A rapid determination method for the highest thermal efficiency range of an engine

Abstract

The present invention is applicable to the technical field of engine thermal efficiency measurement, and provides a method for quickly measuring the highest thermal efficiency range of an engine, including the following steps: setting a basic test for a constant air-fuel ratio condition at a target speed; performing data dimensionality reduction on the test data through the t-SNE algorithm; performing a non-linear correlation test using the Kendall correlation coefficient; performing data inverse transformation on the dimensionality-reduced data using the linear interpolation reconstruction method; performing non-linear fitting on the reconstructed data using the least squares fitting method; and defining the highest thermal efficiency range based on the extreme points of the fitting equation. This method can be used to find the highest thermal efficiency range during the actual engine R & D test process, and can provide a standardized and energy-saving test method, enabling the accurate finding of the high-efficiency thermal efficiency point of the engine while saving a large amount of workload.

Description

Technical Field

[0001] The invention belongs to the technical field of engine thermal efficiency determination, and in particular relates to a method for quickly determining the maximum thermal efficiency range of an engine. Background Art

[0002] The maximum thermal efficiency of an engine refers to the maximum thermal efficiency that the engine can achieve under specific working conditions. The key factors that affect the maximum thermal efficiency of an engine are often multifaceted, including combustion efficiency, heat loss, and system integration. Engine manufacturers and engineers must consider the efficiency performance of the engine under different working conditions and find the best working point under different loads and speeds, that is, the point of maximum thermal efficiency.

[0003] The search for the highest thermal efficiency point is often complex and multi-factorial. It usually involves analyzing the engine's operating characteristics and performance curves and conducting a systematic analysis. Engine manufacturers and engineers generally conduct repeated measurements based on experience and gradually search for the point. The traditional test methods used rely on the actual test experience of test engineers, and then conduct multiple rounds of tests on a large number of operating points, which is often blind and non-standard, time-consuming and labor-intensive, resulting in a large waste of resources. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide a method for quickly determining the maximum thermal efficiency range of an engine, aiming to solve the problems raised in the above-mentioned background technology.

[0005] The embodiment of the present invention is implemented as follows: a method for quickly determining the maximum thermal efficiency range of an engine comprises the following steps:

[0006] Step 1: Set up a basic test at a constant air-fuel ratio at target speed;

[0007] Step 2: Perform data dimension reduction on the test data using the t-SNE algorithm;

[0008] Step 3: Use Kendall correlation coefficient to perform nonlinear correlation test;

[0009] Step 4: Use the linear interpolation reconstruction method to perform inverse transformation on the reduced-dimensional data;

[0010] Step 5: Use the least squares fitting method to perform nonlinear fitting on the reconstructed data;

[0011] Step 6: Define the highest thermal efficiency interval based on the extreme points of the fitting equation.

[0012] According to a further technical solution, step 1 comprises the following steps:

[0013] Carry out basic test of fixed air-fuel ratio under target engine speed. By adjusting injection pressure and injection advance angle, perform variable parameter adjustment test. Statistic the two parameters of cylinder average peak temperature and thermal efficiency in combustion data. Perform engine performance test by continuously changing test parameters.

[0014] A set of in-cylinder average peak temperature and thermal efficiency data is obtained. The in-cylinder average peak temperature is set as the x-axis and the thermal efficiency is set as the y-axis. The in-cylinder average peak temperature and thermal efficiency data of this set of data are recorded as:

[0015] X={(x 1 ,y 1 ),(x 2 ,y 2 ),…(x n ,y n )}

[0016] A further technical solution is that the number of test operating points is not less than 50.

[0017] A further technical solution is that in step 2, the distance between each pair of data points is first calculated, and the conditional probability distribution of the data points in the high-dimensional space is calculated based on the distance between the data points;

[0018] Similarity:

[0019]

[0020] The similarity in high-dimensional space, that is, the conditional probability distribution, is:

[0021]

[0022] Where X i , X j and X k are different data points, ‖(X i -X j )‖ represents the Euclidean distance between data points, σ ​​is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the decay speed of similarity. For low latitudes, it is specified as

[0023] The similarity at low latitude, that is, the conditional probability distribution is:

[0024]

[0025] Optimize the distance KL divergence (Kullback-Leibler divergences) between two distributions. The objective function is:

[0026]

[0027] Through the gradient descent algorithm, the position of the data points in the low-dimensional embedding space is optimized to minimize the KL divergence. The gradient in the low-dimensional space is calculated as:

[0028]

[0029] Convert the Gaussian distribution to a t distribution and optimize the gradient:

[0030]

[0031] Among them, Y i It is the data after dimensionality reduction by t-SNE algorithm.

