A method for determining parameters and component performance in an entire process of engine starting

Abstract

The application belongs to the technical field of engine design, and provides a method for determining full-flow process parameters and component performance of engine starting process, comprising: updating a historical parameter feature set according to a surge margin corresponding to different historical convergence value groups; performing weighted proportion weight distribution of parameters through deviation calculation of the surge margin to determine initial guess values of the parameters; constructing an initial guess range of each parameter based on volatility of the historical convergence values of the parameters, and generating an initial guess value group; performing similarity analysis on the initial guess value group through the initial guess value group and the historical convergence value group whose surge margin does not meet the requirements to screen the initial guess value group; and performing iterative calculation by combining a full-flow test model of the engine and a Newton iteration method, selecting a target initial guess value group according to an iterative calculation result and an iterative representation, and determining starting parameters of the engine; and the application prevents non-physical determination of engine parameters, thereby improving accuracy of full-flow parameter determination of the engine.

Description

Technical Field

[0001] The present invention belongs to the technical field of engine design, and in particular relates to a method for determining full-process parameters and component performance during an engine starting process. Background Art

[0002] Accurately determining engine parameters throughout their entire lifecycle is crucial for ensuring safe and stable operation. Currently, engine parameter determination methods based on solving nonlinear equations face numerous challenges, primarily in initial guess setting, solution space screening, and iterative calculations. These challenges frequently lead to nonphysical solution issues, severely impacting the accuracy of parameter determination.

[0003] When it comes to setting initial guess values, traditional methods often use fixed-proportional fluctuations (such as ±10%) or empirical values, without fully considering the actual distribution characteristics and physical constraints of parameters under different operating conditions. This approach cannot adapt to the volatility of historical data, which can easily cause the initial guess value to deviate from the true solution, causing the iterative calculation to fall into a local optimum or non-physical solution area. For example, under different ambient temperatures, altitudes, and equipment health conditions, using a unified initial guess value setting rule may cause the initial guess values ​​of key parameters such as compressor speed and turbine inlet temperature to be out of line with actual operating requirements, thereby causing subsequent calculation errors.

[0004] Existing solutions often lack systematic control over surge risk and fail to establish a deep correlation between historical data and current operating conditions. This makes it difficult to identify solutions similar to historically dangerous operating conditions and effectively exclude non-physical solutions close to the surge boundary, increasing the risk of engine surge during startup. Furthermore, the lack of consideration of dynamic parameter trends during the iteration process makes it impossible to guarantee the stability and rationality of the solution.

[0005] During the iterative calculation phase, traditional methods often use a single initial guess combined with local optimization algorithms such as the Newton method. Due to the limitations of the initial guess, these methods can easily fall into non-physical solutions. Furthermore, the iterative process fails to deeply integrate physical constraints with the computational process. Consequently, even if the calculations converge, the resulting solution may not conform to actual physical laws.

[0006] To this end, the present invention provides a method for determining the full-process parameters and component performance of an engine starting process. Summary of the Invention

[0007] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0008] The technical solution adopted by the present invention to solve the technical problem is: a method for determining the parameters and component performance of the entire engine starting process, including:

[0009] Through the similarity analysis of working conditions, the historical convergence values ​​of various parameters are retrieved from the engine history start database to construct the historical parameter feature set;

[0010] The historical parameter feature set is updated according to the surge margins corresponding to different historical convergence value groups in the historical parameter feature set;

[0011] Deviation calculation is performed based on the surge margin corresponding to the historical convergence value group of the parameters in the updated historical parameter feature set. Weighted proportional weights of the parameters are allocated based on the surge margin deviation calculation results. The initial guess values ​​of the parameters are calculated based on the historical convergence values ​​of the parameters.

[0012] Based on the volatility of the historical convergence values ​​of the parameters and taking the initial guess values ​​of the parameters as a benchmark, the initial guess range of each parameter is constructed, and multiple initial guess value groups are generated using a random number generator;

[0013] The initial guess value group is screened by performing similarity analysis on the initial guess value group and the historical convergence value group whose surge margin does not meet the requirements contained in the historical parameter feature set;

[0014] The screened initial guess value groups are used as input, and iterative calculations are performed in combination with the full-process test model of the start-up and the Newton iteration method. The target initial guess value group is selected based on the iterative calculation results and the iterative stability performance, and the solution after the iterative calculation of the target initial guess value group is used as the starting parameters of the engine.

