Gas turbine performance prediction method, device and system and computer readable storage medium
By preprocessing and feature extraction of gas turbine performance sampling data, basic and correction models are built, combined with machine learning algorithms, the problem of difficulty in accurately modeling of unified models in gas turbine performance prediction is solved, and the prediction accuracy is improved.
Patent Information
- Application Number
- CN202411182837.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-31
AI Technical Summary
In the existing gas turbine performance prediction methods, it is difficult to balance the accuracy between multiple data points with a unified gas turbine data model, which leads to difficulty in accurately modeling different working conditions, thereby reducing the prediction accuracy of some working conditions.
By preprocessing and feature extraction of performance sampling data, basic models and working condition correction models are built, combined with machine learning algorithms, combined models are established to improve prediction accuracy. The specific steps include data cleaning, normalization processing, feature extraction and clustering, and construct a correction model for different working conditions based on the clustering results, and predicting it in combination with the basic model.
It improves the accuracy of gas turbine performance prediction, can model and predict different working conditions more accurately, and solves the problem of difficulty in balancing the accuracy between multiple data points of a unified model.
Smart Images
Figure CN119226765A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of power machinery, and in particular to a gas turbine performance prediction method, device, system, computer equipment, and computer-readable storage medium. Background Art
[0002] Gas turbines have the advantages of high efficiency, fast start-up, short construction period, low investment, and low environmental pollution. They are widely used in aviation, navigation, power generation and other fields. Therefore, the performance of gas turbines is of great significance in practical applications. Since gas turbines are highly nonlinear, their actual working process is not easy to be mastered by people. Therefore, by establishing a gas turbine model, the power and exhaust temperature of the gas turbine can be predicted to judge the performance of the gas turbine.
[0003] The prediction results of gas turbine performance can be obtained through a unified gas turbine model. However, the use of a unified gas turbine data model involves the balance of accuracy among many data points, which makes it difficult to accurately model different operating conditions, resulting in low prediction accuracy under some operating conditions. Summary of the invention
[0004] An embodiment of the present application provides a method for predicting gas turbine performance, so as to at least solve the problem that the use of a unified gas turbine data model in the related art involves balancing the accuracy among numerous data points, making it difficult to accurately model different operating conditions, resulting in low prediction accuracy under some operating conditions.
[0005] In a first aspect, an embodiment of the present application provides a method for predicting gas turbine performance, comprising:
[0006] The performance sampling data is processed to obtain preprocessed data, the preprocessed data is used as training data, and a basic model for performance prediction is constructed according to a preset machine learning algorithm;
[0007] Extracting features from the performance sampling data to obtain feature variables, and clustering the feature variables to obtain clustering results, wherein the clustering results represent different working conditions divided based on the feature variables;
[0008] Based on the basic model and clustering results, a modified model corresponding to each type of working condition is constructed;
[0009] A prediction model is obtained based on the basic model and the modified model, and a performance prediction value is obtained based on the performance sampling data to be predicted through the prediction model.
[0010] In one embodiment, obtaining preprocessed data includes:
[0011] Selecting the data to be analyzed from the performance sampling data with a preset length window, and obtaining the standard deviation and average value of the data to be analyzed to determine whether the data to be analyzed is in a steady state;
[0012] If the data to be analyzed is in a steady state, the data to be analyzed is cleaned of abnormal data, wherein the abnormal data includes data that deviates from a preset range and outlier data;
[0013] The cleaned data is normalized to obtain preprocessed data.
[0014] In one embodiment, obtaining a characteristic variable includes:
[0015] Obtain the correlation coefficients between different categories of performance sampling data and between performance sampling data and performance prediction values;
[0016] When the absolute value of the correlation coefficient is greater than or equal to a preset threshold, the performance sampling data of the corresponding category is used as a feature variable.
[0017] In one embodiment, based on the basic model and the clustering results, a correction model corresponding to each type of working condition is constructed, including:
[0018] Obtain basic prediction values of basic model outputs based on performance sampling data;
[0019] Determine the deviation between the target value and the basic prediction value for each type of operating condition in the clustering results;
[0020] Based on the deviation, a correction model corresponding to each type of working condition is constructed through a preset machine learning algorithm.
