Vehicle power system thermal efficiency characterization parameter optimization method

Abstract

The invention belongs to the technical field of vehicle power systems, and particularly relates to a vehicle power system thermal efficiency characterization parameter optimization method, which comprises the following steps of S1, constructing a power system data set; s2, screening heat efficiency characterization parameters; s3, performing thermal efficiency evaluation by using the neural network model; s4, comparing thermal efficiency evaluation precision; and S5, outputting a preferred thermal efficiency characterization parameter set. According to the vehicle power system thermal efficiency characterization parameter optimization method, the thermal efficiency characterization parameters are screened through a plurality of correlation coefficient feature extraction methods, and neural network model prediction analysis is combined, so that optimization of the thermal efficiency characterization parameters of the vehicle power system is realized; and a high-confidence representation parameter set is provided for vehicle power system degradation state evaluation. The method can be applied to the degradation state evaluation of any fuel vehicle power system.

Description

Technical Field

[0001] This invention belongs to the technical field of vehicle power systems, specifically to a method for optimizing thermal efficiency characterization parameters for vehicle power systems. This method extracts characterization parameters using three filtering feature extraction methods: the Pearson correlation coefficient, the Spearman correlation coefficient, and the grey relational degree. This method, combined with a neural network model for thermal efficiency prediction and analysis, optimizes the characterization parameter set that provides the most accurate thermal efficiency prediction, providing a high-confidence characterization parameter set for assessing the degradation state of the vehicle power system. This method is applicable to thermal efficiency assessments of any fuel-powered vehicle power system. Background Art

[0002] With the continuous advancement of information technology and intelligentization in vehicle powertrains, monitoring and evaluating powertrain performance degradation has become a key challenge in preventive maintenance. Traditional powertrain performance prediction primarily focuses on component-level testing, using engine sensor parameters such as engine speed, oil pressure, oil temperature, intake air temperature, and output torque for condition prediction and fault diagnosis. However, with the advancement of information technology, data-driven systems have become a powerful tool for equipment condition monitoring. A key component of data-driven powertrain thermal efficiency assessment is the selection of parameters that characterize thermal efficiency. Based on thermodynamic and physical model analysis, 15 parameters related to thermal efficiency have been identified, including circulating oil volume, rail pressure, oil temperature, speed, torque, coolant flow rate, coolant temperature, coolant specific heat, intake air temperature, intake pressure, intake air flow rate, exhaust temperature, exhaust pressure, exhaust flow rate, and ambient temperature. The Pearson correlation coefficient analysis method has been widely used in determining characterization parameters due to its simplicity, intuitiveness, and ease of understanding. However, it has certain drawbacks in analyzing correlations in multivariate and nonlinear data. This results in poor adaptability of the selected thermal efficiency characterization parameters to operating conditions, leading to inaccurate state assessments under certain operating conditions and inhibiting widespread application of powertrain thermal efficiency state assessment. Research on methods for optimizing powertrain thermal efficiency characterization parameters, selecting more adaptable correlation analysis methods, and optimizing powertrain thermal efficiency characterization parameters are crucial for improving vehicle powertrain degradation state assessment capabilities. Summary of the Invention

[0003] (1) Technical issues to be solved

[0004] The technical problem to be solved by the present invention is: how to provide a method for optimizing the thermal efficiency characterization parameters of a vehicle power system, which can be applied to the thermal efficiency status evaluation of any fuel vehicle power system.

[0005] (2) Technical solution

[0006] To solve the above technical problems, the present invention provides a method for optimizing parameters characterizing thermal efficiency of a vehicle power system, the method comprising the following steps:

[0007] Step S1: construct a power system dataset;

[0008] Step S2: screening of thermal efficiency characterization parameters;

[0009] Step S3: using a neural network model to evaluate thermal efficiency;

[0010] Step S4: thermal efficiency evaluation accuracy comparison;

[0011] Step S5: outputting the optimal thermal efficiency characterization parameter set.

[0012] Wherein, in said step S1, a power system data set is constructed;

[0013] The power system test data is input in time series. The input data contains multiple time steps. Each time step includes 15 heat flow characterization parameters, including circulating oil volume, rail pressure, oil temperature, speed, torque, coolant flow, coolant temperature, coolant specific heat, intake temperature, intake pressure, intake flow, exhaust temperature, exhaust pressure, exhaust flow, and ambient temperature.