[0015] Preferably, the process of the operating condition similarity analysis is:

[0016] Obtain the operating condition data corresponding to different historical convergence value groups in the historical start database and the current operating condition data, and after normalization, use Euclidean distance calculation to obtain the operating condition similarity value between the current operating condition data and the operating condition data corresponding to the historical convergence value group;

[0017] If the operating condition similarity value meets the requirements, the historical convergence value group is retrieved and summarized to obtain the historical parameter feature set.

[0018] Preferably, the process of updating the historical parameter feature set is:

[0019] The surge margins corresponding to different historical convergence value groups are obtained. If the surge margins meet the requirements, the historical convergence value groups are retained in the historical parameter feature set; otherwise, they are deleted.

[0020] Preferably, the surge margin deviation is calculated as follows:

[0021] Based on any parameter, the surge margins corresponding to different historical convergence value groups of the historical convergence value of the parameter are obtained, and the difference between the difference and the preset surge margin is calculated to obtain the surge margin deviation.

[0022] Preferably, the process of calculating the initial guess value of the parameter is:

[0023] The surge margin deviations corresponding to the various parameters are summed to obtain a total surge margin deviation. Based on any parameter, the ratio of the surge margin deviation corresponding to the parameter to the total surge margin deviation is calculated to obtain a weighted proportional weight of the parameter.

[0024] The historical convergence values ​​of the parameters in different historical convergence value groups are multiplied by the weighted proportion weights of the parameters to obtain the initial guess values ​​of the parameters.

[0025] Preferably, the process of constructing the initial guess range of each parameter includes:

[0026] Based on any parameter, obtain the historical convergence value of the parameter in different historical convergence value groups, and integrate them to obtain the historical convergence sequence of the parameter;

[0027] Calculate the standard deviation of the historical convergence series of parameters And based on the initial guess value of the parameter, the initial guess range of the parameter is constructed, specifically: [initial guess value - k* , initial guess value + k* ], where k is the dynamic scaling factor.

[0028] Preferably, the process of screening the initial guess value group is:

[0029] Extract the historical convergence value group whose surge margin does not meet the requirements from the historical parameter feature set and mark it as a non-satisfied convergence value group;

[0030] Based on any initial guess value group, calculate the Euclidean distance with each non-convergence value group respectively;

[0031] If, among all non-convergence value groups, the Euclidean distance between any non-convergence value group and the initial guess value group does not meet the requirement, the initial guess value group will be deleted; otherwise, the initial guess value group will be retained.

[0032] Preferably, the process of selecting the target initial guess value group according to the iterative calculation results and the iterative stability performance is:

[0033] Obtain the full process parameters corresponding to each initial guess value group after the iterative calculation is completed, and calculate the surge margin;

[0034] If the surge margin is greater than or equal to the preset surge margin, the initial guess value group is marked as an initial compliance initial guess value group;

[0035] Marking an iteration in which the surge margin of the initial guess value group is greater than or equal to the preset surge margin during the iteration process as a surge qualified iteration;

[0036] By analyzing the number of surge qualified iterations and the corresponding iterative surge deviation, the iterative qualified ratio, iterative surge qualified degree value and iterative surge stability value are obtained, and the iterative performance value is obtained by comprehensive output;

[0037] The initial guess value group with the largest iterative performance value is selected as the target initial guess value group, and the solution after the iterative calculation of the target initial guess value group is completed is used as the full process parameter of the engine.

[0038] Preferably, the qualified iteration ratio is the ratio of the number of occurrences of surge qualified iterations.