[0021] In one embodiment, obtaining a performance prediction value according to the performance sampling data to be predicted by using a prediction model includes:
[0022] Based on the performance sampling data to be predicted, a basic prediction value and a deviation prediction value in the prediction model are obtained;
[0023] The base prediction value and the deviation prediction value are added together to obtain the performance prediction value.
[0024] In one embodiment, before obtaining the basic prediction value and the deviation prediction value in the prediction model based on the performance sampling data to be predicted, the following steps are included:
[0025] Determine the clustering result of the performance sampling data to be predicted;
[0026] Based on the clustering result of the performance sampling data to be predicted, a correction model corresponding to the performance sampling data to be predicted is determined.
[0027] In one embodiment, the preset machine learning algorithm includes linear regression, polynomial regression, support vector regression, neural network, etc., which is characterized in that the preprocessed data is used as training data, and a basic model for performance prediction is constructed according to the preset machine learning algorithm, including:
[0028] Based on the training data and the preset machine learning algorithms, the basic prediction values of different machine learning algorithms are obtained;
[0029] Evaluate the basic prediction value according to the preset evaluation method, determine the machine learning algorithm, and build the corresponding basic model.
[0030] In a second aspect, an embodiment of the present application provides a gas turbine performance prediction device, comprising:
[0031] The first construction module is used to process the performance sampling data to obtain preprocessed data, use the preprocessed data as training data, and construct a basic model for performance prediction according to a preset machine learning algorithm;
[0032] A clustering module is used to extract features from the performance sampling data to obtain feature variables, and to perform clustering processing on the feature variables to obtain clustering results, wherein the clustering results represent different working conditions divided based on the feature variables;
[0033] The second building module is used to build a correction model corresponding to each type of working condition based on the basic model and the clustering results;
[0034] The acquisition module is used to obtain a prediction model based on the basic model and the modified model, and obtain the performance prediction value based on the performance sampling data to be predicted through the prediction model.
[0035] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a gas turbine performance prediction method as described in the first aspect above is implemented.
[0036] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a gas turbine performance prediction method as described in the first aspect above.
[0037] A gas turbine performance prediction method provided in an embodiment of the present application has at least the following technical effects.
[0038] By building a basic model based on all performance sampling data, the performance sampling data is divided into working conditions through a clustering algorithm, and a working condition correction model is built on each working condition, so that the basic model and the working condition correction model constitute a combined model. In this combined model, the performance sampling data to be predicted can be corrected according to the working condition, thereby improving the accuracy of the performance prediction value. This solves the problem that the use of a unified gas turbine data model involves the balance of accuracy among many data points, making it difficult to accurately model different working conditions, resulting in low prediction accuracy in some working conditions.
[0039] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0041] Figure 1 is a flow chart of a method for predicting gas turbine performance according to an exemplary embodiment;
[0042] Figure 2 is a flow chart of performing data processing on performance sampling data to obtain pre-processed data according to an exemplary embodiment;
[0043] Figure 3 is a flowchart of building a prediction model according to an exemplary embodiment;
[0044] Figure 4 is a flow chart of a method for predicting gas turbine performance according to an exemplary embodiment;
[0045] Figure 5 is a block diagram of a gas turbine performance prediction device according to an exemplary embodiment;
[0046] Figure 6 is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION
[0047] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.
[0048] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.
[0049] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.
[0050] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.
[0051] In a first aspect, an embodiment of the present application provides a method for predicting gas turbine performance. Figure 1 is a flow chart of a method for predicting gas turbine performance according to an exemplary embodiment. Figure 1As shown, a gas turbine performance prediction method includes:
[0052] Step S101: Process the performance sampling data to obtain preprocessed data, use the preprocessed data as training data, and build a basic model for performance prediction according to a preset machine learning algorithm.
[0053] The performance sampling data is required according to the actual scenario, wherein the performance sampling data includes compressor inlet temperature, compressor inlet pressure, inlet dew point temperature, turbine exhaust pressure, gas temperature, gas pressure, gas turbine speed, opening of inlet guide vanes, fuel flow, compressor exhaust temperature, compressor exhaust pressure, turbine exhaust temperature, and gas turbine power. In addition, the performance sampling data can also be selected according to the operation requirements. In step S101, the compressor inlet pressure, compressor inlet temperature, inlet dew point temperature, opening of inlet guide vanes, and fuel quantity can be selected as input variables of the basic model, and the gas turbine power, turbine exhaust temperature, and compressor exhaust pressure can be selected as output variables of the basic model.