[0014] Wherein, in said step S2, thermal efficiency characterization parameters are screened;

[0015] Three filtering characterization parameter extraction methods, namely Pearson correlation coefficient, Spearman correlation coefficient and grey relational degree, were established.

[0016] Wherein, in step S2, the Pearson correlation coefficient method is as follows:

[0017] The Pearson correlation coefficient is used to analyze the linear correlation between factors, and its definition is as follows:

[0018]

[0019] In the formula, r is the Pearson correlation coefficient, n is the number of samples, and are the average values ​​of the two samples, σ X and σ Y X i and Y i Standard score of

[0020] When the correlation coefficient |r| ≥ 0.2, the characterization parameters have a weak linear correlation with the thermal efficiency. Therefore, seven parameters, including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature, are selected as the characterization parameters for thermal efficiency evaluation using the Pearson method.

[0021] Wherein, in step S2, the Spearman correlation coefficient method is as follows: The Spearman correlation coefficient is used to analyze the nonlinear correlation between variables, and its definition is as follows:

[0022]

[0023] Where ρ is the Spearman correlation coefficient, n is the number of samples, and d i is the rank difference between the two samples;

[0024] The value range of ρ is [-1, 1]. The closer its absolute value is to 1, the stronger the correlation between variables. When |ρ| ≥ 0.5, the five parameters of torque, circulating oil volume, rail pressure, intake pressure, and exhaust pressure are selected as the characterization parameters of the Spearman method thermal efficiency evaluation.

[0025] Wherein, in step S2, the grey relational method is as follows:

[0026] Grey relational degree is used to analyze the development trend and change pattern of a system over time, that is, the characteristics and degree of dynamic correlation between factors. Its definition is as follows:

[0027]

[0028] In the formula, γ is the grey relational degree, n is the number of samples, and ξ(k) is the grey relational coefficient, as shown in the following formula;

[0029]

[0030] Where, Δ min is the minimum absolute value of the difference between the two sequence values, Δ max is the maximum absolute value of the difference between the two sequence values, ρ is the resolution coefficient, Δ t(k) is the absolute value of the difference between the two sequence values;

[0031] The larger the γ value, the stronger the correlation. When γ ≥ 0.6, six parameters including coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume are selected as the characterization parameters of thermal efficiency evaluation using the grey correlation method.

[0032] Wherein, in said step S3, thermal efficiency evaluation is performed using a neural network model;

[0033] With the vehicle power system test data as input and the thermal efficiency value as output, a neural network model is established. The thermal efficiency evaluation characterization parameters determined by the Pearson correlation coefficient, Spearman correlation coefficient and grey correlation degree are used to evaluate the thermal efficiency of the vehicle power system.

[0034] Wherein, the step S3 specifically includes:

[0035] 1) Input layer: Input the power system test data in time series. The input data contains multiple time steps. Each time step includes nine heat flow characterization parameters: circulating oil volume, rail pressure, speed, torque, coolant temperature, intake temperature, intake pressure, exhaust temperature, and exhaust pressure.

[0036] 2) The data screening layer performs data screening: data screening is performed based on the characterization parameters determined by the Pearson correlation coefficient method, the Spearman correlation coefficient method, and the grey relational method;

[0037] (1) Pearson method verification data screening: Data containing seven characteristic parameters including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature are screened from the power system test data to form the Pearson method verification data;

[0038] (2) Spearman method verification data screening: The data containing five characteristic parameters, namely torque, circulating oil volume, rail pressure, intake pressure and exhaust pressure, are screened from the power system test data to form the Spearman method verification data;

[0039] (3) Grey correlation method validation data screening: Data containing six characterization parameters, namely coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume, are screened from the power system test data to form the grey correlation method validation data;

[0040] 3) Establish CNN layer 1: Establish a first-level convolutional neural network to extract local temporal features of verification data;

[0041] (1) Establish convolution layer 1: Establish a convolution layer with 32 convolution kernels and a convolution kernel size of 3 to extract local temporal features from time series data;