[0032] Further technical solution, the step 3 comprises the following specific steps:

[0033] The average peak temperature data point in the cylinder after dimension reduction is recorded as:

[0034] A={a 1, a 2 ,…a n}

[0035] The corresponding thermal efficiency data points after dimensionality reduction are recorded as:

[0036] B = {b 1, b 2 ,…b n}

[0037] For variables A and B, sort the data points and assign ranks. If there are repeated data points, use the average rank. For each pair of data points (a i ,b i ) and (a j ,b j ), compute the signed difference data:

[0038] d i =sign(A i -A j )×sign(B i -B j )(i <j)

[0039] in:

[0040]

[0041] The Kendall correlation coefficient is:

[0042]

[0043] When τ = 1, it means that the two variables are completely positively correlated; when τ = -1, it means that the two variables are completely negatively correlated; when τ = 0, it means that the two variables are unorderedly correlated;

[0044] The Z statistic for the Kendall correlation coefficient is:

[0045]

[0046] Where τ is the calculated Kendall correlation coefficient, and n is the sample size;

[0047] The P value is calculated from the Z statistic using the cumulative distribution function of the standard normal distribution:

[0048] P=2(1-Φ(|Z|))

[0049]

[0050] If the calculated Kendall correlation is greater than 0.5 and the P value is less than 0.05, it is considered that there is a significant correlation between the two variables, which means that there is a significant nonlinear correlation between the average peak temperature in the cylinder and the thermal efficiency.

[0051] According to a further technical solution, step 4 comprises the following steps:

[0052] Weight calculation:

[0053]

[0054] For each reduced-dimensional data point, ‖(Y i -Y j )‖ represents the Euclidean distance between it and other data points;

[0055] Set its own weight to 0, ω ii =0;

[0056] Normalize the weight of each data point so that the sum of the weights of each data point is 1:

[0057]

[0058] Each data point after dimensionality reduction is transformed into the original data point X 1 ={(x 1 ,y 1 ),(x 2 ,y 2 ),…(x n ,y n ) is used to perform a weighted linear combination to reconstruct the original data point (x j ,y j ):

[0059]

[0060] According to a further technical solution, step 5 comprises the following steps:

[0061] Convert the inverse transformed data points into the design matrix X:

[0062]

[0063] Use the least squares method to calculate the optimal parameters:

[0064]

[0065] in, and are the parameters of the fitted quadratic equation.

[0066] Use the optimal parameters obtained Make predictions for new input data x:

[0067]

[0068] The coefficient of determination R 2 Evaluate the fitting effect, the value range is 0 to 1, the closer to 1, the better the fitting effect, and vice versa; calculate the residual sum of squares and the total sum of squares to calculate the coefficient of determination R 2 :

[0069]

[0070] A further technical solution is that in step 6, the highest point M of the fitting curve, i.e. Point, given temperature range [M-10,M+10], is the highest thermal efficiency range.

[0071] An embodiment of the present invention provides a method for quickly determining the maximum thermal efficiency range of an engine. The method can be used to find the highest thermal efficiency range in actual application in the engine development and testing process, and can provide a standardized and energy-saving testing method, so that the engine's high thermal efficiency point can be accurately found while saving a lot of workload. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 A calculation flow chart of a method for quickly determining the maximum thermal efficiency range of an engine provided by an embodiment of the present invention;

[0073] Figure 2 The relevant data of Example 1 (a: average peak temperature and thermal efficiency in the cylinder; b: data after t-SNE dimension reduction; c: inverse transformation data-fitting);

[0074] Figure 3 These are the relevant data of Example 2 (a: average peak temperature and thermal efficiency in the cylinder; b: data after t-SNE dimensionality reduction; c: inverse transformation data-fitting). DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail 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 invention and are not intended to limit the present invention.