[0054] Preset machine learning algorithms include linear regression, polynomial regression, support vector regression, neural network, etc.
[0055] The preprocessed data is used as training data, and based on the training data and the preset machine learning algorithm, the correlation between the input variables and the output variables is constructed to obtain the basic prediction values obtained by different machine learning algorithms.
[0056] For example, the input variables in the training data are taken as input, and an optimal hyperplane is found through the support vector regression algorithm to regress the data so that most of the data points are located in a preset ε-insensitive band, and the predicted value obtained by the algorithm is used to calculate the loss function with the output variable in the training data. When the value of the loss function is greater than a preset threshold, the position of the support vector and the corresponding multiplier are adjusted to regain the predicted value until the maximum number of iterations is reached or the value of the loss function is less than or equal to the preset threshold; when the value of the loss function is less than or equal to the preset threshold, the basic prediction value is output.
[0057] For example, the input variables in the training data are used as the neural network input, and the mean square error of the output variables in the training data is used as the loss function. When the value of the loss function is greater than a preset threshold, the parameters of the model are adjusted until the maximum number of iterations is reached or the value of the loss function is less than or equal to the preset threshold, and the basic prediction value is output.
[0058] The basic prediction value is evaluated according to the preset evaluation method, and the machine learning algorithm used to construct the basic model is determined based on the evaluation results. The basic model of the corresponding algorithm is constructed, where the input variables include compressor inlet pressure, compressor inlet temperature, inlet dew point temperature, opening of inlet guide vanes, fuel quantity, etc., and the output variables include gas turbine power, turbine exhaust temperature, compressor exhaust pressure, etc.
[0059] Optionally, the preset evaluation method is to evaluate one or more of the mean absolute error, mean square error, mean absolute percentage error, maximum mean absolute percentage error, and coefficient of determination. As an example, the basic prediction value is evaluated using the mean absolute percentage error, and the evaluation result is confirmed based on the smaller mean absolute percentage error as the evaluation principle. In practical applications, indicators such as the maximum error, root mean square error, and coefficient of determination can also be used as references.
[0060] in, Figure 2 is a flow chart of performing data processing on performance sampling data to obtain pre-processed data according to an exemplary embodiment. Figure 2 As shown, the performance sampling data is processed to obtain preprocessed data, and the specific steps include:
[0061] Step S1011: select data to be analyzed from the performance sampling data using a preset length window, and obtain the standard deviation and average value of the data to be analyzed to determine whether the data to be analyzed is in a steady state.
[0062] Select the data to be analyzed from the performance sampling data in the preset length window, and calculate the standard deviation and average value of the data to be analyzed selected in the preset length window. Compare the obtained standard deviation and average value. If the standard deviation is greater than the preset percentage value of the average value, the data to be analyzed fluctuates greatly and is not in a steady state. Delete the data that is not in a steady state; if the standard deviation is less than the preset percentage value of the average value, the data to be analyzed fluctuates slightly and is in a steady state. Perform average filtering on the data to be analyzed and output the data point. After outputting the data point, move one data point backward and repeat the above operation until all the performance sampling data are traversed.
[0063] For example, there are m performance sampling data in total, with a preset length of n (m, n are both positive integers, and m is greater than n). The first data to the n+1th data are obtained from the m data to form the data to be analyzed. The standard deviation and the average value of the data to be analyzed are calculated, and the standard deviation and the average value are compared. If the standard deviation is greater than 5% of the average value, the data to be analyzed is not in a steady state, and the data that is not in a steady state is deleted; if the standard deviation is less than 5% of the average value, the data to be analyzed is in a steady state, and the data is averaged and filtered, and the filtered data is output. After outputting the data, move one data back and repeat the above operation until m data are traversed.
[0064] Step S1012: If the data to be analyzed is in a steady state, then the data to be analyzed is cleaned of abnormal data, wherein the abnormal data includes data deviating from a preset range and outlier data.
[0065] The data to be analyzed is in a steady state, and the data that deviates from the preset range, outlier data, and data with missing values are deleted.