[0039] Preferably, the iterative surge qualification value and the iterative surge stability value are obtained in the following manner:

[0040] Calculate the deviation between the surge margin of each surge qualified iteration and the preset surge margin to obtain an iterative surge deviation, average all iterative surge deviations to obtain an iterative surge qualified degree value;

[0041] The coefficient of variation of all iterative surge deviations is calculated to obtain the iterative surge stability value.

[0042] The beneficial effects of the present invention are as follows: by analyzing the similarity of working conditions, the historical convergence values ​​of various parameters are retrieved from the engine history start database to construct a historical parameter feature set; the historical parameter feature set is updated according to the surge margins corresponding to different historical convergence value groups in the historical parameter feature set; the deviation is calculated by calculating the surge margins corresponding to the historical convergence value groups of the parameters in the updated historical parameter feature set, and the weighted proportion weights of the parameters are allocated according to the surge margin deviation calculation results, and the initial guess values ​​of the parameters are calculated in combination with the historical convergence values ​​of the parameters; based on the volatility of the historical convergence values ​​of the parameters and taking the initial guess values ​​of the parameters as a benchmark, the initial guess values ​​of the parameters are constructed. An initial guess range is obtained, and a random number generator is used to generate multiple initial guess value groups; the initial guess value groups are screened by performing similarity analysis on the initial guess value groups and the historical convergence value groups whose surge margins do not meet the requirements contained in the historical parameter feature set; the screened initial guess value groups are respectively used as inputs, and iterative calculations are performed in combination with the full-process engine test model and the Newton iteration method; a target initial guess value group is selected according to the iterative calculation results and the iterative stability performance, and the solution of the target initial guess value group after iterative calculation is used as the starting parameter of the engine. The present invention prevents the occurrence of non-physical solutions when determining the engine parameters, thereby improving the accuracy of determining the full-process engine parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described below with reference to the accompanying drawings.

[0044] Figure 1 This is a flowchart of the steps of a method for determining parameters and component performance during the entire engine starting process according to an embodiment of the present invention;

[0045] Figure 2 This is a logical judgment diagram for screening the initial guess value group in a method for determining parameters and component performance of the entire engine starting process described in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0047] See also Figure 1-Figure 2 As shown, a method for determining parameters and component performance of the entire engine starting process according to an embodiment of the present invention includes the following steps:

[0048] Step 1: Through operating condition similarity analysis, the historical convergence values ​​of various parameters are retrieved from the engine historical start database to construct a historical parameter feature set;

[0049] In step 1, the engine start history database includes multiple historical convergence value groups after engine starts under different operating conditions. Each historical convergence value group includes the historical convergence value of each parameter. The historical convergence value of a parameter refers to the parameter value that is ultimately stable and conforms to physical laws and operating requirements through iterative calculation during the historical engine start process.

[0050] It should be noted that the historical convergence values ​​of the parameters include but are not limited to the engine speed, turbine inlet temperature, fuel flow rate, pressure ratio, etc.;

[0051] In step 1, the process of the operating condition similarity analysis is as follows:

[0052] Obtain operating condition data corresponding to different historical convergence value groups, where the operating condition data includes but is not limited to environmental parameters (temperature, atmospheric pressure) and equipment status parameters (operating time, component health index);

[0053] Obtaining current operating condition data, normalizing the current operating condition data and the operating condition data corresponding to different historical convergence value groups, and after the normalization, calculating the Euclidean distance between the current operating condition data and the operating condition data corresponding to the historical convergence value groups to obtain an operating condition similarity value;

[0054] If the operating condition similarity value is less than or equal to the operating condition similarity threshold, the historical convergence value group is retrieved;

[0055] If the working condition similarity value is greater than the working condition similarity threshold, no operation is performed;

[0056] Summarize the retrieved historical convergence value groups to form a historical parameter feature set;

[0057] Step 2: updating the historical parameter feature set according to the surge margins corresponding to different historical convergence value groups in the historical parameter feature set;

[0058] In step 2, the process of updating the historical parameter feature set is as follows:

[0059] Obtain surge margins corresponding to different historical convergence value groups, and if the surge margins are less than a preset surge margin, delete the historical convergence value groups from the historical parameter feature set;