[0076] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0077] An embodiment of the present invention provides a method for quickly determining the maximum thermal efficiency range of an engine, comprising the following steps:

[0078] Step 1: Set up a basic test at a constant air-fuel ratio at target speed;

[0079] Step 2: Perform data dimension reduction on the test data using the t-SNE algorithm;

[0080] Step 3: Use Kendall correlation coefficient to perform nonlinear correlation test;

[0081] Step 4: Use the linear interpolation reconstruction method to perform inverse transformation on the reduced-dimensional data;

[0082] Step 5: Use the least squares fitting method to perform nonlinear fitting on the reconstructed data;

[0083] Step 6: Define the highest thermal efficiency interval based on the extreme points of the fitting equation.

[0084] As a preferred embodiment of the present invention, the step 1 comprises the following steps:

[0085] By conducting a basic test of a fixed air-fuel ratio at a certain engine target speed, adjusting the injection pressure and injection advance angle, and conducting a variable parameter adjustment test, the two parameters of the average peak temperature in the cylinder and the thermal efficiency in the combustion data are statistically analyzed, and the test parameters are continuously changed to conduct engine performance testing;

[0086] A set of in-cylinder average peak temperature and thermal efficiency data is obtained. The in-cylinder average peak temperature is set as the x-axis and the thermal efficiency is set as the y-axis. The in-cylinder average peak temperature and thermal efficiency data of this set of data are recorded as:

[0087] X={(x 1 ,y 1 ),(x 2 ,y 2 ),…(x n ,y n )}

[0088] In the embodiment of the present invention, the number of test operating points is not less than 50.

[0089] As a preferred embodiment of the present invention, in step 2, the distance between each pair of data points is first calculated, and the conditional probability distribution of the data points in the high-dimensional space is calculated based on the distance between the data points. This can be done by converting the distance into probability through a Gaussian kernel function.

[0090] Similarity:

[0091]

[0092] The similarity in high-dimensional space, that is, the conditional probability distribution, is:

[0093]

[0094] Where X i , X j and X k are different data points, ‖(X i -X j )‖ represents the Euclidean distance between data points, σ ​​is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the decay speed of similarity. For low latitudes, it is specified as

[0095] The similarity at low latitude, that is, the conditional probability distribution is:

[0096]

[0097] If the dimensionality reduction effect is good and the local features are preserved intact, then p j|i =q j|i , so the distance KL divergence (Kullback-Leibler divergences) between the two distributions is optimized, and the objective function is:

[0098]

[0099] Through the gradient descent algorithm, the position of the data points in the low-dimensional embedding space is optimized to minimize the KL divergence. The gradient in the low-dimensional space is calculated as:

[0100]

[0101] Convert the Gaussian distribution to a t distribution and optimize the gradient:

[0102]

[0103] Among them, Y i It is the data after dimensionality reduction by t-SNE algorithm.

[0104] As a preferred embodiment of the present invention, step 3 includes the following specific steps:

[0105] The average peak temperature data point in the cylinder after dimension reduction is recorded as:

[0106] A={a 1, a 2 ,…a n}

[0107] The corresponding thermal efficiency data points after dimensionality reduction are recorded as:

[0108] B = {b 1, b 2 ,…b n}

[0109] For variables A and B, sort the data points and assign ranks. If there are duplicate data points, use the average rank. i ,b i ) and (a j ,b j ), compute the signed difference data:

[0110] d i =sign(A i -A j )×sign(B i -B j ) (i <j)

[0111] in:

[0112]

[0113] The Kendall correlation coefficient is:

[0114]

[0115] When τ=1, it means that the two variables are completely positively correlated; when τ=-1, it means that the two variables are completely negatively correlated; when τ=0, it means that the two variables are unorderedly correlated.

[0116] The correlation needs to be tested, and the test requires calculating the P value representing its confidence level. When calculating the Kendall correlation coefficient, the following assumptions are usually tested:

[0117] Null hypothesis (H 0 ): There is no correlation between the two variables (the correlation coefficient is 0).

[0118] The alternative hypothesis (H 1 ): There is a correlation between the two variables (the correlation coefficient is not 0).