[0066] Step S1013: normalize the cleaned data to obtain preprocessed data.
[0067] The cleaned data is subjected to data unit unification and normalization processing to obtain preprocessed data. The data unit is the international unit of the corresponding data type.
[0068] The performance sampling data is processed to obtain preprocessed data, abnormal data is removed, and the accuracy of the basic prediction value of the basic prediction model is improved. In addition, the performance sampling data is processed to obtain preprocessed data, and the preprocessed data is used as training data to build a basic model for performance prediction according to a preset machine learning algorithm.
[0069] The basic model is obtained based on performance sampling data and machine learning algorithms, maintaining certain global characteristics and avoiding local overfitting, providing global guarantees for subsequent prediction models.
[0070] Continue to refer to Figure 1 , after step S101, step S102 is performed, which is as follows:
[0071] Step S102: extracting features from the performance sampling data to obtain feature variables, and clustering the feature variables to obtain clustering results, wherein the clustering results represent different operating conditions divided based on the feature variables.
[0072] Regarding the method of obtaining characteristic variables, optionally, the Pearson coefficient method is used to extract features from the performance sampling data to obtain characteristic variables. Based on the Pearson coefficient method, the correlation coefficients between different categories of performance sampling data and between the performance sampling data and the performance prediction values are obtained. The correlation coefficient characterizes the linear correlation between different categories of performance sampling data and between the performance sampling data and the performance prediction values. The larger the linear correlation, the closer the relationship between the data and other data, thereby being able to characterize the data related to it.
[0073] When the absolute value of the correlation coefficient is greater than or equal to the preset threshold, the performance sampling data of the corresponding category is used as the characteristic variable. For example, the value range of the correlation coefficient is -1 to 1. When the correlation coefficient is a negative number, the data is negatively correlated; when the correlation coefficient is a positive number, the data is positively correlated. The closer the correlation coefficient is to -1 or 1, the higher the linear correlation. Therefore, the preset correlation coefficient absolute value threshold is set in the range of 0.3 to 1. Preferably, the preset threshold is set to 0.5.
[0074] Regarding the method of obtaining characteristic variables, optionally, performance sampling data that has a great impact on the performance prediction value is selected as the characteristic variable based on empirical knowledge and mechanism knowledge.
[0075] Regarding clustering processing, optionally, the clustering processing is selected from any one of a K-MEANS clustering algorithm, a mean shift clustering algorithm, and a DBSCAN clustering algorithm.
[0076] There are similar variables in the characteristic variables, and the distribution of similar variables is also similar, so similar variables are clustered together through clustering processing, and the number of divided working conditions is determined according to actual needs, so as to obtain clustering results. Among them, the clustering results represent different working conditions divided based on the characteristic variables.
[0077] Each type of operating condition includes multiple characteristic variables, and the characteristic variables include target values of the performance to be predicted. The target value is determined according to the performance to be predicted. For example, if the power of a gas turbine needs to be predicted, then each type of operating condition includes sampling the actual power of the gas turbine in the corresponding operating condition, and the actual power of the gas turbine is used as the target value.
[0078] The characteristic variables are obtained based on the correlation coefficients between performance sampling data of different categories and between performance sampling data and performance prediction values, ensuring the accuracy of working condition division, thereby avoiding overfitting, data sparsity, dimensionality disaster, etc.
[0079] Step S103: construct a correction model corresponding to each type of working condition based on the basic model and the clustering result.
[0080] Optionally, step S103 specifically includes: obtaining a basic prediction value output by a basic model according to the performance sampling data. Based on the performance sampling data, there are multiple target values collected in each type of working condition, and the basic prediction value is subtracted from the multiple target values to obtain the deviation between the basic prediction value and each target value, so as to perform data fitting based on a machine learning algorithm and construct a correction model corresponding to each type of working condition.
[0081] The preset machine learning algorithms include linear regression, polynomial regression, support vector regression, neural network, etc.
[0082] For example, the basic prediction value and the target value in each type of working condition are taken as input, and an optimal hyperplane is found through the support vector regression algorithm to regress the data so that most of the data points are located in a preset ε-insensitive band, and the deviation prediction value obtained by the algorithm and the deviation between the basic prediction value and the target value are calculated by loss function. When the value of the loss function is greater than the preset threshold, the position of the support vector and the corresponding multiplier are adjusted to regain the prediction value until the maximum number of iterations is reached or the value of the loss function is less than or equal to the preset threshold; when the value of the loss function is less than or equal to the preset threshold, the deviation prediction value is output.