[0060] If the surge margin is greater than or equal to the preset surge margin, the historical convergence value group is retained in the historical parameter feature set;

[0061] After all historical convergence value groups are deleted or retained, the update of the historical parameter feature set is completed;

[0062] It should be noted that the surge margin corresponding to the historical convergence value group represents the minimum surge margin during a successful engine start under the operating conditions corresponding to the historical convergence value group;

[0063] Exemplarily, the preset surge margin is set by a technician. For example, a surge margin SM corresponding to a historical convergence value group is obtained. If SM is less than 1.1, a surge phenomenon is directly triggered, and the historical convergence value group is deleted. If 1.1≤SM<1.2, it indicates that the surge boundary is approached but a surge phenomenon is not triggered, and the historical convergence value group is retained. If 1.2≤SM, it indicates that the surge margin is high, and the historical convergence value group is retained.

[0064] Step 3: Calculate the deviation of the surge margin corresponding to the historical convergence value group of the parameters in the updated historical parameter feature set, assign the weighted proportion of the parameters according to the surge margin deviation calculation result, and calculate the initial guess value of the parameter in combination with the historical convergence value of the parameter;

[0065] In step 3, the surge margin deviation is calculated as follows:

[0066] Based on any parameter, the surge margin corresponding to different historical convergence value groups of the historical convergence value of the parameter is obtained, and the difference between the difference and the preset surge margin is calculated to obtain the surge margin deviation;

[0067] In step 3, the weighted proportion of the parameters is allocated according to the calculation results of the surge margin deviation, and the initial guess value of the parameters is calculated in combination with the historical convergence value of the parameters as follows:

[0068] The surge margin deviations corresponding to the various parameters are summed to obtain a total surge margin deviation. Based on any parameter, the ratio of the surge margin deviation corresponding to the parameter to the total surge margin deviation is calculated to obtain a weighted proportional weight of the parameter.

[0069] The historical convergence values ​​of the parameters in different historical convergence value groups are multiplied by the weighted proportion weights of the parameters to obtain the initial guess values ​​of the parameters;

[0070] It should be noted that the historical convergence value of the parameter corresponds to the weighted proportional weight of the parameter. Specifically, the weighted proportional weight of the parameter is obtained by performing deviation calculation and proportional calculation based on the surge margin corresponding to the historical convergence value group to which the historical convergence value of the parameter belongs.

[0071] It can be understood that the purpose of allocating the weighted proportion of the parameters according to the surge margin deviation calculation results and calculating the initial guess value of the parameters in combination with the historical convergence value of the parameters is to:

[0072] Function 1: Improve operational safety by assigning higher weight to data with higher surge margins, making the calculated initial guess closer to the parameters under safe operating conditions. For example, when determining the initial guess of engine speed, more reference is made to the speed under high surge margin conditions. This ensures that the initial guess can maintain stable engine operation during starting calculations, avoids surge caused by unreasonable parameters, and ensures safe engine starting and operation.

[0073] Function 2: Optimizing the reliability of the initial guess: Weighted calculations comprehensively consider the importance of data under different surge margin conditions, allowing the initial guess to more accurately reflect the parameter characteristics under safe and stable operating conditions. The resulting initial guess is more reliable, effectively reducing the probability of subsequent iterative calculations falling into non-physical solution regions due to unreasonable initial guesses, thereby improving iteration efficiency and calculation result accuracy.

[0074] Function 3: Strengthen the utilization of historical data. Surge margin is used as the key basis for data screening and weighting, making historical data screening more targeted. Instead of simply screening data based on environmental or equipment status, it is combined with core indicators of engine operating stability to allow historical data to better serve the generation of initial guesses, providing basic data that better meets actual operating needs for full-process parameter determination.

[0075] Function 4: Adapting to complex operating conditions. The engine surge risk varies under different operating conditions. The weighting mechanism can flexibly adjust the initial guess value. In harsh environments or when the equipment is in poor condition, the initial guess value can be calculated more based on high surge margin data, leaving sufficient safety margin. Under normal operating conditions, the weight of each data can also be reasonably balanced to ensure that the initial guess value is both safe and efficient, enhancing the engine's adaptability to complex operating conditions.