[0119] The Z statistic for the Kendall correlation coefficient is:

[0120]

[0121] Where τ is the calculated Kendall correlation coefficient and n is the sample size.

[0122] The P value is calculated from the Z statistic using the cumulative distribution function (CDF) of the standard normal distribution:

[0123] P=2(1-Φ(|Z|))

[0124]

[0125] If the calculated Kendall correlation is greater than 0.5 and the P value is less than 0.05, it is considered that there is a significant correlation between the two variables, which means that there is a significant nonlinear correlation between the average peak temperature in the cylinder and the thermal efficiency.

[0126] As a preferred embodiment of the present invention, step 4 comprises the following steps:

[0127] Weight calculation:

[0128]

[0129] For each reduced-dimensional data point, ‖(Y i -Y j )‖ represents the Euclidean distance between it and other data points;

[0130] Set its own weight to 0, ω ii =0;

[0131] Normalize the weight of each data point so that the sum of the weights of each data point is 1:

[0132]

[0133] Each data point after dimensionality reduction is transformed into the original data point X 1 ={(x 1 ,y 1 ),(x 2 ,y 2 ),…(x n ,y n ) is used to perform a weighted linear combination to reconstruct the original data point (x j ,y j ):

[0134]

[0135] As a preferred embodiment of the present invention, step 5 comprises the following steps:

[0136] Convert the inverse transformed data points into the design matrix X:

[0137]

[0138] Use the least squares method to calculate the optimal parameters:

[0139]

[0140] in, and are the parameters of the fitted quadratic equation.

[0141] Use the optimal parameters obtained Make predictions for new input data x:

[0142]

[0143] The coefficient of determination R 2 Evaluate the fitting effect, the value range is 0 to 1, the closer to 1, the better the fitting effect, and vice versa. Calculate the residual sum of squares and the total sum of squares to calculate the coefficient of determination R 2 :

[0144]

[0145] As a preferred embodiment of the present invention, in step 6, the highest point M of the fitting curve, i.e. Point, given temperature range [M-10,M+10], is the highest thermal efficiency range.

[0146] Two specific examples are provided below to verify the effectiveness of this method:

[0147] The test operating point of Example 1 is 50°C, and the average peak temperature and thermal efficiency in the cylinder are as follows: Figure 2 a; the test operating point of Example 2 is 70, and the average peak temperature and thermal efficiency in the cylinder are as shown in Figure 3 The data of Example 1 and Example 2 after t-SNE dimension reduction are shown in Fig. Figure 2 b and Figure 3 As shown in b, the Kendall correlation coefficient in Example 1 is 0.78575; the Kendall correlation coefficient in Example 2 is 0.751553. The P value in Example 1 is 2.9125e -35 , then the null hypothesis is rejected, indicating that there is a significant correlation between the two variables; the P value in Example 2 is 3.7028e -20 , then the null hypothesis is rejected, indicating that there is a significant correlation between the two variables.

[0148] In Example 1 and -0.0015, 0.8954, -15.0194, R2 is 0.9387; in Example 2 and They are -1.78e respectively -4 , 0.6748, -1.6078, R 2 It is 0.7551.

[0149] Finally Figure 2 c and Figure 3 As shown in c, it can be seen that the highest thermal efficiency ranges of Examples 1 and 2 are [1773-1793] at an air-fuel ratio of 26 and [1750-1770] at an air-fuel ratio of 28.5, respectively.

[0150] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for quickly determining the maximum thermal efficiency range of an engine, characterized in that: The following steps are involved: Step 1: Set up a basic test at a constant air-fuel ratio at target speed; Step 2: Perform data dimension reduction on the test data using the t-SNE algorithm; Step 3: Use the Kendall correlation coefficient to perform a coherence test on the data processed in step 2. The data that pass the test in step 3 proceed to step 4 for processing. If the data fail the test in step 3, return to step 1 for additional testing. Step 4: Use the linear interpolation reconstruction method to perform inverse transformation on the reduced-dimensional data; Step 5: Use the least squares fitting method to perform nonlinear fitting on the reconstructed data; Step 6: Define the highest thermal efficiency interval based on the extreme points of the fitting equation.

