Artificial intelligence monitoring method and system for heavy-duty gas turbine material performance test data

Abstract

The application relates to a heavy gas turbine material performance test data artificial intelligence monitoring method and system, which comprises the following steps: obtaining a material test data set; obtaining a quality evaluation value of each dimension material test data, and dividing the material test data into valid data and invalid data based on the quality evaluation value; integrating the valid data into a valid data set, obtaining a structured data set of each material test data in the valid data set at different temperatures; calling an editing and approval process, executing the editing and approval process on the structured data set of any material test data, and storing the material test data set corresponding to the structured data set to a gas turbine design material database in the case that the structured data set is approved. Through the four steps of data extraction, quality evaluation, calculation processing and editing and approval, a complete data processing system is formed, the accuracy and reliability of the product in the use process are guaranteed, and efficient processing and effective utilization of the multi-dimensional test data of the gas turbine material are realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an artificial intelligence monitoring method and system for test data of material performance of heavy-duty gas turbines. Background Technology

[0002] Typically, the material performance data required for the design, life prediction, and safety assessment of gas turbines covers multiple aspects, including physical properties, elastic properties, tensile properties, high-cycle fatigue, low-cycle fatigue, crack propagation, fracture toughness, creep rupture, and oxidation properties. This data requires sampling and testing from multiple batches and samples to obtain accurate and valid test results. These results then undergo a series of data processing steps to ultimately form the material performance data used for design.

[0003] However, the requirements for material data differ at different design stages during the gas turbine product design process. For example, when designing the overall structural integrity, the focus is primarily on the average material values; while in the life prediction and safety assessment stages, in addition to average values, relatively conservative minimum material performance values ​​are more important to ensure the safe and stable operation of the gas turbine product. The minimum material performance values ​​directly affect the results of life prediction and safety assessment. If the minimum values ​​are handled too conservatively, while safety can be ensured, it will lead to material waste and increased gas turbine costs; conversely, if the handling is too lenient, while costs can be reduced, it will bring significant safety risks and fail to guarantee the safety and stability of the product during use.

[0004] Therefore, there is an urgent need to develop a data processing method, system, electronic device, and storage medium to solve one or more of the aforementioned problems. Summary of the Invention

[0005] In view of this, in order to solve the above-mentioned technical problems or some of the technical problems, the embodiments of this application provide a data processing method, system, electronic device and storage medium. The method forms a complete data processing system through four steps: data extraction, quality assessment, calculation and processing and editing and approval, to ensure the safety and stability of the product during use. It realizes the efficient processing and effective utilization of multi-dimensional test data of gas turbine materials, and can effectively solve the technical problem that the requirements for material data are different at different design stages in the gas turbine product design process.

[0006] In a first aspect, this application provides a data processing method, the method comprising:

[0007] Obtain a materials testing dataset, which includes multi-dimensional materials testing data for gas turbine materials;

[0008] The quality assessment value of the material test data for each dimension is obtained through a machine learning model, and the material test data is divided into valid data and invalid data based on the quality assessment value.

[0009] The effective data are integrated into an effective dataset, and a structured dataset of each material test data in the effective dataset is obtained at different temperatures.

[0010] The editing and approval process is invoked to execute the editing and approval process on any structured dataset of material test data. If the structured dataset is approved, the material test dataset corresponding to the structured dataset is stored in the gas turbine design material database.

[0011] In one possible implementation, obtaining a structured dataset of material test data from the valid dataset at different temperatures includes:

[0012] For any material test data in the valid dataset, obtain the statistical average and statistical minimum values ​​at different temperatures to obtain a structured dataset including the statistical average and statistical minimum values ​​of material properties at different temperatures.

[0013] In one possible implementation, obtaining the statistical average at different temperatures includes:

[0014] Obtain the relative average value at different temperatures;

[0015] The relative average value at different temperatures is fitted to the temperature to obtain the fitted value of the relative average value at different temperatures;

[0016] Obtain the arithmetic mean of the test data at the reference temperature to get the statistical average of the reference temperature;

[0017] The fitted value of the relative average at different temperatures is multiplied by the statistical average at the reference temperature to obtain the statistical average at different temperatures.

[0018] In one possible implementation, obtaining the relative average value at different temperatures includes:

[0019] Determine a reference temperature, which is the temperature at which the material's corresponding performance is stable or the temperature at which the material's corresponding performance data is the largest and the highest proportion of high-quality test data.

[0020] Calculate the relative performance values ​​of each batch at different temperatures relative to a reference temperature;

[0021] The average value of the relative performance of each batch at different temperatures is obtained, thus obtaining the relative average value at different temperatures.

[0022] In one possible implementation, obtaining the statistical minimum at different temperatures includes:

[0023] For any given temperature, calculate the difference between the test data of each effective material property at that temperature and the statistical average value at that temperature;

[0024] Based on the data differences at different temperatures, calculate the standard deviation corresponding to each temperature.

[0025] By fitting the trend of the standard deviation with temperature, the fitted and corrected standard deviations at different temperatures are obtained.

[0026] Based on the fitted and corrected standard deviations at different temperatures, the statistical minimum value at each temperature is obtained.

[0027] In one possible implementation, the process of building the machine learning model is as follows:

[0028] Obtain historical test datasets for gas turbine materials;

[0029] The historical test dataset is preprocessed to remove abnormal test data.

[0030] The preprocessed historical test dataset is divided into a training set and a test set, wherein the training set is used for model training and the test set is used for model evaluation.