[0076] Step 4: Based on the volatility of the historical convergence values ​​of the parameters and the initial guess values ​​of the parameters, construct the initial guess range of each parameter and use the random number generator to generate multiple initial guess value groups;

[0077] In step 4, the process of constructing the initial guess range of each parameter based on the volatility of the historical convergence value of the parameter and taking the initial guess value of the parameter as a benchmark includes:

[0078] Based on any parameter, obtain the historical convergence value of the parameter in different historical convergence value groups, and integrate them to obtain the historical convergence sequence of the parameter;

[0079] Calculate the standard deviation of the historical convergence series of parameters And based on the initial guess value of the parameter, the initial guess range of the parameter is constructed, specifically: [initial guess value - k* , initial guess value + k* ], where k is the dynamic scaling factor;

[0080] Wherein, k is a dynamic scaling factor, which is set by those skilled in the art according to the distribution characteristics of the historical convergence value data of the parameter. For example, the dynamic scaling factor k can be obtained by:

[0081] A1. Calculate the mean, standard deviation, and quartile statistics of the historical convergence series, and use the boxplot method or a method based on the multiple of the standard deviation (e.g., data exceeding the mean ± 3 times the standard deviation is considered an outlier) to identify outliers in the historical convergence series.

[0082] A2. Formulate the rules for adjusting the k value, including:

[0083] A21, adjustment based on dispersion: If the standard deviation is small and the proportion of outliers is low, it means that the data distribution is concentrated. In this case, a smaller k value (such as 1.0-1.5) can be set to make the fluctuation range relatively narrow and the generated initial guess values ​​more concentrated around the benchmark value. Conversely, if the standard deviation is large and the proportion of outliers is high, it means that the data is highly dispersed. In this case, the k value should be increased (such as 2.0-3.0) to widen the fluctuation range and cover more possible parameter values.

[0084] A22, combined with the interquartile range (IQR) interval: calculate the ratio of IQR / (1.35×σ). This ratio reflects the discrete characteristics of the data distribution. If the ratio is small, it means that the data are relatively concentrated within the interquartile range, and the k value can be appropriately reduced. If the ratio is large, it means that the data outside the interquartile range fluctuates greatly, and the k value needs to be increased. For example, when IQR / (1.35×σ) < 1, the k value can be set to 1.2× the original k value; when IQR / (1.35×σ) > 1.5, the k value is set to 1.8× the original k value.

[0085] A23, set the k value boundary: in order to prevent the k value from being too small to cause the fluctuation range to be too narrow, miss reasonable initial guess value, or the k value is too large to make the fluctuation range too wide, increase the number of invalid initial guess value, need to set the upper and lower limit of k value, usually limit the k value between 1.0-3.0, if the calculated k value is less than 1.0, take 1.0; if greater than 3.0, take 3.0;

[0086] Based on the initial guess range of each parameter, a plurality of initial guess value groups are generated by using a random number generator;

[0087] It should be noted that each initial guess value group contains the initial guess value of each parameter, and is within the initial guess range of each parameter;

[0088] It can be understood that the role of generating multiple initial guess value groups is:

[0089] Effect one: expand the search range of solution space: since the determination of engine full-flow parameters involves complex nonlinear equations and physical constraints, the true solution may be in a relatively wide range. Generating multiple initial guess value groups, each of which represents a different search starting point, can explore the solution space from multiple directions, avoiding missing the true solution due to inaccurate single initial guess value. For example, if only one initial guess value is used, once the initial guess value deviates far from the true solution, the iterative calculation may converge to a local optimal solution or a non-physical solution. Multiple initial guess value groups can increase the probability of finding a global optimal solution and improve the accuracy of parameter determination;

[0090] Effect two: reduce the influence of initial guess value error: the initial guess value is estimated based on historical data and has certain error. By generating multiple initial guess value groups, the idea of "probability average" can be used to reduce the influence of single initial guess value error on the final result. Even if some initial guess value groups have large initial errors and lead to unsatisfactory calculation results, other initial guess value groups may still obtain results close to the true solution. Through the subsequent screening mechanism, more reliable solutions can be retained, improving the reliability of the overall calculation result;