[0042] (2) Establish pooling layer 1: Establish a pooling layer with a pooling window size of 2 to reduce the resolution of the time dimension;

[0043] 4) Establish CNN layer 2: Establish a two-level convolutional neural network to achieve high-level extraction of temporal features of verification data;

[0044] (1) Establish convolution layer 2: Establish a convolution layer with 64 convolution kernels and a convolution kernel size of 5 to further extract higher-level temporal features of the verification data;

[0045] (2) Establishing pooling layer 2: Establishing a pooling layer with a pooling window size of 2 to highlight the timing characteristics of key time segments;

[0046] 5) Establishing the LSTM layer: Establishing a secondary long-short-term memory network to capture the dynamic trend of time series changes;

[0047] (1) Establish LSTM layer 1: Establish a long short-term memory network with 128 units to capture long-term dependencies in time series;

[0048] (2) Establish LSTM layer 2: Establish a long short-term memory network with 64 units to capture the dynamic trend of time series;

[0049] 6) Establish a fully connected layer: Establish a fully connected layer to map the output of LSTM layer 2 to the thermal efficiency value and generate the thermal efficiency prediction result;

[0050] 7) Output layer performs output: the generated power system thermal efficiency prediction value is output in time series. The output data contains multiple time steps that are the same as the input layer, and each time step has one thermal efficiency prediction value.

[0051] Wherein, in said step S4, thermal efficiency evaluation accuracy is compared;

[0052] Compare the thermal efficiency prediction value evaluated by the neural network model with the actual thermal efficiency value calculated from the test data to optimize the thermal efficiency characterization parameter screening method;

[0053] 1) Calculation of the actual value of thermal efficiency: The thermal efficiency of the power system is calculated using the vehicle power system test data. The calculation formula is as follows:

[0054]

[0055] Where η r is the true value of thermal efficiency, T tq is the engine torque, n is the engine speed, is the amount of oil circulating in the engine, Hv is the lower heating value of the fuel;

[0056] 2) Calculation of prediction accuracy: Compare the predicted value of thermal efficiency with the actual value of thermal efficiency to calculate the prediction accuracy. The calculation formula is as follows:

[0057]

[0058] Where A is the prediction accuracy, η p is the predicted value of thermal efficiency.

[0059] Wherein, in the step S5, the preferred thermal efficiency characterization parameter set is output; the characterization parameters screened out by the thermal efficiency characterization parameter screening method with the highest prediction accuracy are output as the preferred thermal efficiency characterization parameter set.

[0060] (3) Beneficial effects

[0061] Compared with the existing technology, the present invention provides a method for optimizing the thermal efficiency characterization parameters of a vehicle power system. First, the thermal efficiency characterization parameters are screened through a variety of correlation coefficient feature extraction methods, and then a neural network model is used to perform thermal efficiency prediction analysis. Then, the deviation between the thermal efficiency prediction value and the true value is compared, thereby realizing the optimization of the thermal efficiency characterization parameters of the vehicle power system and providing a high-confidence characterization parameter set for the degradation state assessment of the vehicle power system.

[0062] The present invention demonstrates the following: This method for optimizing vehicle powertrain thermal efficiency parameters uses multiple correlation coefficient feature extraction methods to screen these parameters, combined with neural network model prediction and analysis. This method optimizes vehicle powertrain thermal efficiency parameters and provides a high-confidence parameter set for evaluating vehicle powertrain degradation. This method is applicable to evaluating the degradation of any fuel-powered vehicle powertrain. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart showing the optimization of thermal efficiency characterization parameters of a vehicle power system according to the present invention;

[0064] Figure 2 This is a flow chart of the thermal efficiency prediction method based on the neural network model of the present invention;

[0065] Figure 3-Figure 5 This is a data display diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, content, and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below with reference to the accompanying drawings and examples.

[0067] To solve the above technical problems, the present invention provides a method for optimizing parameters characterizing thermal efficiency of a vehicle power system, the method comprising the following steps:

[0068] Step S1: construct a power system dataset;

[0069] Step S2: screening of thermal efficiency characterization parameters;

[0070] Step S3: using a neural network model to evaluate thermal efficiency;

[0071] Step S4: thermal efficiency evaluation accuracy comparison;

[0072] Step S5: outputting the optimal thermal efficiency characterization parameter set.