[0031] Using the chemical composition and test conditions of gas turbine materials as input parameters and material properties as output parameters, a machine learning model based on the support vector machine algorithm is constructed.

[0032] In one possible implementation, the method further includes:

[0033] The parameters of the machine learning model are iteratively adjusted and optimized using the 10-fold cross-validation method to improve the model's prediction accuracy.

[0034] Secondly, this application provides a data processing system, the system comprising:

[0035] The data extraction module is used to acquire a material testing dataset, which includes multi-dimensional material testing data of gas turbine materials.

[0036] The quality assessment module is used to obtain the quality assessment value of the material test data for each dimension through a machine learning model, and to divide the material test data into valid data and invalid data based on the quality assessment value.

[0037] The calculation and processing module is used to integrate the effective data into an effective dataset and obtain a structured dataset of each material test data in the effective dataset at different temperatures.

[0038] The editing and approval module is used to call the editing and approval process, execute the editing and approval process on any structured dataset of material test data, and store the material test dataset corresponding to the structured dataset into the gas turbine design material database if the structured dataset is approved.

[0039] Thirdly, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the data processing method described in any embodiment of the first aspect.

[0040] Fourthly, this application also provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the data processing method described in any embodiment of the first aspect.

[0041] Compared with the prior art, the technical solution provided in this application has the following advantages: The method provided in this application, through four steps—data extraction, quality assessment, calculation processing, and editing and approval—forms a complete data processing system to ensure the safety and stability of the product during use, and achieves efficient processing and effective utilization of multi-dimensional test data of gas turbine materials. Specifically, in the data extraction stage, a comprehensive material test dataset is obtained, providing a rich data foundation for subsequent processing. In the quality assessment stage, a machine learning model is used to accurately divide the material test data into valid and invalid data, ensuring the reliability of the data used subsequently. The calculation processing step integrates the valid data into a valid dataset and further obtains structured datasets of each material test data at different temperatures, meeting the different material data requirements of different design stages of gas turbines. The editing and approval process ensures the accuracy and usability of the data, ensuring that the data finally stored in the gas turbine design material database can provide strong support for gas turbine design, life prediction, and safety assessment. Attached Figure Description

[0042] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0045] Figure 1 A flowchart illustrating a data processing method provided in an embodiment of this application;

[0046] Figure 2 A flowchart illustrating a method for obtaining a statistical average value, provided in an embodiment of this application;

[0047] Figure 3 A flowchart illustrating a method for obtaining a statistical minimum value provided in an embodiment of this application;

[0048] Figure 4 A flowchart illustrating a machine learning model building method provided in an embodiment of this application;

[0049] Figure 5 This is a schematic diagram illustrating the steps of a data processing method provided in an embodiment of this application;

[0050] Figure 6 This is a schematic diagram of the structure of a data processing system provided in an embodiment of this application;

[0051] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0053] The following disclosure provides numerous different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of the invention. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.

[0054] To address the technical challenge of varying material data requirements at different stages of gas turbine product design in existing technologies, this application proposes a data processing method, system, electronic equipment, and storage medium. In the data extraction stage, a comprehensive material test dataset is acquired, providing a rich data foundation for subsequent processing. In the quality assessment stage, a machine learning model is used to accurately distinguish between valid and invalid material test data, ensuring the reliability of the data used later. In the calculation and processing step, valid data is integrated into a valid dataset, and structured datasets of each material test data at different temperatures are further extracted to meet the diverse material data requirements at different stages of gas turbine design. The editing and approval process ensures the accuracy and usability of the data, ensuring that the data ultimately stored in the gas turbine design material database can provide strong support for gas turbine design, life prediction, and safety assessment. Through these four steps—data extraction, quality assessment, calculation and processing, and editing and approval—a complete data processing system is constructed, ensuring the safety and stability of the product during use and achieving efficient processing and effective utilization of multi-dimensional test data for gas turbine materials.

[0055] Figure 1 This is a flowchart illustrating a data processing method provided in an embodiment of this application, as shown below. Figure 1 As shown, the method specifically includes:

[0056] S101. Obtain a material test dataset, which includes multi-dimensional material test data of gas turbine materials;

[0057] Materials testing datasets refer to test data on gas turbine materials across multiple dimensions, including physical properties, elastic properties, tensile properties, high-cycle fatigue, low-cycle fatigue, crack propagation, fracture toughness, creep durability, and oxidation properties. This includes information such as material ID, material grade, material state, chemical composition, furnace batch, and performance data.

[0058] In this embodiment, the material test data is stored in a structured manner, and the structured data is obtained in batches by directly matching the material ID, material grade, and material state.

[0059] S102. Using a machine learning model, obtain the quality assessment value of the material test data for each dimension, and divide the material test data into valid data and invalid data based on the quality assessment value;

[0060] Machine learning models are models built on the support vector machine algorithm that can evaluate the quality of multi-dimensional test data of gas turbine materials.

[0061] In this embodiment, the data quality assessment method based on machine learning obtains the prediction error of each data point through a machine learning model, thereby obtaining a quality assessment value, and then filters out abnormal data points, i.e. invalid data, based on the quality assessment value.

[0062] For example, the prediction error for each data point is obtained by calculating the absolute percentage error between the model's predicted value and the actual test value for each data point, and then the quality of each data point is evaluated based on the data error.