[0091] Effect three: adapt to complex working conditions and model uncertainty: the engine operating conditions are complex and variable, and the model also has certain simplification and uncertainty. Multiple initial guess value groups can better adapt to these changes. Each initial guess value group can be calculated for different potential working conditions or model error conditions. For example, at different environmental temperatures and altitudes, the characteristics of the engine will differ. Multiple initial guess value groups can cover more possible working condition scenarios, ensuring that reasonable parameter solutions can be found under various conditions;

[0092] Step five: through similarity analysis on the initial guess value groups and the historical parameter feature set containing the surge margin unsatisfied historical convergence value groups, the initial guess value groups are screened;

[0093] Please refer to Figure 2As shown, in step 5, the process of screening the initial guess value group is:

[0094] Extract the historical convergence value group whose surge margin does not meet the requirements from the historical parameter feature set and mark it as a non-satisfied convergence value group;

[0095] Based on any initial guess value group, similarity calculation is performed with each non-satisfied convergence value group respectively. Specifically, the Euclidean distance between the initial guess value group and the non-satisfied convergence value group is calculated;

[0096] If, among all non-convergence value groups, the Euclidean distance between any non-convergence value group and the initial guess value group is less than the Euclidean distance threshold, the initial guess value group will be deleted;

[0097] If among all non-convergence value groups, there is no non-convergence value group whose Euclidean distance with the initial guess value group is less than the Euclidean distance threshold, then the initial guess value group will be retained;

[0098] It can be understood that the purpose of screening the preliminary guess value group by performing similarity analysis on the preliminary guess value group and the historical convergence value group whose surge margin does not meet the requirements contained in the historical parameter feature set is as follows: since the preliminary guess value group is randomly generated within the preliminary guess range of multiple parameters, although the preliminary guess range has taken the issue of surge margin into consideration, the preliminary guess range is only for individual parameters, so there is a problem: the preliminary guess value group randomly generated within the preliminary guess range of each parameter may have a distribution similarity of preliminary guess values ​​with the historical convergence value group whose surge margin does not meet the requirements. Therefore, by performing similarity analysis and deleting the preliminary guess value group that has a high distribution similarity with the historical convergence value group whose surge margin does not meet the requirements, the accuracy of parameter determination can be further improved;

[0099] Step 6: Use the selected initial guess value groups as input, combine the engine full-process test model and Newton iteration method to perform iterative calculations, select the target initial guess value group based on the iterative calculation results and iterative stability performance, and use the solution of the target initial guess value group after iterative calculation as the engine starting parameters;

[0100] In step 6, the full-process test model includes multiple established balance equations, including a mass balance equation (based on the flow balance between the compressor inlet and outlet, turbine, and combustion chamber within the engine), an energy balance equation (energy balance of the engine combustion chamber, turbine, and compressor), a speed balance equation, and a pressure balance equation.

[0101] The filtered initial guess value groups are used as input, and iterative calculations are performed in combination with the full-process engine test model and the Newton iteration method. After the iterative calculations are completed, the full-process parameters corresponding to each initial guess value group are obtained, and the surge margin is calculated.

[0102] If the surge margin is greater than or equal to the preset surge margin, the initial guess value group is marked as an initial compliance initial guess value group;

[0103] If the surge margin is less than the preset surge margin, the initial guess value group is marked as a non-compliant initial guess value group;

[0104] It should be noted that the iterative calculation process of the Newton iteration method is: using the full-process test model and taking the initial guess value group as input, comparing the output results with the engine test data, obtaining the deviation, and correcting and updating the initial guess values ​​in the initial guess value group, and then re-inputting the corrected and updated initial guess value group into the full-process test model until the deviation meets the residual accuracy requirements. This technical method is an existing theoretical technology and will not be described in detail here.

