Gas turbine similar working condition screening method based on characteristic parameter correlation analysis

Abstract

According to the characteristic parameter correlation analysis-based gas turbine similar working condition screening method provided by the invention, a gas turbine long-term historical operation database is constructed, key characteristic parameters are screened, and a rank space mapping technology is innovatively adopted to convert multi-dimensional operation parameters into rank coordinates with unified dimensions; in combination with self-adaptive weight distribution based on information entropy and a weighted Cabepla distance algorithm, accurate quantitative analysis of multi-dimensional dynamic coupling characteristics of the operation condition of the gas turbine is realized, and the problem of low matching accuracy caused by static condition division and multi-parameter independent processing in a traditional method is effectively solved. The accuracy and timeliness of similar working condition screening are remarkably improved, and high-precision and full-working-condition coverage data support is provided for performance monitoring, fault diagnosis and intelligent operation and maintenance of the gas turbine.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent operation and maintenance of gas turbines, and particularly relates to a gas turbine similar working condition screening method based on characteristic parameter correlation analysis. BACKGROUND

[0002] Due to the influence of factors such as ambient temperature, load fluctuation and fuel characteristics, the gas turbine often deviates from the design working condition during actual operation, resulting in dynamic changes of operating parameters. This dynamic variable working condition characteristic makes the unit performance monitoring, fault feature extraction and fault diagnosis seriously dependent on the accurate matching of historical similar working conditions.

[0003] The prior art mostly divides the working conditions by presetting the load interval or the ambient parameter range, ignoring the continuous gradual change characteristics of the parameters, resulting in significant differences in working conditions within the same interval and failing to accurately reflect the actual operating state. In addition, the prior art mostly processes each parameter independently without fully considering the dynamic correlation between parameters, resulting in the inability to accurately screen similar working conditions of the gas turbine. SUMMARY

[0004] The first aspect of the present disclosure provides a gas turbine similar working condition screening method based on characteristic parameter correlation analysis, comprising the following steps:

[0005] Obtain the historical operating data of the gas turbine, including atmospheric environment data, operating state data and monitoring data, construct a historical operating database, and screen a first characteristic parameter representing the operating working condition from the historical operating database;

[0006] Obtain the current operating data of the gas turbine, and screen operating data of the same type as the first characteristic parameter in the current operating data as a second characteristic parameter;

[0007] Construct a historical working condition characteristic data set and a historical operating characteristic parameter vector based on the first characteristic parameter, generate a rank mapping function according to the historical working condition characteristic data set, and calculate the rank coordinates of the historical operating characteristic parameter vector through the rank mapping function;

[0008] Construct a real-time operating characteristic parameter vector based on the second characteristic parameter, and calculate the rank coordinates of the real-time operating characteristic parameter vector through the rank mapping function;

[0009] Calculate the weighted Cambera distance between the real-time operating characteristic parameter vector and the historical operating characteristic parameter vector according to the rank coordinates of the real-time operating characteristic parameter vector and the rank coordinates of the historical operating characteristic parameter vector;

[0010] Screen the historical operating working condition similar to the current working condition according to the weighted Cambera distance.

[0011] In conjunction with the first aspect, the first characteristic parameter includes at least five of the following: gas turbine power, speed, IGV opening, atmospheric pressure, atmospheric humidity, compressor inlet pressure loss, compressor inlet air temperature, and gas turbine exhaust temperature.

[0012] In conjunction with the first aspect, the construction of the historical operating condition feature dataset and the historical operation feature parameter vector based on the first feature parameter is achieved through the following formula:

[0013]

[0014] Where n is the number of samples and k is the number of feature parameters. Let D be the set of historical operating feature parameters, t be the historical operating condition feature dataset, and v be the sampling time. tk This represents the k-th historical running feature parameter vector at time t.

[0015] In conjunction with the first aspect, generating the rank mapping function based on the historical working condition feature dataset includes independently sorting each variable to generate the rank mapping function:

[0016]

[0017] Where v j Let rank be the j-th feature parameter in the historical data. j (v j ) for v j Sort position in historical data

[0018]

[0019]

[0020] rank mapping function f j (v j The output value range of ) is [0,1], representing v j The quantile position in the historical distribution.

[0021] In conjunction with the first aspect, the rank coordinates of the historical operational feature parameter vector are calculated using the rank mapping function according to the following formula:

[0022]

[0023] in, The rank coordinates of the historical running characteristic parameter vector.