[0063] The formula for calculating the prediction error for each data point is as follows:

[0064] ;

[0065] in, For predicted values, The actual value;

[0066] The prediction error range for each data point is between 0 and 1. The closer the error value is to 1, the higher the degree of data anomaly. When the absolute percentage error (APE) of the data is ≤0.3, it is considered to be of high data quality. If APE>0.3 and APE<0.8, manual intervention is required to determine the data quality. If APE≥0.8, it is directly determined to be an outlier and the data is removed.

[0067] S103. Integrate the effective data into an effective dataset, and obtain a structured dataset of each material test data in the effective dataset at different temperatures;

[0068] To meet the material data requirements for gas turbine design, life prediction, and safety assessment, data processing is also required for the effective data, including the processing of statistical averages and minimum values.

[0069] In this embodiment, the statistical average and statistical minimum values ​​of each material test data in the effective dataset are calculated according to different temperatures. Through data processing, the effective data are integrated into an effective dataset, and a structured dataset including the statistical average and statistical minimum values ​​of material properties at different temperatures is obtained, thereby better meeting the diverse needs of material data at different design stages of gas turbines.

[0070] S104. Invoke the editing and approval process, execute the editing and approval process on any structured dataset of material test data, and if the structured dataset is approved, store the material test dataset corresponding to the structured dataset in the gas turbine design material database.

[0071] The editing, proofreading, and approval process refers to a series of operational procedures for editing, proofreading, and approving structured datasets.

[0072] In this embodiment, the editing and approval process includes multiple rounds of review. Through an intelligent four-level data control process of editing, proofreading, reviewing, and approving, only when the structured dataset passes all stages of the editing and approval process will its corresponding material test dataset be stored in the gas turbine design material database. This ensures the data quality in the database and provides reliable data support for gas turbine-related work. At the same time, the data stored in the database can be easily queried, analyzed, and used later, further improving the efficiency of gas turbine product design and development, realizing the effectiveness of the final design material data, and ensuring the reliability and authority of the design data.

[0073] The data processing method provided in this application embodiment achieves standardized and procedural processing of batch test data of various performance characteristics of gas turbine materials through four core steps: data extraction, quality assessment, calculation processing, and editing, proofreading, approval and effectiveness. In the calculation processing stage, machine learning algorithms are introduced to ensure that the material data for design purposes generated after processing has accuracy, reliability and authority. Through intelligent and procedural processing, a large amount of material data can be processed efficiently and intelligently, which significantly improves the efficiency and accuracy of data processing.

[0074] In an optional embodiment of the present invention, obtaining a structured dataset of material test data from the effective dataset at different temperatures includes:

[0075] For any material test data in the valid dataset, obtain the statistical average and statistical minimum values ​​at different temperatures to obtain a structured dataset including the statistical average and statistical minimum values ​​of material properties at different temperatures.

[0076] In this embodiment, for the material performance data in the effective dataset, the corresponding statistical average and statistical minimum values ​​are calculated separately according to different temperature dimensions. This generates a structured dataset containing the statistical average and statistical minimum values ​​of material performance at each temperature, and associated with core information such as material ID, furnace batch, and test conditions. This satisfies the specific needs of different tasks such as gas turbine design, life prediction, and safety assessment for material data at different temperatures. The structured dataset clearly presents the changes in material performance under different temperature conditions, providing accurate data support for gas turbine design and optimization. Simultaneously, the structured data format facilitates data storage, management, and retrieval, improving data utilization efficiency.

[0077] In order to obtain reliable and accurate average values ​​for design purposes, embodiments of this application propose a method for obtaining statistical average values ​​based on reference temperature and batch trend lines.

[0078] Figure 2 This is a flowchart illustrating a method for obtaining a statistical average value, as provided in an embodiment of this application. Figure 2 As shown, obtaining the statistical average at different temperatures includes:

[0079] S201. Obtain the relative average value at different temperatures;

[0080] In this embodiment, the relative average value refers to the average value of the material performance data at each temperature.

[0081] Specifically, for the material performance data in the valid dataset, the data is grouped according to different temperatures, and the average value of each group is calculated. This average value is used as the relative average value at that temperature, which can accurately reflect the average level of the material's performance under different temperature conditions, providing a basis for subsequent data analysis and processing.

[0082] S202. Fit the relative average value at different temperatures to the temperature to obtain the fitted value of the relative average value at different temperatures;

[0083] In this embodiment, linear fitting or polynomial fitting methods are used to fit the relative average value at different temperatures to the temperature, forming a relative value trend curve. This can more accurately reflect the trend of material properties changing with temperature, eliminate random errors in the data, and make the obtained data more regular and accurate.

[0084] For example, when the material properties and temperature have an approximately linear relationship, a linear fitting method can be used. Let the fitting equation be y = ax + b, where y is the relative average value and x is the temperature. The coefficients a and b are determined by methods such as the least squares method, so as to obtain the fitted value of the relative average value at different temperatures.

[0085] S203. Obtain the arithmetic mean of the test data at the reference temperature to get the statistical average of the reference temperature;

[0086] The reference temperature is typically selected from a representative temperature point observed during actual operation of the gas turbine. It can be a temperature where the material's properties are relatively stable, or a temperature with the most performance data, especially a temperature providing high-quality test data.

[0087] In this embodiment, by performing an arithmetic mean calculation on the test data at the reference temperature, a representative statistical average value can be obtained, which can serve as an important indicator for measuring the performance of the material at that specific temperature.