[0105] Based on the initial match and initial guess value group, a surge margin calculated after each iteration of the initial match and initial guess value group in the iteration process is obtained, and compared with a preset surge margin;

[0106] If the surge margin after iterative calculation is greater than or equal to the preset surge margin, the iteration is marked as a surge qualified iteration;

[0107] If the surge margin after iterative calculation is less than the preset surge margin, the iteration is marked as a non-surge qualified iteration;

[0108] Count the number of surge qualified iterations to obtain the qualified iteration ratio;

[0109] Based on the surge qualified iteration, the surge margin of each surge qualified iteration is calculated to deviate from the preset surge margin to obtain an iterative surge deviation, and all iterative surge deviations are averaged to obtain an iterative surge qualified degree value;

[0110] Calculate the coefficient of variation of all iterative surge deviations to obtain the iterative surge stability value;

[0111] The iteration qualified ratio, the iteration surge qualified degree value and the iteration surge stability value are summed to obtain the iteration performance value;

[0112] The initial guess value group with the largest iterative performance value is selected as the target initial guess value group, and the solution after the iterative calculation of the target initial guess value group is used as the full process parameter of the engine;

[0113] It should be noted that the iterative performance value is obtained by comprehensively outputting the qualified ratio, qualified degree and qualified stability of the initial guess value group during the iterative calculation process. The initial qualified initial guess value group with the largest iterative performance value is selected as the target initial guess value group, and the solution after the iterative calculation of the target initial guess value group is used as the full process parameter of the engine, which can improve the accuracy of parameter determination.

[0114] The technical solution of the embodiment of the present invention is as follows: by analyzing the similarity of working conditions, the historical convergence values ​​of various parameters are retrieved from the engine history start database to construct a historical parameter feature set; the historical parameter feature set is updated according to the surge margins corresponding to different historical convergence value groups in the historical parameter feature set; the deviation is calculated by the surge margins corresponding to the historical convergence value groups of the parameters in the updated historical parameter feature set, the weighted proportion weight of the parameters is allocated according to the surge margin deviation calculation result, and the initial guess value of the parameter is calculated in combination with the historical convergence value of the parameter; based on the volatility of the historical convergence value of the parameter and taking the initial guess value of the parameter as a benchmark, the initial guess value of the parameter is constructed. The method adopts a novel method for determining the initial guess value group and a random number generator to generate multiple initial guess value groups; the initial guess value group is screened by performing similarity analysis on the initial guess value group and the historical convergence value group whose surge margin does not meet the requirements contained in the historical parameter feature set; the screened initial guess value groups are respectively used as input, and iterative calculations are performed in combination with the full-process engine test model and the Newton iteration method; a target initial guess value group is selected according to the iterative calculation results and the iterative stability performance; and the solution of the target initial guess value group after iterative calculation is used as the starting parameter of the engine. The present invention prevents the occurrence of non-physical solutions when determining the engine parameters, thereby improving the accuracy of determining the full-process engine parameters.

[0115] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for determining parameters and component performance during the entire engine starting process, characterized by: include: Through the similarity analysis of working conditions, the historical convergence values ​​of various parameters are retrieved from the engine history start database to construct the historical parameter feature set; The historical parameter feature set is updated according to surge margins corresponding to different historical convergence value groups in the historical parameter feature set; Deviation calculation is performed based on the surge margin corresponding to the historical convergence value group of the parameters in the updated historical parameter feature set. Weighted proportional weights of the parameters are allocated based on the surge margin deviation calculation results. The initial guess values ​​of the parameters are calculated based on the historical convergence values ​​of the parameters. The surge margin deviation is calculated as follows: Based on any parameter, the surge margin corresponding to different historical convergence value groups of the historical convergence value of the parameter is obtained, and the difference between the difference and the preset surge margin is calculated to obtain the surge margin deviation; The process of calculating the initial guess value of the parameter is as follows: The surge margin deviations corresponding to the various parameters are summed to obtain a total surge margin deviation. Based on any parameter, the ratio of the surge margin deviation corresponding to the parameter to the total surge margin deviation is calculated to obtain a weighted proportional weight of the parameter. The historical convergence values ​​of the parameters in different historical convergence value groups are multiplied by the weighted proportion weights of the parameters to obtain the initial guess values ​​of the parameters; Based on the volatility of the historical convergence values ​​of the parameters and taking the initial guess values ​​of the parameters as a benchmark, the initial guess range of each parameter is constructed, and multiple initial guess value groups are generated using a random number generator; The process of constructing the initial guess range of each parameter includes: Based on any parameter, obtain the historical convergence value of the parameter in different historical convergence value groups, and integrate them to obtain the historical convergence sequence of the parameter; Calculate the standard deviation of the historical convergence series of parameters And based on the initial guess value of the parameter, the initial guess range of the parameter is constructed, specifically: [initial guess value - k* , initial guess value + k* ], where k is the dynamic scaling factor; The initial guess value group is screened by performing similarity analysis on the initial guess value group and the historical convergence value group whose surge margin does not meet the requirements contained in the historical parameter feature set; The screened initial guess value groups are used as input, and iterative calculations are performed in combination with the engine full-process test model and the Newton iteration method. The target initial guess value group is selected based on the iterative calculation results and iterative stability performance, and the solution after iterative calculation of the target initial guess value group is used as the engine starting parameters.