[0024] In conjunction with the first aspect, the step of constructing a real-time running feature parameter vector based on the second feature parameter, and calculating the rank coordinates of the real-time running feature parameter vector through the rank mapping function, includes:

[0025]

[0026] in, To obtain the rank coordinates of the feature parameter vector in real-time operation. This is the vector of the k-th real-time running feature parameters.

[0027] In conjunction with the first aspect, the formula for calculating the weighted Canberra distance is as follows:

[0028]

[0029] Where d tw For the weighted Canberra distance, w j The weight coefficient for the j-th feature parameter; Let R be the rank coordinate of the j-th real-time running feature parameter vector. tj Let ε be the rank coordinate of the j-th historical running characteristic parameter vector at time t, and let ε be a small constant to prevent division by zero.

[0030] In conjunction with the first aspect, the step of filtering historical operating conditions similar to the current operating condition based on the weighted Canberra distance includes:

[0031] Set a similarity threshold δ to filter historical operating conditions that are similar to the current operating condition. The filtering condition is d. tw <δ.

[0032] A second aspect of this disclosure provides an electronic device, characterized in that it comprises:

[0033] One or more processors;

[0034] A storage unit is used to store one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the gas turbine similar operating condition screening method based on feature parameter correlation analysis.

[0035] A third aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it can implement the gas turbine similar operating condition screening method based on feature parameter correlation analysis.

[0036] Beneficial Effects: This disclosure provides a method for screening similar operating conditions of gas turbines based on the correlation analysis of characteristic parameters. By constructing a long-term historical operating database of gas turbines and screening key characteristic parameters, it innovatively adopts rank space mapping technology to convert multi-dimensional operating parameters into rank coordinates with unified dimensions. Combined with adaptive weight allocation based on information entropy and the weighted Canberra distance algorithm, it achieves accurate quantitative analysis of the multi-dimensional dynamic coupling characteristics of gas turbine operating conditions. This effectively solves the problem of low matching accuracy caused by static operating condition division and independent processing of multiple parameters in traditional methods, significantly improving the accuracy and timeliness of similar operating condition screening. It provides high-precision, full-condition data support for gas turbine performance monitoring, fault diagnosis, and intelligent operation and maintenance. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a method for screening similar operating conditions of gas turbines based on feature parameter correlation analysis according to an embodiment of this disclosure.

[0038] Figure 2 An electronic device according to an embodiment of this disclosure. Detailed Implementation

[0039] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those disclosed herein.

[0040] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the present disclosure. The singular forms “a,” “the,” and “the” as used in this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0041] like Figure 1 The diagram shown is a flowchart illustrating a method for screening similar operating conditions of gas turbines based on correlation analysis of characteristic parameters, according to an embodiment of this disclosure. The method includes:

[0042] S101: Acquire historical operating data of the gas turbine, including atmospheric environment data, operating status data and monitoring data, construct a historical operating database, and select a first characteristic parameter characterizing the operating condition from the historical operating database;

[0043] This step first requires collecting operating records of the gas turbine over a considerable period of time to form a historical operating database. These records cover a variety of information that affects and reflects the operating status of the gas turbine, specifically including: atmospheric environmental data reflecting external conditions, such as ambient temperature, atmospheric pressure, and air humidity; core operating status data reflecting the overall output of the unit, such as the gas turbine's output power and rotor speed; and detailed operating monitoring data of key internal components, such as temperatures at various stages (e.g., compressor inlet temperature and exhaust temperature), pressures (e.g., compressor pressure loss), vibration amplitude, fuel or air flow, and the opening positions of various valves (e.g., inlet guide vanes IGV).

[0044] These data typically originate from the historical data storage unit of the gas turbine control system or the plant-level monitoring information system, ensuring their integrity and continuity. The time span is generally required to cover at least one complete annual cycle to include the operating status under different seasons and environmental conditions.

[0045] After establishing a historical operational database containing the aforementioned multi-dimensional data, the next step is to identify and filter out the core characteristic parameters that best represent the operating conditions of the gas turbine from this massive amount of data, namely the first characteristic parameter. The filtering process is not arbitrary; rather, it involves a scientific evaluation based on the parameter's sensitivity and representativeness to changes in the gas turbine's operating state.

[0046] The core idea is to prioritize key physical quantities that are strongly correlated with the overall performance, efficiency, and safety of the gas turbine and exhibit stable variation patterns under different operating conditions. Typically, through engineering practice and analysis, these key characteristic parameters include power and speed, which directly reflect the unit's load status; atmospheric pressure, atmospheric humidity, and compressor inlet pressure loss, which affect intake conditions; compressor inlet air temperature and gas turbine exhaust temperature, which reflect the thermodynamic cycle state; and IGV opening, which characterizes the core control actions.