[0088] For example, if the reference temperature is 500℃, the statistical average value of the reference temperature is obtained by summing the multiple material performance test data at that temperature and then dividing by the number of data.

[0089] S204. Multiply the fitted value of the relative average at different temperatures by the statistical average of the reference temperature to obtain the statistical average at different temperatures.

[0090] The fitted value of the relative average reflects the trend of material properties changing with temperature, while the statistical average of the reference temperature is a representative benchmark value. Multiplying the two allows for a comprehensive consideration of both the changing trend of material properties and the benchmark performance level.

[0091] In this embodiment, the fitted value of the relative average value at different temperatures is multiplied by the statistical average value of the reference temperature. The result can more accurately reflect the performance of the material at different temperatures, providing more accurate data support for the design, life prediction and safety assessment of gas turbines.

[0092] The statistical average value acquisition method provided in this application comprehensively considers the trend of material performance with temperature and its benchmark performance level by processing material performance data at different temperatures. By obtaining the relative average value at different temperatures, and then fitting it to the temperature to obtain the fitted value, and combining it with the statistical average value at the reference temperature, the precise statistical average value at different temperatures is finally obtained. This method not only accurately reflects the average performance level of the material under different temperature conditions, but also demonstrates the law of material performance change with temperature, providing more accurate and valuable data support for gas turbine design, life prediction, and safety assessment.

[0093] In one optional embodiment of the present invention, obtaining the relative average value at different temperatures includes: determining a reference temperature, wherein the reference temperature is the temperature at which the material's corresponding performance is stable or the temperature at which the material's corresponding performance data is the most abundant and the proportion of high-quality test data is the highest; calculating the relative performance value of each batch at different temperatures relative to the reference temperature; obtaining the average value of the relative performance value of each batch at different temperatures, thereby obtaining the relative average value at different temperatures.

[0094] In this embodiment, for the furnace batch trend line, a furnace batch with a complete temperature series is selected, and the temperature values ​​of each furnace batch are compared with the reference temperature to obtain the relative value, thereby constructing a relative value trend curve.

[0095] Specifically, the first step is to determine a reference temperature. This reference temperature is selected based on the material's stable performance or the temperature with the most abundant and highest-quality performance data, ensuring its representativeness and reliability. Then, for each batch of materials, the relative performance values ​​at different temperatures are calculated relative to the reference temperature. This clearly shows how the performance of each batch of materials changes relative to the reference temperature. Next, the relative performance values ​​of each batch at different temperatures are summarized, and their average value is calculated. This relative average value more accurately reflects the average performance level of the material under different temperature conditions, providing a data foundation for further data analysis and processing. It also better meets the material data requirements in gas turbine design, life prediction, and safety assessment, contributing to improved accuracy and reliability in gas turbine design and providing strong support for stable gas turbine operation.

[0096] For example, the specific method for calculating the statistical average includes the following steps:

[0097] Step 1: Determine the reference temperature;

[0098] Step 2: Select furnace batches for the entire temperature series and calculate the relative value of each furnace batch with respect to the reference temperature;

[0099] Step 3: Calculate the relative average value of each temperature. The specific method is to calculate the arithmetic mean of the relative values ​​of each batch.

[0100] Step 4: Fit the calculated relative average value to temperature T. The fitting process needs to consider the variation of the corresponding performance with temperature. After fitting, the fitted value of the relative average value at each temperature can be obtained.

[0101] Step 5: Calculate the reference temperature T ref The statistical average is calculated by taking the arithmetic mean of the test data at all reference temperatures.

[0102] Step 6: Multiply the fitted value of the relative average at each temperature with the average at the reference temperature to calculate the statistical average at each temperature, which is the design average.

[0103] By introducing the concepts of fitting and reference temperature statistical averages, data processing becomes more scientific and reasonable, better meeting the diverse material data needs of gas turbines in practical applications. This helps improve the design efficiency and quality of gas turbines, reduce design costs and risks, and provide strong technical support for the advancement of the gas turbine industry.

[0104] The design philosophy for the safe operation of heavy-duty gas turbine products adopts a 95% confidence level and a 99% probability (C95 / P99). Therefore, in life prediction and safety assessment, the conservative value (minimum value) of material properties also adopts the same philosophy, that is, through a large amount of test data, the statistically significant lower limit of C95 / P99 is selected as the minimum value of material property data.

[0105] The materials used in gas turbines mainly include cast high-temperature alloys, wrought high-temperature alloys, stainless steel, structural steel, and alloy steel, whose material properties conform to a normal distribution. Therefore, when calculating the minimum value of a heavy-duty gas turbine, the C95 / P99 one-sided confidence lower bound method based on the normal distribution is used.

[0106] Since the dispersion of material performance data for heavy-duty gas turbines varies at different temperatures, this application proposes a scheme to correlate the standard deviation of material performance with temperature. After calculating the standard deviation of material performance at different temperatures, a fitting process is further performed to obtain a corrected standard deviation.

[0107] Figure 3 This is a flowchart illustrating a method for obtaining a statistical minimum value according to an embodiment of this application, as shown below. Figure 3 As shown, obtaining the statistical minimum value at different temperatures includes:

[0108] S301. For any given temperature, calculate the difference between the test data of each effective material performance at that temperature and the statistical average value at that temperature;

[0109] In this embodiment, by calculating the data difference, the degree of deviation of each effective material performance test data from the statistical average value at that temperature can be seen intuitively.

