preprocessing module; S2. Preprocess the original dataset to obtain a cleaned dataset; S3. Based on the cleaned dataset, perform fusion and feature extraction to obtain the feature fusion dataset; The steps for fusion and feature extraction based on a cleaned dataset include: S3.

1. Extract temperature gradient features based on temperature type data in the cleaning dataset to obtain temperature gradient feature vectors; S3.

2. Based on the pressure type data in the cleaning dataset, pressure fluctuation features are extracted to obtain a pressure fluctuation feature vector; S3.3 Calculate the average value of the flow velocity type data in the cleaning dataset to obtain the average flow velocity characteristic data; S3.

4. Based on the vibration type data in the cleaning dataset, select the maximum value to obtain the vibration peak characteristic data; S3.5, package temperature gradient feature vector, pressure fluctuation feature vector, average flow velocity feature data and vibration peak feature data to obtain feature fusion dataset; S3.6 Output the feature fusion dataset to the hydraulic performance prediction model training module; S4. Train the hydraulic performance prediction model based on the feature fusion dataset to obtain the final performance prediction model; The steps for training a hydraulic performance prediction model based on a feature fusion dataset include: S4.1 Divide the feature fusion dataset into a 70% training set and a 30% test set; S4.2 Construct a hydraulic performance prediction model. The hydraulic performance prediction model is used to predict hydraulic performance. The hydraulic performance prediction model includes an input layer, a hidden layer, and an output layer. S4.

3. Train the hydraulic performance prediction model using the training set; S4.4 Calculate the evaluation index of the hydraulic performance prediction model using the test set and obtain the model evaluation value. When the model evaluation value is greater than or equal to the evaluation threshold, the final performance prediction model is obtained and proceed to step S4.

5. When the model evaluation value is less than the evaluation threshold, return to step S4.1 to retrain the model. S4.5 Store the final performance prediction model in the database and output it to the dynamic model calibration module; S5. Based on the feature fusion dataset and the final performance prediction model, correction is performed to obtain the corrected performance prediction model and the corrected prediction dataset; The steps for calibration based on the feature fusion dataset and the final performance prediction model include: S5.1 Retrieve and run the traditional physical model based on the NS equation from the database, input the feature fusion dataset into the traditional physical model, and output the first prediction performance value; S5.2 Run the final performance prediction model, input the feature fusion dataset, and output the second predicted performance value; S5.3 Calculate the first and second prediction performance values ​​using the weighted average method to obtain the corrected prediction value; S5.4 Adjust the model parameters of the final performance prediction model based on the corrected prediction values ​​to obtain the corrected performance prediction model; S5.

5. Pack the corrected predicted values ​​and the feature fusion dataset to obtain the corrected predicted dataset; S5.6 Output the corrected performance prediction model and the corrected prediction dataset to the hydraulic performance prediction simulation module; S6. Based on the corrected performance prediction model and the corrected prediction dataset, predict the hydraulic performance of the equipment and obtain the equipment performance report; The steps for predicting the hydraulic performance of equipment based on a corrected performance prediction model and a corrected prediction dataset include: S6.1 Run the correction performance prediction model, retrieve the feature fusion dataset within the preset time window from the database, obtain the historical feature fusion dataset, input the historical feature fusion dataset into the correction performance prediction module, and output the key performance dataset, which includes head data values, correction prediction performance values ​​and life loss values. S6.2 Calculate 95% confidence intervals based on all data items in the key performance dataset to obtain a set of confidence intervals; S6.

3. Perform time series analysis based on key performance datasets to obtain a performance decline trend report. The analysis includes calculating the moving average and rate of change. S6.

4. Based on the confidence interval set of the key performance dataset and the performance decline trend report, compile them into chart formats and package them to obtain the device performance report; S6.5 Output the equipment performance report to the decision report visualization module; S7. Analyze the equipment performance report to obtain a maintenance prediction report and a user decision suggestion report, and provide a visual interface for processing; The steps involved in analyzing equipment performance reports to obtain maintenance prediction reports and user decision-making suggestion reports, and providing a visual interface for processing, include: S7.1 Based on the equipment performance report, when the corrected predicted performance value is less than the efficiency threshold or the life loss value is greater than the life loss threshold, a maintenance alarm report is generated. The maintenance alarm report includes a description of poor performance or insufficient remaining life of the current circulating pump, requesting staff to contact maintenance personnel as soon as possible for inspection and maintenance; S7.2 Calculate the most recent maintenance time based on the life loss value in the equipment performance report; S7.

3. The estimated maintenance cost is obtained by summing the products of parts cost and labor cost multiplied by the estimated downtime. S7.

4. Retrieve the equipment maintenance plan from the database and calculate the optimal inventory level. S7.

5. The failure risk assessment value is obtained by calculating the product of the failure probability and the cost of generating loss. S7.6 Displays maintenance alarm reports, recent maintenance time, estimated maintenance cost, optimal inventory level, failure risk assessment value, and equipment performance report based on a visualization panel.

2. The method for predicting the multi-field coupled hydraulic performance of a high-temperature liquid metal circulating pump according to claim 1, characterized in that, The steps for preprocessing the original dataset include: S2.1 Based on the original dataset, and using a moving average filter to denoise all data items of all data types in the original dataset, a denoised dataset is obtained. S2.2 Based on the denoised dataset, the Z-score method is used to detect outliers. When the detection result is an outlier, the data items of the outlier result are removed and replaced with the median of the corresponding data type to obtain the outlier detection dataset. S2.

3. Based on the anomaly detection dataset, missing values ​​are filled using linear interpolation to obtain the missing values ​​filled dataset. S2.

4. Use the min-max normalization method to scale all data items in the missing data imputation dataset to the range of [0, 1] to obtain the cleaned dataset; S2.5 Output the cleaned dataset to the multi-source data fusion and extraction module.

Images