[0047] By focusing on these selected, highly representative primary characteristic parameters, subsequent analysis can more effectively capture the core characteristics of the gas turbine's operating status, avoid interference from redundant information, and lay the foundation for accurate operating condition matching. At the same time, the database itself is not static; its content is periodically updated as the unit continues to operate, incorporating the latest operating data and removing outdated records to ensure the database's timeliness and adaptability to the current operating status.

[0048] S102: Obtain the current operating data of the gas turbine, and filter the operating data of the same type as the first feature parameter in the current operating data as the second feature parameter;

[0049] This step focuses on the real-time capture and feature extraction of the gas turbine's current operating status. In practice, a sensor network deployed in the gas turbine itself and auxiliary systems is used to collect dynamic data streams during unit operation in real time.

[0050] These data must strictly correspond to the data types in the historical database, covering atmospheric environmental parameters (such as current ambient temperature, atmospheric pressure, and humidity), operating status parameters (such as real-time power output and rotor speed), and key component monitoring parameters (such as temperature, pressure, vibration, flow rate, and valve opening at various levels).

[0051] To ensure comparability between subsequent and historical operating conditions, the data acquisition source, sampling frequency, and measurement accuracy of real-time data must be completely consistent with the construction standards of the historical database to avoid matching deviations due to differences in data quality.

[0052] After obtaining complete real-time running data, it is necessary to accurately extract physical quantities that are exactly the same as the first characteristic parameters determined in S101 to form the second characteristic parameter set.

[0053] This process is essentially a dynamic mapping: if the first characteristic parameter includes specific items such as power, speed, IGV opening, atmospheric pressure, and exhaust temperature, then the second characteristic parameter must strictly extract the corresponding data points from the real-time data stream according to the same name, the same unit, and the same physical meaning.

[0054] For example, if the first characteristic parameter is selected as "compressor inlet air temperature", then the temperature value of the same measuring point and the same dimension must be extracted synchronously from the real-time data as the second characteristic parameter.

[0055] This strict type alignment ensures that historical and real-time data reside in the same feature space, providing a foundation for subsequent quantitative comparisons. Simultaneously, this step implicitly includes a data validity verification mechanism—when a value corresponding to a first feature parameter is missing in the real-time data, an exception handling process (such as interpolation compensation or deferred analysis) will be triggered to ensure the completeness and reliability of the second feature parameter set input into subsequent algorithms.

[0056] S103: Construct a historical operating condition feature dataset and a historical operating feature parameter vector based on the first feature parameter; generate a rank mapping function based on the historical operating condition feature dataset; and calculate the rank coordinates of the historical operating feature parameter vector through the rank mapping function.

[0057] The core objective of this step is to transform historical operational data into quantifiable and comparable rank-space coordinates. First, based on the first feature parameter selected by S101, all relevant data points are extracted from the historical database to form a structured historical operating condition feature dataset.

[0058] The construction of the historical operating condition feature dataset and the historical operation feature parameter vector based on the first feature parameter is achieved through the following formula:

[0059]

[0060] Where n is the number of samples and k is the number of feature parameters. Let D be the set of historical operating feature parameters, t be the historical operating condition feature dataset, and v be the sampling time. tk This represents the k-th historical running feature parameter vector at time t.

[0061] This dataset is essentially a multidimensional feature matrix, with each row corresponding to a complete snapshot of a historical moment, and each column representing a numerical sequence of a specific feature parameter (such as power, exhaust temperature, etc.) over time.

[0062] After obtaining the historical working condition feature dataset, it is necessary to construct a rank mapping rule for unifying the units of measurement.

[0063] The step of generating a rank mapping function based on the historical working condition feature dataset includes independently sorting each variable to generate the rank mapping function:

[0064]

[0065] Where v j Let rank be the j-th feature parameter in the historical data. j (v j ) for v j Sort position in historical data

[0066]

[0067] rank mapping function f j (v j The output value range of ) is [0,1], representing v j The quantile position in the historical distribution.