[0110] Specifically, for each valid material performance test data, the statistical average value at that temperature is subtracted from the value. If the difference is positive, it means that the data is higher than the average value; if it is negative, it means that the data is lower than the average value.

[0111] S302. Based on the data differences at different temperatures, calculate the standard deviation corresponding to each temperature.

[0112] In this embodiment, the standard deviation is the square root of the variance, used to measure the degree of data dispersion. By taking the square root of the variance calculated at different temperatures, the standard deviation corresponding to different temperatures can be obtained. This provides a more intuitive view of the dispersion of material performance data relative to the statistical mean at different temperatures, and provides a more accurate data foundation for subsequent calculations and analyses.

[0113] S303. Fit the trend of the standard deviation with temperature to obtain the fitted and corrected standard deviation at different temperatures;

[0114] In this embodiment, appropriate fitting methods such as linear fitting and polynomial fitting are used to fit the standard deviation to the temperature, thereby obtaining the fitted and corrected standard deviation at different temperatures.

[0115] For example, when the standard deviation and temperature have an approximately linear relationship, let the fitting equation be y=mx+n, where y is the standard deviation and x is the temperature. The coefficients m and n are determined by methods such as the least squares method, so as to obtain the fitted and corrected standard deviation at different temperatures. This eliminates random fluctuations in the data and makes the standard deviation more accurately reflect the discrete characteristics of material performance data at different temperatures, providing a more reliable data basis for subsequent calculation of the statistical minimum.

[0116] S304. Based on the fitted and corrected standard deviation at different temperatures, obtain the statistical minimum value at different temperatures.

[0117] In this embodiment, based on the concept of 95% confidence level and 99% probability (C95 / P99) for the safe service of heavy-duty gas turbine products, the statistical minimum value at different temperatures is obtained by combining the fitted and corrected standard deviation at different temperatures and the C95 / P99 one-sided confidence lower limit method of normal distribution.

[0118] Specifically, using the fitted and corrected standard deviation, and through specific statistical formulas and algorithms, the lower limit of the material performance data corresponding to each temperature under the C95 / P99 requirements is calculated, which is the statistical minimum value at that temperature.

[0119] The statistical minimum value acquisition method provided in this application first calculates the difference between the effective material performance test data and the statistical average value at each temperature to clarify the degree of data deviation; then, it calculates the standard deviation based on the difference to intuitively display the data dispersion; next, it fits the standard deviation to the temperature and corrects the standard deviation to eliminate random fluctuations and make it more accurately reflect the dispersion characteristics; finally, it combines the safe service concept of heavy-duty gas turbines and the one-sided confidence lower limit method of normal distribution to obtain the statistical minimum value at different temperatures; by comprehensively considering the dispersion difference of material performance data at different temperatures, it can provide more conservative material performance data that meets actual needs for the life prediction and safety assessment of heavy-duty gas turbines, ensuring the safety and reliability of gas turbines during operation.

[0120] For example, the specific method for calculating the statistical minimum includes the following steps:

[0121] Step 1: Calculate the difference between the performance data at each temperature and the average value at the corresponding temperature;

[0122] Step 2: Calculate the standard deviation of each temperature point;

[0123] Step 3: Fit the standard deviation to temperature. Linear, bilinear, polynomial, exponential, and power functions can be used for fitting. After fitting, the fitted and corrected standard deviation s(T) at each temperature is obtained.

[0124] Step 4: Calculate the minimum value of the material properties based on the average value. The calculation formula is as follows:

[0125] ;

[0126] in, x represents the minimum material properties at various temperatures. mean (T) represents the statistical average of material properties at various temperatures, k stat With a 95% confidence level and a 99% probability (C95 / P99), it is generally obtained by looking up a table, and s(T) is the standard deviation after fitting correction.

[0127] The processing method based on temperature correlation and fitting correction demonstrates the scientific rigor of data processing, which helps to improve the accuracy of gas turbine design and operation, reduce the risks caused by uncertainties in material properties, and provide strong technical support for the stable development of the gas turbine industry.

[0128] Figure 4 This is a flowchart illustrating a machine learning model building method provided in an embodiment of this application, as shown below. Figure 4 As shown, the process of establishing the machine learning model is as follows:

[0129] S401. Obtain historical test datasets for gas turbine materials;

[0130] Historical test datasets refer to the collection of test data accumulated during the long-term operation and testing of gas turbines, covering various aspects such as material properties and operating parameters. This includes the performance of different types of materials under various operating conditions, such as the strength, toughness, and fatigue life of materials under different temperatures, pressures, and speeds, as well as the operating parameters of the gas turbines, such as power output and efficiency.

[0131] In this embodiment, by collecting and organizing these historical test data, a comprehensive dataset can be constructed, providing a data foundation for training machine learning models.

[0132] S402. Preprocess the historical test dataset and delete abnormal test data in the historical test dataset;

[0133] Abnormal test data may be caused by test equipment failure, human error, or other reasons, which can negatively affect the training of machine learning models and reduce the accuracy and reliability of the models.

[0134] In this embodiment, the historical test dataset is cleaned, and abnormal data is identified and deleted.

[0135] Specifically, statistical analysis methods, such as standard deviation-based methods, can be used to consider data that deviates from the mean by a certain multiple of standard deviation as outliers. Alternatively, machine learning algorithms, such as the Isolation Forest algorithm and the Local Anomaly Factor algorithm, can be used to automatically detect and label outliers and then remove them from the dataset.