2. The method for rapidly determining the engine's maximum thermal efficiency range according to claim 1, characterized in that: The step 1 comprises the following steps: Carry out basic test of fixed air-fuel ratio under target engine speed. By adjusting injection pressure and injection advance angle, perform variable parameter adjustment test. Statistic the two parameters of cylinder average peak temperature and thermal efficiency in combustion data. Perform engine performance test by continuously changing test parameters. A set of in-cylinder average peak temperature and thermal efficiency data is obtained. The in-cylinder average peak temperature is set as the x-axis and the thermal efficiency is set as the y-axis. The in-cylinder average peak temperature and thermal efficiency data of this set of data are recorded as: 。 3. The method for quickly determining the maximum thermal efficiency range of an engine according to claim 2, characterized in that: The number of test conditions shall not be less than 50.

4. The method for quickly determining the maximum thermal efficiency range of an engine according to claim 2, characterized in that: In step 2, the distance between each pair of data points is first calculated, and the conditional probability distribution of the data points in the high-dimensional space is calculated based on the distance between the data points; Similarity: ； The similarity in high-dimensional space, that is, the conditional probability distribution, is: ； in X i , X j and X k are different data points, represents the Euclidean distance between data points, is the bandwidth parameter of the Gaussian kernel function, which is used to adjust the decay rate of similarity. For low latitudes, it is specified as ; The similarity at low latitude, that is, the conditional probability distribution is: ； Optimize the distance KL divergence between two distributions, the objective function is: ； Through the gradient descent algorithm, the position of the data points in the low-dimensional embedding space is optimized to minimize the KL divergence. The gradient in the low-dimensional space is calculated as: ； Convert the Gaussian distribution to a t distribution and optimize the gradient: ； in, Y i It is the data after dimensionality reduction by t-SNE algorithm.

5. The method for quickly determining the engine's maximum thermal efficiency range according to claim 4, characterized in that: The step 3 comprises the following specific steps: The average peak temperature data point in the cylinder after dimension reduction is recorded as: ； The corresponding thermal efficiency data points after dimensionality reduction are recorded as: ； For variables A and B , sort the data points and assign ranks. If there are repeated data points, use the average rank. For each pair of data points and , compute the signed difference data: ； in: ； The Kendall correlation coefficient is: ； when When , it means that the two variables are completely positively correlated; when When , it means that the two variables are completely negatively correlated; when When , it means that the two variables are not related in order; Kendall's correlation coefficient Z The statistics are: ； in, is the calculated Kendall correlation coefficient, n is the sample size; according to Z Statistics, calculated using the cumulative distribution function of the standard normal distribution P value: ； ； If the calculated Kendall correlation is greater than 0.5, P If the value is less than 0.05, it is considered that there is a significant correlation between the two variables, which means that there is a significant nonlinear correlation between the average peak temperature in the cylinder and the thermal efficiency.

6. The method for quickly determining the engine's highest thermal efficiency range according to claim 5, characterized in that: The step 4 comprises the following steps: Weight calculation: ； For each reduced data point, Represents the Euclidean distance between it and other data points; Set its own weight to 0, ; Normalize the weight of each data point so that the sum of the weights of each data point is 1: ； Each data point after dimensionality reduction is transformed into the original data point Perform weighted linear combination to reconstruct the original data points : 。 7. The method for quickly determining the engine's highest thermal efficiency range according to claim 6, characterized in that: The step 5 comprises the following steps: Convert the inverse transformed data points into the design matrix X : ； Use the least squares method to calculate the optimal parameters: ； ； in, , and are the parameters of the fitted quadratic equation; Use the optimal parameters obtained Make predictions for new input data x: ； Coefficient of determination R 2 Evaluate the fitting effect, the value range is 0 to 1, the closer to 1, the better the fitting effect, and vice versa; calculate the residual sum of squares and the total sum of squares to calculate the determination coefficient R 2 : ； ； 。 8. The method for quickly determining the maximum thermal efficiency range of an engine according to claim 7, characterized in that: In step 6, the highest point M of the fitting curve, i.e. Point, given temperature range [M-10, M+10], is the highest thermal efficiency range sought.

Images