[0073] Wherein, in said step S1, a power system data set is constructed;

[0074] The power system test data is input in time series. The input data contains multiple time steps. Each time step includes 15 heat flow characterization parameters, including circulating oil volume, rail pressure, oil temperature, speed, torque, coolant flow, coolant temperature, coolant specific heat, intake temperature, intake pressure, intake flow, exhaust temperature, exhaust pressure, exhaust flow, and ambient temperature.

[0075] Wherein, in said step S2, thermal efficiency characterization parameters are screened;

[0076] Three filtering characterization parameter extraction methods, namely Pearson correlation coefficient, Spearman correlation coefficient and grey relational degree, were established.

[0077] Wherein, in step S2, the Pearson correlation coefficient method is as follows:

[0078] The Pearson correlation coefficient is used to analyze the linear correlation between factors, and its definition is as follows:

[0079]

[0080] In the formula, r is the Pearson correlation coefficient, n is the number of samples, and are the average values ​​of the two samples, σ X and σ Y X i and Y i Standard score of

[0081] When the correlation coefficient |r| ≥ 0.2, the characterization parameters have a weak linear correlation with the thermal efficiency. Therefore, seven parameters, including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature, are selected as the characterization parameters for thermal efficiency evaluation using the Pearson method.

[0082] Wherein, in step S2, the Spearman correlation coefficient method is as follows: The Spearman correlation coefficient is used to analyze the nonlinear correlation between variables, and its definition is as follows:

[0083]

[0084] Where ρ is the Spearman correlation coefficient, n is the number of samples, and d i is the rank difference between the two samples;

[0085] The value range of ρ is [-1, 1]. The closer its absolute value is to 1, the stronger the correlation between variables. When |ρ| ≥ 0.5, the five parameters of torque, circulating oil volume, rail pressure, intake pressure, and exhaust pressure are selected as the characterization parameters of the Spearman method thermal efficiency evaluation.

[0086] Wherein, in step S2, the grey relational method is as follows:

[0087] Grey relational degree is used to analyze the development trend and change pattern of a system over time, that is, the characteristics and degree of dynamic correlation between factors. Its definition is as follows:

[0088]

[0089] In the formula, γ is the grey relational degree, n is the number of samples, and ξ(k) is the grey relational coefficient, as shown in the following formula;

[0090]

[0091] Where, Δ min is the minimum absolute value of the difference between the two sequence values, Δ max is the maximum absolute value of the difference between the two sequence values, ρ is the resolution coefficient, Δ t(k) is the absolute value of the difference between the two sequence values;

[0092] The larger the γ value, the stronger the correlation. When γ ≥ 0.6, six parameters including coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume are selected as the characterization parameters of thermal efficiency evaluation using the grey correlation method.

[0093] Wherein, in said step S3, thermal efficiency evaluation is performed using a neural network model;

[0094] With the vehicle power system test data as input and the thermal efficiency value as output, a neural network model is established. The thermal efficiency evaluation characterization parameters determined by the Pearson correlation coefficient, Spearman correlation coefficient and grey correlation degree are used to evaluate the thermal efficiency of the vehicle power system.

[0095] Wherein, the step S3 specifically includes:

[0096] 1) Input layer: Input the power system test data in time series. The input data contains multiple time steps. Each time step includes nine heat flow characterization parameters: circulating oil volume, rail pressure, speed, torque, coolant temperature, intake temperature, intake pressure, exhaust temperature, and exhaust pressure.

[0097] 2) The data screening layer performs data screening: data screening is performed based on the characterization parameters determined by the Pearson correlation coefficient method, the Spearman correlation coefficient method, and the grey relational method;

[0098] (1) Pearson method verification data screening: Data containing seven characteristic parameters including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature are screened from the power system test data to form the Pearson method verification data;

[0099] (2) Spearman method verification data screening: The data containing five characteristic parameters, namely torque, circulating oil volume, rail pressure, intake pressure and exhaust pressure, are screened from the power system test data to form the Spearman method verification data;

[0100] (3) Grey correlation method validation data screening: Data containing six characterization parameters, namely coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume, are screened from the power system test data to form the grey correlation method validation data;