[0136] S403. Divide the preprocessed historical test dataset into a training set and a test set, wherein the training set is used for model training and the test set is used for model evaluation.

[0137] In this embodiment, the preprocessed historical test dataset is divided according to a preset ratio. Most of the data is used as the training set, which is used to enable the machine learning model to learn the relationship and rules between material properties and operating parameters. By continuously adjusting the model's parameters, the model can predict material properties as accurately as possible. A small portion of the data is used as the test set, which is used to evaluate the performance of the trained model. By inputting the input data from the test set into the model, the model's prediction results are obtained and compared with the actual results in the test set. Evaluation indicators such as mean squared error and mean absolute error are calculated to measure the model's accuracy and generalization ability.

[0138] For example, 80% of the data is used as the training set and 20% of the data is used as the test set.

[0139] S404. Using the chemical composition and test conditions of the gas turbine materials as input parameters and the material properties as output parameters, construct a machine learning model based on the support vector machine algorithm.

[0140] Support Vector Machine (SVM) is a powerful machine learning algorithm with good generalization ability and high prediction accuracy.

[0141] In this embodiment, a machine learning model based on the support vector machine algorithm is constructed using the chemical composition and test conditions of the gas turbine material as input parameters and the material properties as output parameters to effectively predict the performance of the gas turbine material.

[0142] The machine learning model building method provided in this application comprehensively considers multiple aspects such as data acquisition, processing, and model building, forming a complete and scientific process. It can obtain a machine learning model that can quickly and accurately predict material properties based on the input chemical composition and test conditions, providing important decision-making basis for the design, maintenance, and optimization of gas turbines.

[0143] In an optional embodiment of the present invention, the method further includes:

[0144] The parameters of the machine learning model are iteratively adjusted and optimized using the 10-fold cross-validation method to improve the model's prediction accuracy.

[0145] In this embodiment, based on the support vector machine algorithm, ten-fold cross-validation is used to tune the model parameters, and the mean absolute percentage error (MAPE) is used as the evaluation function to establish a high-precision machine learning regression model.

[0146] The formula for calculating MAPE is as follows:

[0147] ;

[0148] Where n is the amount of data, For predicted values, This is the actual value.

[0149] It should be noted that the MAPE of the established machine learning model should be below 15% in order to be considered to have high predictive accuracy and be applicable to the actual prediction of gas turbine material performance.

[0150] Figure 5 This is a schematic diagram illustrating the steps of a data processing method provided in an embodiment of this application, as shown below. Figure 5 As shown, the method specifically includes the following steps:

[0151] Step 1: Data Input:

[0152] Collect multi-dimensional raw test data of gas turbine materials.

[0153] Step 2: Data Preprocessing Stage

[0154] First, the raw test data is "structured" to extract key information such as material performance data from the unstructured raw records. Then, the extracted information is organized into a "structured dataset" in a unified format to achieve data standardization. Finally, combined with "assessment based on machine learning models", the structured data is screened for quality, outliers are removed, and "high-quality data" is obtained.

[0155] Step 3: Data Calculation and Processing Stage

[0156] Based on high-quality data, complete the statistical calculation of core performance indicators: through the "Data Calculation and Processing" module, simultaneously execute "Average Statistical Calculation" and "Minimum Statistical Calculation" to finally obtain "Design Data".

[0157] Step Four: Data Editing and Storage Stage

[0158] A "design dataset to be effective" is formed, and then the compliance and accuracy of the data are verified through the "editing, proofreading, review and approval process" (which includes multiple stages of editing, proofreading, review and approval). The approved design data is finally stored in the "gas turbine design material database" for direct use in the subsequent design of gas turbine components.

[0159] The artificial intelligence monitoring method for heavy-duty gas turbine material performance test data provided in this application integrates machine learning algorithms and programmed processing techniques through four key processes: data extraction, quality assessment, calculation and processing, and data activation. This method efficiently and effectively processes batch test data from multiple furnace batches and multiple samples into accurate, reliable, and authoritative material data for design, providing a solid data foundation for heavy-duty gas turbine design simulation, life prediction, and safety assessment.

[0160] The machine learning algorithm introduced in data quality assessment correlates material composition with material properties. The determination of material property data quality is no longer a purely statistical method. Combining it with material composition is more in line with the characteristics of the material itself, and the assessment results are more accurate and reliable. It achieves highly reliable and accurate data quality assessment and ensures the quality of the test data involved in the processing.

[0161] In data processing, the calculation of average material properties incorporates reference temperature and batch trend lines, which better reflects the changing characteristics of gas turbine materials. Compared to the traditional method of using the arithmetic mean of test values, the processing method proposed in this application yields more scientific and reasonable results. Furthermore, the proposed reference temperature and batch trend line processing logic requires testing multiple batches and samples at the reference temperature, while only a few samples from the same batch need to be tested at other temperatures. This reduces the number of tests, lowers testing costs, shortens the testing cycle, and improves the efficiency of material property data acquisition while ensuring data quality.

[0162] In the minimum value calculation process, combined with the design requirements of heavy-duty gas turbines, the minimum value with a confidence level of 95% and a probability of 99% is defined as the minimum value of the material performance for design. Considering the inherent characteristic that the dispersion of material performance varies at different temperatures, a processing logic for the standard deviation changing with temperature is proposed. The standard deviation is then fitted and corrected in combination with the actual calculated standard deviation to obtain a more scientific and reasonable standard deviation. This results in a more scientific, accurate, and reliable minimum value of the material for design, ensuring the safe operation of gas turbine products while maximizing the exploitation and utilization of material characteristics, thus maximizing the utilization value of the material.