[0083] For example, the basic prediction value and the target value in each type of working condition are used as the input of the neural network, and the loss function adopts the mean square error. The basic prediction value and the target value in each type of working condition are calculated by the loss function. When the loss value of the loss function is greater than the preset threshold, the model parameters in the neural network are adjusted, and the calculation is iterated until the maximum number of iterations is reached or the loss value is less than or equal to the preset threshold, and the deviation prediction value is output. Among them, the model parameters include the input variable coefficients and intercepts of the linear model or the connection weights and neuron biases between neurons of the multi-layer perceptron model.
[0084] The training is performed according to the machine learning algorithm to obtain the training results, and the training results are compared through a preset evaluation method to select the best deviation prediction value in the training results, and then the corresponding machine learning algorithm is selected to construct a correction model under the corresponding working conditions.
[0085] Optionally, the preset evaluation method is an indicator such as mean absolute error, mean square error, mean absolute percentage error, maximum mean absolute percentage error, and determination coefficient.
[0086] A correction model corresponding to each type of working condition is constructed so that corrections can be made for different working conditions, thereby improving the accuracy of the prediction for each type of working condition.
[0087] Step S104: obtaining a prediction model according to the basic model and the modified model, and obtaining a performance prediction value according to the performance sampling data to be predicted through the prediction model.
[0088] Figure 3 is a flowchart of constructing a prediction model according to an exemplary embodiment, such as Figure 3 As shown, the basic model and the modified model are combined to obtain a prediction model, wherein the basic model obtains a basic prediction value based on the preprocessed data, and the modified model obtains a deviation prediction value based on the working conditions of the performance sampling data. Thus, the prediction model adds the basic prediction value and the deviation prediction value based on the basic prediction value and the deviation prediction value to obtain a performance prediction value.
[0089] Determine the clustering result of the performance sampling data to be predicted, that is, confirm the working condition of the performance sampling data to be predicted, so as to determine the correction model corresponding to the performance sampling data to be predicted, and correct the performance sampling data to be predicted with the correction model.
[0090] As an example, clustering processing is based on performance sampling data divided into N types of working conditions, which is equivalent to dividing the data into N parts. The basic model predicts the N parts of data to obtain N basic prediction values, subtracts the N basic prediction values from the target values in the N parts of data, and obtains the deviation prediction value in the N parts of data through a machine learning algorithm. The deviation prediction value is added to the basic prediction value to obtain the performance prediction value.
[0091] The basic model and the modified model are combined to obtain the prediction model. On the one hand, the basic model is used as the basis, so that the model maintains the global characteristics and avoids local overfitting. On the other hand, the modified model is established under different working conditions, and the predicted values are modified according to different working conditions, thereby improving the prediction accuracy of different working conditions.
[0092] Figure 4 is a flow chart of a method for predicting gas turbine performance according to an exemplary embodiment, such as Figure 4 As shown, the performance sampling data to be predicted obtains the basic prediction value through the basic model, and determines the working condition of the predicted performance sampling data, obtains the deviation prediction value according to the correction model of the corresponding working condition, and finally obtains the performance prediction value by adding the basic prediction value and the deviation prediction value through the prediction model.
[0093] In summary, a gas turbine performance prediction method provided in an embodiment of the present application constructs a basic model for performance sampling data, and a basic prediction value can be obtained through the basic model; feature selection is performed on the performance sampling data, and at least two working conditions are divided based on a clustering algorithm, and a correction model is established in each type of working condition, and the basic prediction value is corrected by the correction model. The basic model and the correction model of each type of working condition are combined to construct a performance prediction model, so that the prediction values under different working conditions can be obtained through the performance prediction model, and accurate modeling for different working conditions is achieved, and the prediction accuracy under different working conditions is improved.