[0101] 3) Establish CNN layer 1: Establish a first-level convolutional neural network to extract local temporal features of verification data;

[0102] (1) Establish convolution layer 1: Establish a convolution layer with 32 convolution kernels and a convolution kernel size of 3 to extract local temporal features from time series data;

[0103] (2) Establish pooling layer 1: Establish a pooling layer with a pooling window size of 2 to reduce the resolution of the time dimension;

[0104] 4) Establish CNN layer 2: Establish a two-level convolutional neural network to achieve high-level extraction of temporal features of verification data;

[0105] (1) Establish convolution layer 2: Establish a convolution layer with 64 convolution kernels and a convolution kernel size of 5 to further extract higher-level temporal features of the verification data;

[0106] (2) Establishing pooling layer 2: Establishing a pooling layer with a pooling window size of 2 to highlight the timing characteristics of key time segments;

[0107] 5) Establishing the LSTM layer: Establishing a secondary long-short-term memory network to capture the dynamic trend of time series changes;

[0108] (1) Establish LSTM layer 1: Establish a long short-term memory network with 128 units to capture long-term dependencies in time series;

[0109] (2) Establish LSTM layer 2: Establish a long short-term memory network with 64 units to capture the dynamic trend of time series;

[0110] 6) Establish a fully connected layer: Establish a fully connected layer to map the output of LSTM layer 2 to the thermal efficiency value and generate the thermal efficiency prediction result;

[0111] 7) Output layer performs output: the generated power system thermal efficiency prediction value is output in time series. The output data contains multiple time steps that are the same as the input layer, and each time step has one thermal efficiency prediction value.

[0112] Wherein, in said step S4, thermal efficiency evaluation accuracy is compared;

[0113] Compare the thermal efficiency prediction value evaluated by the neural network model with the actual thermal efficiency value calculated from the test data to optimize the thermal efficiency characterization parameter screening method;

[0114] 1) Calculation of the actual value of thermal efficiency: The thermal efficiency of the power system is calculated using the vehicle power system test data. The calculation formula is as follows:

[0115]

[0116] Where η r is the true value of thermal efficiency, T tq is the engine torque, n is the engine speed, is the amount of oil circulating in the engine, Hv is the lower heating value of the fuel;

[0117] 2) Calculation of prediction accuracy: Compare the predicted value of thermal efficiency with the actual value of thermal efficiency to calculate the prediction accuracy. The calculation formula is as follows:

[0118]

[0119] Where A is the prediction accuracy, η p is the predicted value of thermal efficiency.

[0120] Wherein, in the step S5, the preferred thermal efficiency characterization parameter set is output; the characterization parameters screened out by the thermal efficiency characterization parameter screening method with the highest prediction accuracy are output as the preferred thermal efficiency characterization parameter set.

[0121] Example 1

[0122] In order to better understand the present invention, the following is a detailed description of the present invention. Figure 1 Vehicle power system thermal efficiency characterization parameter optimization process, appendix Figure 2 The present invention is described in detail based on the thermal efficiency prediction method of the neural network model.

[0123] (1) Construct a power system data set; input the power system test data in time series. The input data contains multiple time steps. Each time step has 15 heat flow characterization parameters, including circulating oil volume, rail pressure, oil temperature, speed, torque, coolant flow, coolant temperature, coolant specific heat, intake temperature, intake pressure, intake flow, exhaust temperature, exhaust pressure, exhaust flow, and ambient temperature.

[0124] (2) Screening of thermal efficiency characterization parameters; establishing three filtering characterization parameter extraction methods: Pearson correlation coefficient, Spearman correlation coefficient, and grey relational degree;

[0125] (201) Pearson correlation coefficient method: The Pearson correlation coefficient method is used to screen the thermal efficiency characterization parameters. The definition of the Pearson correlation coefficient is as follows:

[0126]

[0127] In the formula, r is the Pearson correlation coefficient, n is the number of samples, and are the average values ​​of the two samples, σ X and σ Y X i and Y i The standard score of .

[0128] The correlation coefficient |r|≥0.2 was taken, and seven parameters including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature were selected as the characterization parameters for Pearson thermal efficiency evaluation. Figure 3 shown.