[0163] Figure 6 This is a schematic diagram of the structure of a data processing system provided in an embodiment of this application, such as... Figure 6 As shown, the system specifically includes:

[0164] Data extraction module 601 is used to acquire material test dataset, which includes multi-dimensional material test data of gas turbine materials;

[0165] The quality assessment module 602 is used to obtain the quality assessment value of the material test data for each dimension through a machine learning model, and to divide the material test data into valid data and invalid data based on the quality assessment value;

[0166] The calculation and processing module 603 is used to integrate the effective data into an effective dataset and obtain a structured dataset of the test data of each material in the effective dataset at different temperatures.

[0167] The editing and approval module 604 is used to call the editing and approval process, execute the editing and approval process on any structured dataset of material test data, and store the material test dataset corresponding to the structured dataset into the gas turbine design material database if the structured dataset is approved.

[0168] In one possible implementation, the calculation processing module 603 is further configured to obtain the statistical average and statistical minimum values ​​at different temperatures for any material test data in the effective dataset, thereby obtaining a structured dataset including the statistical average and statistical minimum values ​​of material properties at different temperatures.

[0169] In one possible implementation, the calculation processing module 603 is further configured to obtain the relative average value at different temperatures; fit the relative average value at different temperatures to the temperature respectively to obtain the fitted value of the relative average value at different temperatures; obtain the arithmetic mean of the test data at the reference temperature to obtain the statistical average value of the reference temperature; and multiply the fitted value of the relative average value at different temperatures by the statistical average value of the reference temperature to obtain the statistical average value at different temperatures respectively.

[0170] In one possible implementation, the calculation processing module 603 is further configured to determine a reference temperature, which is either the temperature at which the material's corresponding performance is stable or the temperature at which the material's corresponding performance data is most abundant and the proportion of high-quality test data is highest; calculate the relative performance values ​​of each batch at different temperatures relative to the reference temperature; obtain the average value of the relative performance values ​​of each batch at different temperatures, and obtain the relative average value at different temperatures.

[0171] In one possible implementation, the calculation processing module 603 is further configured to calculate, for any given temperature, the difference between the effective material performance test data at that temperature and the statistical average value at that temperature; calculate the standard deviation corresponding to different temperatures based on the data differences at different temperatures; fit the trend of the standard deviation with temperature to obtain the fitted and corrected standard deviation at different temperatures; and obtain the statistical minimum value at different temperatures based on the fitted and corrected standard deviation at different temperatures.

[0172] In one possible implementation, the system further includes a module 605 (not shown in the figure) for acquiring historical test datasets of gas turbine materials; preprocessing the historical test datasets to remove abnormal test data; dividing the preprocessed historical test datasets into training and testing sets, wherein the training set is used for model training and the testing set is used for model evaluation; and constructing a machine learning model based on the support vector machine algorithm using the chemical composition and test conditions of the gas turbine materials as input parameters and the material properties as output parameters.

[0173] In one possible implementation, the establishment module 605 is further configured to iteratively adjust and optimize the parameters of the machine learning model using ten-fold cross-validation to improve the model's prediction accuracy.

[0174] The data processing system provided in this embodiment can be as follows: Figure 6 The data processing system shown can perform, for example... Figures 1-5 All steps of data processing in China, thereby achieving Figures 1-5 For details on the technical effects of the data processing shown, please refer to [link / reference]. Figures 1-5 The relevant descriptions are presented concisely and will not be elaborated upon here.

[0175] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0176] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 7As shown, this application provides an electronic device including a processor 701, a communication interface 702, a memory 703, and a communication bus 704. The processor 701, communication interface 702, and memory 703 communicate with each other via the communication bus 704. The memory 703 stores computer programs. When the processor 701 executes the program stored in the memory 703, it implements the data processing steps provided in any of the aforementioned method embodiments.

[0177] A materials testing dataset is obtained, comprising multi-dimensional materials testing data for gas turbine materials. A quality assessment value for each dimension of the materials testing data is obtained using a machine learning model, and the data is divided into valid and invalid data based on the quality assessment value. The valid data is integrated into a valid dataset, and a structured dataset of each materials testing data in the valid dataset is obtained at different temperatures. An editing and approval process is invoked, executing the editing and approval process on the structured dataset of any materials testing data. If the structured dataset is approved, the corresponding materials testing dataset is stored in the gas turbine design materials database.

[0178] In one possible implementation, for any material test data in the effective dataset, the statistical average and statistical minimum values ​​at different temperatures are obtained to obtain a structured dataset including the statistical average and statistical minimum values ​​of material properties at different temperatures.

[0179] In one possible implementation, the relative average values ​​at different temperatures are obtained; the relative average values ​​at different temperatures are fitted to the temperatures respectively to obtain fitted values ​​of the relative average values ​​at different temperatures; the arithmetic mean of the test data at a reference temperature is obtained to obtain the statistical average value of the reference temperature; the fitted values ​​of the relative average values ​​at different temperatures are multiplied by the statistical average value of the reference temperature to obtain the statistical average values ​​at different temperatures respectively.

[0180] In one possible implementation, a reference temperature is determined, which is either the temperature at which the material's corresponding performance is stable or the temperature at which the material's corresponding performance data is most abundant and the proportion of high-quality test data is highest; the relative performance values ​​of each batch at different temperatures relative to the reference temperature are calculated; the average value of the relative performance values ​​of each batch at different temperatures is obtained, thus obtaining the relative average value at different temperatures.