[0068] Specifically, for each selected feature parameter, its numerical distribution in the historical dataset is analyzed independently. For example, for the parameter "gas turbine power," the power values ​​recorded at all historical moments are sorted from smallest to largest to determine the relative position of each power value in the historical sequence. This sorting process assigns a rank number to each historical data point—the smallest value has a rank of 1, the second smallest rank is 2, and so on; if there are identical values, the average rank is taken. Subsequently, through normalization, the rank number is converted into a quantile position value between 0 and 1, forming the rank mapping function for that feature parameter. The significance of this function is that regardless of the original unit and magnitude of the parameter (e.g., temperature is in degrees Celsius while pressure is in megapascals), it is ultimately converted into a standardized rank coordinate representing "where this value is located in the historical distribution."

[0069] After constructing the rank mapping function, it is applied to the historical dataset itself. Each parameter value in the historical feature parameter vector (i.e., the combination of all feature parameter values ​​at a certain moment) is transformed using the corresponding rank mapping function.

[0070] The rank coordinates of the historical running feature parameter vector are calculated using the rank mapping function according to the following formula:

[0071]

[0072] in, The rank coordinates of the historical running characteristic parameter vector.

[0073] For example, the power value at a certain historical moment is converted to 0.35 using a power mapping function, and the exhaust temperature is converted to 0.72 using a temperature mapping function; other parameters are converted similarly. This ultimately generates a rank coordinate vector for that historical moment, where each dimension represents the relative position of the corresponding parameter over a long historical period. This transformation makes parameters with originally different dimensions comparable in the rank space, laying the foundation for subsequent precise quantification of similarity.

[0074] S104: Construct a real-time running feature parameter vector based on the second feature parameter, and calculate the rank coordinates of the real-time running feature parameter vector through the rank mapping function;

[0075] The core task of this step is to map the real-time gas turbine operating status to the rank space constructed from historical data.

[0076] First, based on the second feature parameter determined in S102 (i.e., the real-time physical quantity that strictly corresponds to the historical first feature parameter), the parameter values ​​collected at the current moment are organized into a real-time running feature parameter vector in a fixed order. The dimension of this vector is completely consistent with the historical feature vector. For example, when the first feature parameter includes k parameters such as power, speed, and exhaust temperature, the real-time vector is also arranged in the same order of power value, speed value, exhaust temperature value, etc., forming a comparable structured data object.

[0077] Subsequently, the real-time vector is transformed using the rank mapping function generated in S103.

[0078] The rank coordinates of the real-time running feature parameter vector include:

[0079]

[0080] in, To obtain the rank coordinates of the feature parameter vector in real-time operation. This is the vector of the k-th real-time running feature parameters.

[0081] This process must strictly adhere to the mapping rules established using historical data: for each feature parameter value in the real-time vector, its corresponding historical rank mapping function is called for independent calculation. For example, the current measured power value is input into the rank mapping function of the power parameter, and the output is the relative quantile of that power value in the historical power distribution; simultaneously, the current exhaust temperature value is input into the rank mapping function of the temperature parameter, and the output is its quantile in the historical temperature sequence. This transformation essentially converts the absolute physical quantity values ​​of real-time parameters into "positional labels" relative to long-term historical operation.

[0082] The resulting real-time rank coordinate vector has values ​​between 0 and 1 for each dimension, representing the relative position of the current operating state with respect to historical records across each feature dimension. For example, a rank coordinate of 0.8 indicates that the current value is higher than 80% of historical samples; a value of 0.1 indicates that it is only higher than 10% of historical samples. This transformation not only eliminates the dimensional differences between different parameters, but more importantly, it places the real-time operating state under a unified reference system of historical experience, creating conditions for subsequent accurate quantification of its similarity to historical operating conditions. It is particularly important to emphasize that this process completely reuses the mapping rules constructed from historical datasets, ensuring that historical and real-time data are compared under the same standards.

[0083] S105: Calculate the weighted Campbell's distance between the real-time running feature parameter vector and the historical running feature parameter vector based on the rank coordinates of the real-time running feature parameter vector and the rank coordinates of the historical running feature parameter vector.

[0084] The core of this step lies in quantifying the similarity between the current operating state and historical operating conditions. First, it's important to understand that although S104 has converted both real-time and historical data into a unified rank coordinate system, different characteristic parameters have varying degrees of influence on the gas turbine's operating state. For example, power variations typically reflect fundamental differences in operating conditions more effectively than atmospheric humidity variations. Therefore, directly comparing the rank differences of all parameters without considering their importance can lead to misjudgment.

[0085] To address this issue, this step introduces an adaptive weighting mechanism based on information entropy. This mechanism stems from in-depth analysis of historical data: by calculating the information entropy value exhibited by each feature parameter over a long historical period, the "importance" of that parameter is objectively assessed.