[0129] (202) Spearman correlation coefficient method: The Spearman correlation coefficient method is used to screen the thermal efficiency characterization parameters. The definition of the Spearman correlation coefficient is as follows:

[0130]

[0131] Where ρ is the Spearman correlation coefficient, n is the number of samples, and d i is the rank difference between the two samples.

[0132] Taking |ρ|≥0.5, we select five parameters, namely torque, circulating oil volume, rail pressure, intake pressure, and exhaust pressure, as the Spearman thermal efficiency evaluation parameters. Figure 4 shown.

[0133] (203) Grey correlation method: The grey correlation method is used to screen the thermal efficiency characterization parameters. The grey correlation definition is as follows:

[0134]

[0135] In the formula, γ is the grey relational degree, n is the number of samples, and ξ(k) is the grey relational coefficient, as shown in the following formula.

[0136]

[0137] Where, Δ min is the minimum absolute value of the difference between the two sequence values, Δ max is the maximum absolute value of the difference between the two sequence values, ρ is the resolution coefficient, Δ t(k) It is the absolute value of the difference between the two series values.

[0138] When γ≥0.6, six parameters including coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume are selected as the thermal efficiency evaluation parameters of the grey correlation method. Figure 5 shown.

[0139] (3) Use a neural network model to evaluate thermal efficiency; use vehicle power system test data as input and thermal efficiency value as output to build a neural network model, and use the thermal efficiency evaluation characterization parameters determined by the Pearson correlation coefficient, Spearman correlation coefficient, and grey correlation degree to evaluate the thermal efficiency of the vehicle power system;

[0140] (301) Input layer: The power system test data is input in time series. The input data contains multiple time steps. Each time step has nine heat flow characterization parameters, including circulating oil volume, rail pressure, speed, torque, coolant temperature, intake temperature, intake pressure, exhaust temperature, and exhaust pressure.

[0141] (302) Data screening layer: Data screening is performed based on the characterization parameters determined by the Pearson correlation coefficient method, the Spearman correlation coefficient method, and the grey relational method;

[0142] (3021) Pearson method verification data screening: Filter out data containing seven characterization parameters such as torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature from the power system test data to form the Pearson method verification data;

[0143] (3022) Spearman method validation data screening: Filter out data containing five characterization parameters such as torque, circulating oil volume, rail pressure, intake pressure, and exhaust pressure from the power system test data to form the Spearman method validation data;

[0144] (3023) Grey correlation method validation data screening: Filter out data containing six characterization parameters such as coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume from the power system test data to form the grey correlation method validation data;

[0145] (303) Establish CNN layer 1: Establish a first-level convolutional neural network layer to extract local temporal features of verification data;

[0146] (3031) Establish convolution layer 1: Establish convolution layer 1 with 32 convolution kernels and 3 convolution kernel size to extract local temporal features from time series data;

[0147] (3032) Establish pooling layer 1: Establish pooling layer 1 with a pooling window size of 2 to reduce the resolution of the time dimension;

[0148] (304) Establish CNN layer 2: Establish a secondary convolutional neural network layer to achieve high-level extraction of temporal features of verification data;

[0149] (3041) Establish convolution layer 2: Establish convolution layer 2 with 64 convolution kernels and 5 convolution kernel size to further extract higher-level temporal features of the verification data;

[0150] (3042) Establish pooling layer 2: Establish pooling layer 2 with a pooling window size of 2 to highlight the timing characteristics of key time segments;

[0151] (305) Establish LSTM layer: Establish a secondary long short-term memory network to capture the dynamic trend of time series;

[0152] (3051) Establish LSTM layer 1: Establish a long short-term memory network layer with 128 units to capture long-term dependencies in time series;

[0153] (3052) Establish LSTM layer 2: Establish a long short-term memory network layer with 64 units to capture the dynamic trend of time series;

[0154] (306) Fully connected layer: Establish a fully connected layer to map the output of LSTM layer 2 to the thermal efficiency value and generate the predicted result of thermal efficiency;

[0155] (307) Output layer: Output the generated power system thermal efficiency prediction value in time series. The output data contains the same multiple time steps as the input layer, and each time step has one thermal efficiency prediction value.