2. The method for determining parameters and component performance during the entire engine starting process according to claim 1, characterized in that: The process of the working condition similarity analysis is as follows: Obtain the operating condition data corresponding to different historical convergence value groups in the historical start database and the current operating condition data, and after normalization, use Euclidean distance calculation to obtain the operating condition similarity value between the current operating condition data and the operating condition data corresponding to the historical convergence value group; If the operating condition similarity value meets the requirements, the historical convergence value group is retrieved and summarized to obtain the historical parameter feature set.

3. The method for determining parameters and component performance during the entire engine starting process according to claim 1, characterized in that: The process of updating the historical parameter feature set is as follows: The surge margins corresponding to different historical convergence value groups are obtained. If the surge margins meet the requirements, the historical convergence value groups are retained in the historical parameter feature set; otherwise, they are deleted.

4. The method for determining parameters and component performance during the entire engine starting process according to claim 1, characterized in that: The process of screening the initial guess value group is as follows: Extract the historical convergence value group whose surge margin does not meet the requirements from the historical parameter feature set and mark it as a non-satisfied convergence value group; Based on any initial guess value group, calculate the Euclidean distance with each non-convergence value group respectively; If, among all non-convergence value groups, the Euclidean distance between any non-convergence value group and the initial guess value group does not meet the requirement, the initial guess value group will be deleted; otherwise, the initial guess value group will be retained.

5. The method for determining parameters and component performance during the entire engine starting process according to claim 1, characterized in that: The process of selecting the target initial guess value group based on the iterative calculation results and iterative stability performance is as follows: Obtain the full process parameters corresponding to each initial guess value group after the iterative calculation is completed, and calculate the surge margin; If the surge margin is greater than or equal to the preset surge margin, the initial guess value group is marked as an initial compliance initial guess value group; Marking an iteration in which the surge margin of the initial guess value group is greater than or equal to the preset surge margin during the iteration process as a surge qualified iteration; By analyzing the number of surge qualified iterations and the corresponding iterative surge deviation, the iterative qualified ratio, iterative surge qualified degree value and iterative surge stability value are obtained, and the iterative performance value is obtained by comprehensive output; The initial guess value group with the largest iterative performance value is selected as the target initial guess value group, and the solution after the iterative calculation of the target initial guess value group is completed is used as the full process parameter of the engine.

6. The method for determining parameters and component performance during the entire engine starting process according to claim 5, characterized in that: The qualified iteration ratio is the ratio of the number of surge qualified iterations.

7. The method for determining parameters and component performance during the entire engine starting process according to claim 5, characterized in that: The iterative surge qualification value and the iterative surge stability value are obtained as follows: Calculate the deviation between the surge margin of each surge qualified iteration and the preset surge margin to obtain an iterative surge deviation, average all iterative surge deviations to obtain an iterative surge qualified degree value; The coefficient of variation of all iterative surge deviations is calculated to obtain the iterative surge stability value.

Images