[0086] Specifically, if a parameter fluctuates frequently and significantly in history (such as power changing dramatically with load), its information entropy is high, indicating that the parameter carries more information about state changes and should be given a higher weight in similarity assessment. Conversely, if a parameter remains stable over a long period (such as certain environmental parameters), its weight will be reduced accordingly. This weighting method is entirely determined by the characteristics of the historical data itself and requires no manual intervention.

[0087] After obtaining the weight coefficients of each feature parameter, the improved Canberra distance algorithm is used to calculate the similarity.

[0088] The formula for calculating the weighted Canberra distance is:

[0089]

[0090] Where d tw For the weighted Canberra distance, w j The weight coefficient for the j-th feature parameter; Let R be the rank coordinate of the j-th real-time running feature parameter vector. tj Let ε be the rank coordinate of the j-th historical running characteristic parameter vector at time t, and let ε be a small constant to prevent division by zero.

[0091] The core advantage of this algorithm lies in its high sensitivity to small differences, making it particularly suitable for quantile comparisons in rank space. In practice, for the rank coordinate vectors of real-time and historical conditions, the relative difference in rank values ​​is calculated parameter by parameter: the difference in rank coordinates for each parameter is divided by the sum of its absolute rank coordinate values ​​to obtain a standardized difference measure for that parameter. This calculation method ensures that both large-scale and small-scale rank differences can be reasonably captured.

[0092] Finally, the standardized difference measures of all parameters are multiplied by their corresponding weighting coefficients and summed to obtain a comprehensive weighted Canberra distance value. The smaller this distance value, the closer the real-time operating condition is to the historical operating condition in terms of the relative distribution of core characteristic parameters; the larger the distance value, the more significant the difference. This calculation method considers both the differences in importance of different parameters and the coupling relationship between parameters, thereby achieving a precise quantitative assessment of the similarity of gas turbine operating states.

[0093] S106: Filter historical operating conditions that are similar to the current operating conditions based on the weighted Canberra distance.

[0094] This step is the final decision-making stage for operating condition matching. Its core objective is to accurately locate historical operating conditions that are highly similar to the current operating state from massive amounts of historical data. The key to achieving this goal lies in establishing scientific similarity judgment rules:

[0095] A dynamic similarity threshold δ is set as the filtering criterion to filter historical operating conditions that are similar to the current operating condition. The filtering condition is d. tw <δ.

[0096] This threshold is not a fixed value, but is determined comprehensively based on the characteristics of historical data and engineering requirements. By analyzing the distribution patterns of distance values ​​between historical operating conditions, and combining this with the tolerance requirements for operating condition similarity in actual gas turbine operation (for example, performance analysis requires high similarity, while fault warning can be appropriately relaxed), a reasonable distance boundary is finally defined. The threshold essentially defines a quantitative standard for "similarity": if the weighted Canberra distance between a real-time operating condition and a certain historical operating condition is lower than this threshold, then the two are considered to be in a similar state that can be mutually referenced.

[0097] During the specific screening process, the system calculates the distance between the rank coordinates of the current real-time operating condition and the rank coordinates of each historical operating condition in the historical database (this distance value has already been completed in S105). Then, automated comparison is initiated: the distance value corresponding to each historical operating condition is compared with a preset threshold in real time, automatically marking all historical records that meet the condition of a distance below the threshold. This process is like casting a "similarity filter" across the river of history, accurately capturing historical fragments that meet the criteria.

[0098] The final output set of similar operating conditions not only includes historical time points that meet the similarity criteria, but also synchronously associates complete snapshots of operating parameters at that moment (such as raw values ​​of power, temperature, valve opening, etc.). The output results are typically sorted from smallest to largest distance value—the earlier the operating condition is ranked, the higher its similarity to the current state, allowing maintenance personnel to prioritize the most relevant historical records. For example, when a gas turbine experiences abnormal vibration, the system can immediately retrieve vibration data and corresponding handling measures for dozens of similar historical operating conditions, providing a direct reference for fault analysis.

[0099] Through dynamic quantitative similarity assessment, it can find highly consistent operating conditions during steady-state operation and capture similar transition processes under changing operating conditions, truly achieving "full-state, refined" operating condition matching and building a high-value historical experience database for intelligent operation and maintenance of gas turbines.