[0094] In a second aspect, an embodiment of the present application provides a gas turbine performance prediction device. Figure 5 FIG. 1 is a block diagram of a gas turbine performance prediction device according to an exemplary embodiment. Figure 5 As shown, a gas turbine performance prediction device comprises:
[0095] A first construction module is used to process the performance sampling data to obtain preprocessed data, use the preprocessed data as training data, and construct a basic model for performance prediction according to a preset machine learning algorithm;
[0096] A clustering module, used for performing feature extraction on the performance sampling data to obtain feature variables, and performing clustering processing on the feature variables to obtain clustering results, wherein the clustering results represent different operating conditions divided based on the feature variables;
[0097] A second construction module is used to construct a correction model corresponding to each type of working condition based on the basic model and the clustering result;
[0098] The obtaining module is used to obtain a prediction model according to the basic model and the modified model, and obtain a performance prediction value according to the performance sampling data to be predicted through the prediction model.
[0099] In summary, the present application provides a gas turbine performance prediction device, which, in the first construction module, constructs a basic model based on performance sampling data, divides the performance sampling data into operating conditions by a clustering algorithm in the clustering module, and constructs an operating condition correction model on each operating condition in the second construction model, thereby forming a combined model with the basic model and the operating condition correction model. In the acquisition module, in this combined model, the performance sampling data to be predicted can be corrected according to the operating conditions, thereby improving the accuracy of the performance prediction value. This solves the problem that the use of a unified gas turbine data model involves a balance of accuracy among many data points, making it difficult to accurately model different operating conditions, resulting in low prediction accuracy in some operating conditions.
[0100] It should be noted that the gas turbine performance prediction device provided in this embodiment is used to implement the above-mentioned implementation mode, and the descriptions that have been made will not be repeated. As used above, the terms "module", "unit", "subunit" and the like can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the above embodiment is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0101] In a third aspect, an embodiment of the present application provides an electronic device, Figure 6 FIG. 1 is a block diagram of an electronic device according to an exemplary embodiment. Figure 6 As shown, the electronic device may include a processor 81 and a memory 82 storing computer program instructions.
[0102] Specifically, the processor 81 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.
[0103] Among them, the memory 82 may include a large-capacity memory for data or instructions. By way of example and not limitation, the memory 82 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 82 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 82 may be inside or outside the data processing device. In a specific embodiment, the memory 82 is a non-volatile memory. In a specific embodiment, the memory 82 includes a read-only memory (ROM) and a random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, PROM for short), an erasable PROM (Erasable ProgrammableRead-Only Memory, EPROM for short), an electrically erasable PROM (Electrically Erasable ProgrammableRead-Only Memory, EEPROM for short), an electrically alterable ROM (Electrically Alterable Read-Only Memory, EAROM for short) or a flash memory (FLASH) or a combination of two or more of these. Under appropriate circumstances, the RAM can be a static random access memory (SRAM) or a dynamic random access memory (DRAM), wherein the DRAM can be a fast page mode dynamic random access memory (FPMDRAM), an extended data output dynamic random access memory (EDODRAM), a synchronous dynamic random access memory (SDRAM), etc.
[0104] The memory 82 may be used to store or cache various data files required for processing and / or communication, as well as possible computer program instructions executed by the processor 81 .
[0105] The processor 81 implements any one of the gas turbine performance prediction methods in the above embodiments by reading and executing computer program instructions stored in the memory 82 .
[0106] In one embodiment, a gas turbine performance prediction device may further include a communication interface 83 and a bus 80. Figure 6 As shown, the processor 81, the memory 82, and the communication interface 83 are connected via a bus 80 and communicate with each other.
[0107] The communication interface 83 is used to implement communication between the modules, devices, units and / or equipment in the embodiment of the present application. The communication port 83 can also implement data communication with other components such as: external devices, image / data acquisition equipment, databases, external storage, and image / data processing workstations.
[0108] The bus 80 includes hardware, software or both, and couples the components of a gas turbine performance prediction device to each other. The bus 80 includes but is not limited to at least one of the following: a data bus, an address bus, a control bus, an expansion bus, and a local bus. By way of example and not limitation, bus 80 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses or a combination of two or more of these. Where appropriate, bus 80 may include one or more buses. Although embodiments of the present application describe and illustrate a particular bus, the present application contemplates any suitable bus or interconnect.
[0109] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a program stored thereon, and when the program is executed by a processor, a gas turbine performance prediction method provided in the first aspect is implemented.
[0110] The readable storage medium may include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device or any suitable combination of the above.