[0156] (4) Comparison of thermal efficiency evaluation accuracy: Compare the thermal efficiency prediction value evaluated by the neural network model with the actual thermal efficiency value calculated from the test data, and optimize the thermal efficiency characterization parameter screening method;

[0157] (401) Calculation of actual value of thermal efficiency: The thermal efficiency of the power system is calculated using the vehicle power system test data. The calculation formula is as follows:

[0158]

[0159] Where η r is the true value of thermal efficiency, T tq is the engine torque, n is the engine speed, is the amount of oil circulating in the engine, and Hv is the lower heating value of the fuel.

[0160] (402) Prediction accuracy calculation: The predicted value of thermal efficiency is compared with the actual value of thermal efficiency to calculate the prediction accuracy. The calculation formula is as follows:

[0161]

[0162] Where A is the prediction accuracy, η p is the predicted value of thermal efficiency.

[0163] Comparison of prediction accuracy of different characterization parameter screening methods

[0164]

[0165] (5) Outputting the optimal thermal efficiency characterization parameter set; outputting the six characterization parameters, namely, coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume, which are selected by the grey correlation thermal efficiency characterization parameter screening method with the highest prediction accuracy, as the optimal thermal efficiency characterization parameter set.

[0166] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for optimizing parameters characterizing thermal efficiency of a vehicle power system, characterized in that: The method comprises the following steps: Step S1: construct a power system dataset; Step S2: screening of thermal efficiency characterization parameters; Step S3: using a neural network model to evaluate thermal efficiency; Step S4: thermal efficiency evaluation accuracy comparison; Step S5: outputting the optimal thermal efficiency characterization parameter set.

2. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 1, characterized in that: In the step S1, a power system data set is constructed; The power system test data is input in time series. The input data contains multiple time steps. Each time step includes 15 heat flow characterization parameters, including circulating oil volume, rail pressure, oil temperature, speed, torque, coolant flow, coolant temperature, coolant specific heat, intake temperature, intake pressure, intake flow, exhaust temperature, exhaust pressure, exhaust flow, and ambient temperature.

3. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 2, characterized in that: In the step S2, thermal efficiency characterization parameters are screened; Three filtering characterization parameter extraction methods, namely Pearson correlation coefficient, Spearman correlation coefficient and grey relational degree, were established.

4. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 3, characterized in that: In step S2, the Pearson correlation coefficient method is as follows: The Pearson correlation coefficient is used to analyze the linear correlation between factors, and its definition is as follows: In the formula, r is the Pearson correlation coefficient, n is the number of samples, and are the average values ​​of the two samples, σ X and σ Y X i and Y i Standard score of When the correlation coefficient |r| ≥ 0.2, the characterization parameters have a weak linear correlation with the thermal efficiency. Therefore, seven parameters, including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature, are selected as the characterization parameters for thermal efficiency evaluation using the Pearson method.

5. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 4, characterized in that: In step S2, the Spearman correlation coefficient method is as follows: The Spearman correlation coefficient is used to analyze the nonlinear correlation between variables, and its definition is as follows: Where ρ is the Spearman correlation coefficient, n is the number of samples, and d is the i is the rank difference between the two samples; The value range of ρ is [-1, 1]. The closer its absolute value is to 1, the stronger the correlation between variables. When |ρ| ≥ 0.5, the five parameters of torque, circulating oil volume, rail pressure, intake pressure, and exhaust pressure are selected as the characterization parameters of the Spearman method thermal efficiency evaluation.

6. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 5, characterized in that: In step S2, the grey relational method is as follows: Grey relational degree is used to analyze the development trend and change pattern of a system over time, that is, the characteristics and degree of dynamic correlation between factors. Its definition is as follows: In the formula, γ is the grey relational degree, n is the number of samples, and ξ(k) is the grey relational coefficient, as shown in the following formula; Where, Δ min is the minimum absolute value of the difference between the two sequence values, Δ max is the maximum absolute value of the difference between the two sequence values, ρ is the resolution coefficient, Δ t(k) is the absolute value of the difference between the two series; The larger the γ value, the stronger the correlation. When γ ≥ 0.6, six parameters including coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume are selected as the characterization parameters of thermal efficiency evaluation using the grey correlation method.

7. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 6, characterized in that: In step S3, thermal efficiency evaluation is performed using a neural network model; With the vehicle power system test data as input and the thermal efficiency value as output, a neural network model is established. The thermal efficiency evaluation characterization parameters determined by the Pearson correlation coefficient, Spearman correlation coefficient and grey correlation degree are used to evaluate the thermal efficiency of the vehicle power system.

8. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 7, characterized in that: The step S3 specifically includes: 1) Input layer: Input the power system test data in time series. The input data contains multiple time steps. Each time step includes nine heat flow characterization parameters: circulating oil volume, rail pressure, speed, torque, coolant temperature, intake temperature, intake pressure, exhaust temperature, and exhaust pressure. 2) The data screening layer performs data screening: data screening is performed based on the characterization parameters determined by the Pearson correlation coefficient method, the Spearman correlation coefficient method, and the grey relational method; (1) Pearson method verification data screening: Data containing seven characteristic parameters including torque, circulating oil volume, intake temperature, exhaust temperature, speed, rail pressure, and coolant temperature are screened from the power system test data to form the Pearson method verification data; (2) Spearman method verification data screening: The data containing five characteristic parameters, namely torque, circulating oil volume, rail pressure, intake pressure and exhaust pressure, are screened from the power system test data to form the Spearman method verification data; (3) Grey correlation method validation data screening: Data containing six characterization parameters, namely coolant temperature, intake temperature, exhaust temperature, intake pressure, exhaust pressure, and circulating oil volume, are screened from the power system test data to form the grey correlation method validation data; 3) Establish CNN layer 1: Establish a first-level convolutional neural network to extract local temporal features of verification data; (1) Establish convolution layer 1: Establish a convolution layer with 32 convolution kernels and a convolution kernel size of 3 to extract local temporal features from time series data; (2) Establish pooling layer 1: Establish a pooling layer with a pooling window size of 2 to reduce the resolution of the time dimension; 4) Establish CNN layer 2: Establish a two-level convolutional neural network to achieve high-level extraction of temporal features of verification data; (1) Establish convolution layer 2: Establish a convolution layer with 64 convolution kernels and a convolution kernel size of 5 to further extract higher-level temporal features of the verification data; (2) Establishing pooling layer 2: Establishing a pooling layer with a pooling window size of 2 to highlight the timing characteristics of key time segments; 5) Establishing the LSTM layer: Establishing a secondary long short-term memory network to capture the dynamic trend of time series changes; (1) Establish LSTM layer 1: Establish a long short-term memory network with 128 units to capture long-term dependencies in time series; (2) Establish LSTM layer 2: Establish a long short-term memory network with 64 units to capture the dynamic trend of time series; 6) Establish a fully connected layer: Establish a fully connected layer to map the output of LSTM layer 2 to the thermal efficiency value and generate the thermal efficiency prediction result; 7) Output layer performs output: the generated power system thermal efficiency prediction value is output in time series. The output data contains multiple time steps that are the same as the input layer, and each time step has one thermal efficiency prediction value.

9. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 8, characterized in that: In step S4, thermal efficiency evaluation accuracy is compared; Compare the thermal efficiency prediction value evaluated by the neural network model with the actual thermal efficiency value calculated from the test data to optimize the thermal efficiency characterization parameter screening method; 1) Calculation of the actual value of thermal efficiency: The thermal efficiency of the power system is calculated using the vehicle power system test data. The calculation formula is as follows: Where η r is the true value of thermal efficiency, T tq is the engine torque, n is the engine speed, is the amount of oil circulating in the engine, Hv is the lower heating value of the fuel; 2) Calculation of prediction accuracy: Compare the predicted value of thermal efficiency with the actual value of thermal efficiency to calculate the prediction accuracy. The calculation formula is as follows: Where A is the prediction accuracy, η p is the predicted value of thermal efficiency.

10. The method for optimizing vehicle power system thermal efficiency characterization parameters according to claim 9, characterized in that: In the step S5, the preferred thermal efficiency characterization parameter set is output; the characterization parameters screened out by the thermal efficiency characterization parameter screening method with the highest prediction accuracy are output as the preferred thermal efficiency characterization parameter set.

Images