[0100] Electronic device 200 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 200 may include, but is not limited to, processor 201 and memory 202. Those skilled in the art will understand that... Figure 2 This is merely an example of electronic device 200 and does not constitute a limitation on electronic device 200. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0101] The processor 201 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0102] The memory 202 can be an internal storage unit of the electronic device 200, such as a hard disk or RAM of the electronic device 300. The memory 202 can also be an external storage device of the electronic device 200, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the electronic device 200. Furthermore, the memory 202 can include both internal and external storage units of the electronic device 200. The memory 202 is used to store the computer program 203 and other programs and data required by the electronic device. The memory 202 can also be used to temporarily store data that has been output or will be output.

[0103] The above embodiments are only used to illustrate the technical solutions of this disclosure, and are not intended to limit it. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be included within the protection scope of this disclosure.

Claims

1. A method for screening similar operating conditions of gas turbines based on correlation analysis of characteristic parameters, characterized in that, Includes the following steps: Historical operating data of the gas turbine is acquired, including atmospheric environment data, operating status data and monitoring data, and a historical operating database is constructed. First characteristic parameters characterizing the operating conditions are selected from the historical operating database. Obtain the current operating data of the gas turbine, and filter the operating data in the current operating data that are of the same type as the first feature parameter, and use them as the second feature parameter; Based on the first feature parameter, a historical operating condition feature dataset and a historical operation feature parameter vector are constructed. A rank mapping function is generated according to the historical operating condition feature dataset, and the rank coordinates of the historical operation feature parameter vector are calculated through the rank mapping function. A real-time running feature parameter vector is constructed based on the second feature parameter, and the rank coordinates of the real-time running feature parameter vector are calculated through the rank mapping function; The weighted Campbell distance between the real-time running feature parameter vector and the historical running feature parameter vector is calculated based on the rank coordinates of the real-time running feature parameter vector and the rank coordinates of the historical running feature parameter vector. Based on the weighted Canberra distance, historical operating conditions similar to the current operating conditions are selected.

2. The method according to claim 1, characterized in that, The first characteristic parameter includes at least five of the following: gas turbine power, speed, IGV opening, atmospheric pressure, atmospheric humidity, compressor inlet pressure loss, compressor inlet air temperature, and gas turbine exhaust temperature.

3. The method according to claim 1, characterized in that, The construction of the historical operating condition feature dataset and the historical operation feature parameter vector based on the first feature parameter is achieved through the following formula: Where n is the number of samples, and k is the number of feature parameters. Let D be the set of historical operating feature parameters, t be the historical operating condition feature dataset, and v be the sampling time. tk This represents the vector of the k-th historical running feature parameters at time t.

4. The method according to claim 3, characterized in that, The step of generating a rank mapping function based on the historical working condition feature dataset includes independently sorting each variable to generate the rank mapping function: Where v j Let rank be the j-th feature parameter in the historical data. j (v j ) for v j Sort position in historical data rank mapping function f j (v j The output value range of ) is [0,1], representing v j The quantile position in the historical distribution.

5. The method according to claim 4, characterized in that, The rank coordinates of the historical running feature parameter vector are calculated using the rank mapping function according to the following formula: in, The rank coordinates of the historical running characteristic parameter vector.

6. The method according to claim 5, characterized in that, The step of constructing a real-time running feature parameter vector based on the second feature parameter, and calculating the rank coordinates of the real-time running feature parameter vector through the rank mapping function, includes: in, To obtain the rank coordinates of the feature parameter vector in real-time operation. This is the vector of the k-th real-time running feature parameters.

7. The method according to claim 6, characterized in that, The formula for calculating the weighted Canberra distance is: Where d tw For the weighted Canberra distance, w j The weight coefficient for the j-th feature parameter; Let R be the rank coordinate of the j-th real-time running feature parameter vector. tj Let ε be the rank coordinate of the j-th historical running characteristic parameter vector at time t, and let ε be a small constant to prevent division by zero.

8. The method according to claim 1, characterized in that, The step of filtering historical operating conditions similar to the current operating condition based on the weighted Canberra distance includes: Set a similarity threshold δ to filter historical operating conditions that are similar to the current operating condition. The filtering condition is d. tw <δ.

9. An electronic device, characterized in that, include: One or more processors; A storage unit is used to store one or more programs, which, when executed by one or more processors, enable the one or more processors to implement the gas turbine similar operating condition screening method based on feature parameter correlation analysis as described in claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it can implement the gas turbine similar operating condition screening method based on feature parameter correlation analysis as described in claims 1-8.

Images