[0111] In a possible implementation, the present invention can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the steps of a gas turbine performance prediction method provided in the first aspect.
[0112] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on a user device, partially on a user device, as an independent software package, partially on a user device and partially on a remote device, or entirely on a remote device.
[0113] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0114] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.
Claims
1. A gas turbine performance prediction method, characterized in that: include: Processing the performance sampling data to obtain preprocessed data, using the preprocessed data as training data, and building a basic model for performance prediction according to a preset machine learning algorithm; Performing feature extraction on the performance sampling data to obtain feature variables, and performing clustering processing on the feature variables to obtain clustering results, wherein the clustering results represent different working conditions divided based on the feature variables; Based on the basic model and the clustering results, construct a correction model corresponding to each type of working condition; A prediction model is obtained according to the basic model and the modified model, and a performance prediction value is obtained according to the performance sampling data to be predicted through the prediction model.
2. The gas turbine performance prediction method according to claim 1, characterized in that: The step of processing the performance sampling data to obtain preprocessed data includes: Selecting data to be analyzed from the performance sampling data using a preset length window, and obtaining a standard deviation and an average value of the data to be analyzed to determine whether the data to be analyzed is in a steady state; If the data to be analyzed is in a steady state, then the data to be analyzed is cleaned of abnormal data, wherein the abnormal data includes data deviating from a preset range and outlier data; The cleaned data is normalized to obtain the preprocessed data.
3. The gas turbine performance prediction method according to claim 1, characterized in that: The obtaining of characteristic variables comprises: Obtaining correlation coefficients between the performance sampling data of different categories and between the performance sampling data and performance prediction values; When the absolute value of the correlation coefficient is greater than or equal to a preset threshold, the performance sampling data corresponding to the category is used as the feature variable.
4. The gas turbine performance prediction method according to claim 1, characterized in that: The constructing a correction model corresponding to each type of working condition based on the basic model and the clustering result includes: Obtaining a basic prediction value output by the basic model according to the performance sampling data; Determining the deviation between the target value of each type of operating condition in the clustering result and the basic predicted value; Based on the deviation, a correction model corresponding to each type of working condition is constructed through a preset machine learning algorithm.
5. The gas turbine performance prediction method according to claim 1, characterized in that: The obtaining of the performance prediction value according to the performance sampling data to be predicted by the prediction model includes: Based on the performance sampling data to be predicted, a basic prediction value and a deviation prediction value in the prediction model are obtained; The basic prediction value and the deviation prediction value are added to obtain a performance prediction value.
6. The gas turbine performance prediction method according to claim 5, characterized in that: Before obtaining the basic prediction value and the deviation prediction value in the prediction model based on the performance sampling data to be predicted, the method includes: Determine the clustering result of the performance sampling data to be predicted; Based on the clustering result of the performance sampling data to be predicted, a correction model corresponding to the performance sampling data to be predicted is determined.
7. The gas turbine performance prediction method according to claim 1, wherein the preset machine learning algorithm comprises linear regression, polynomial regression, support vector regression, neural network, etc., characterized in that: The method of using the pre-processed data as training data and constructing a basic model for performance prediction according to a preset machine learning algorithm includes: Based on the training data and the preset machine learning algorithm, basic prediction values of different machine learning algorithms are obtained; The basic prediction value is evaluated according to a preset evaluation method, a machine learning algorithm is determined, and a corresponding basic model is constructed.
8. A gas turbine performance prediction device, characterized in that: include: A first construction module is used to process the performance sampling data to obtain preprocessed data, use the preprocessed data as training data, and construct a basic model for performance prediction according to a preset machine learning algorithm; A clustering module, used for performing feature extraction on the performance sampling data to obtain feature variables, and performing clustering processing on the feature variables to obtain clustering results, wherein the clustering results represent different operating conditions divided based on the feature variables; A second construction module is used to construct a correction model corresponding to each type of working condition based on the basic model and the clustering result; The acquisition module is used to obtain a prediction model based on the basic model and the modified model, and obtain a performance prediction value based on the performance sampling data to be predicted through the prediction model.
9. An electronic device, characterized in that: The invention comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a gas turbine performance prediction method according to any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a gas turbine performance prediction method according to any one of claims 1 to 7 is implemented.