[0181] In one possible implementation, for any given temperature, the difference between the effective material performance test data at that temperature and the statistical average value at that temperature is calculated; based on the data differences at different temperatures, the standard deviation corresponding to different temperatures is calculated respectively; the trend of the standard deviation with temperature is fitted to obtain the fitted and corrected standard deviation at different temperatures; based on the fitted and corrected standard deviation at different temperatures, the statistical minimum value at different temperatures is obtained.

[0182] In one possible implementation, a historical test dataset of gas turbine materials is obtained; the historical test dataset is preprocessed to remove abnormal test data; the preprocessed historical test dataset is divided into a training set and a test set, wherein the training set is used for model training and the test set is used for model evaluation; a machine learning model based on the support vector machine algorithm is constructed using the chemical composition and test conditions of the gas turbine materials as input parameters and the material properties as output parameters.

[0183] In one possible implementation, the parameters of the machine learning model are iteratively adjusted and optimized using ten-fold cross-validation to improve the model's prediction accuracy.

[0184] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented using software plus a general-purpose hardware platform, or of course, using hardware. Based on this understanding, the above technical solutions, in essence or the parts that contribute to the related technology, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0185] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0186] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. An artificial intelligence monitoring method for test data of heavy-duty gas turbine materials, characterized in that, include: Obtain a materials testing dataset, which includes multi-dimensional materials testing data for gas turbine materials; The quality assessment value of material test data for each dimension is obtained through a machine learning model, and the material test data is divided into valid data and invalid data based on the quality assessment value. In the machine learning-based data quality assessment method, the prediction error of each data point is obtained through a machine learning model to obtain the quality assessment value, and abnormal data points are screened out as invalid data based on the quality assessment value. The effective data is integrated into an effective dataset, and a structured dataset of material test data in the effective dataset at different temperatures is obtained. This includes: for any material test data in the effective dataset, obtaining the statistical average and statistical minimum at different temperatures, resulting in a structured dataset including the statistical average and statistical minimum of material performance at different temperatures; obtaining the statistical minimum at different temperatures includes: for any temperature, calculating the data difference between each effective material performance test data at that temperature and the statistical average at that temperature; calculating the standard deviation corresponding to different temperatures based on the data difference at different temperatures; fitting the trend of the standard deviation with temperature to obtain the fitted and corrected standard deviation at different temperatures; and obtaining the statistical minimum at different temperatures based on the fitted and corrected standard deviation at different temperatures. The editing and approval process is invoked to execute the editing and approval process on any structured dataset of material test data. If the structured dataset is approved, the material test dataset corresponding to the structured dataset is stored in the gas turbine design material database.

2. The method according to claim 1, characterized in that, The process of obtaining statistical averages at different temperatures includes: Obtain the relative average value at different temperatures; The relative average value at different temperatures is fitted to the temperature to obtain the fitted value of the relative average value at different temperatures; Obtain the arithmetic mean of the test data at the reference temperature to get the statistical average of the reference temperature; The fitted value of the relative average at different temperatures is multiplied by the statistical average at the reference temperature to obtain the statistical average at different temperatures.

3. The method according to claim 2, characterized in that, Obtaining the relative average value at different temperatures includes: Determine a reference temperature, which is the temperature at which the material's corresponding performance is stable or the temperature at which the material's corresponding performance data is the largest and the highest proportion of high-quality test data. Calculate the relative performance values ​​of each batch at different temperatures relative to a reference temperature; The average value of the relative performance of each batch at different temperatures is obtained, thus obtaining the relative average value at different temperatures.

4. The method according to claim 1, characterized in that, The process of establishing the machine learning model is as follows: Obtain historical test datasets for gas turbine materials; The historical test dataset is preprocessed to remove abnormal test data. The preprocessed historical test dataset is divided into a training set and a test set, wherein the training set is used for model training and the test set is used for model evaluation. Using the chemical composition and test conditions of gas turbine materials as input parameters and material properties as output parameters, a machine learning model based on the support vector machine algorithm is constructed.

5. The method according to claim 1, characterized in that, The method further includes: The parameters of the machine learning model are iteratively adjusted and optimized using the 10-fold cross-validation method to improve the model's prediction accuracy.

6. An artificial intelligence monitoring system for heavy-duty gas turbine material performance test data, employing the artificial intelligence monitoring method for heavy-duty gas turbine material performance test data as described in any one of claims 1-5, characterized in that, include: The data extraction module is used to acquire a material testing dataset, which includes multi-dimensional material testing data of gas turbine materials. The quality assessment module is used to obtain the quality assessment value of the material test data for each dimension through a machine learning model, and to divide the material test data into valid data and invalid data based on the quality assessment value. The calculation and processing module is used to integrate the effective data into an effective dataset and obtain a structured dataset of each material test data in the effective dataset at different temperatures. The editing and approval module is used to call the editing and approval process, execute the editing and approval process on any structured dataset of material test data, and store the material test dataset corresponding to the structured dataset into the gas turbine design material database if the structured dataset is approved.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the artificial intelligence monitoring method for test data of heavy-duty gas turbine materials as described in any one of claims 1-5.

8. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the artificial intelligence monitoring method for test data of heavy-duty gas turbine materials as described in any one of claims 1-5.

Images