Method for optimizing intake and exhaust phases of small rotary engine
By constructing a multidimensional mapping relationship and active subspace optimization using the Gaussian process regression method, the problem of difficult parameter combination identification in the intake and exhaust phase optimization of rotary engines is solved, thereby improving optimization efficiency and effectiveness and reducing computational costs.
Patent Information
- Application Number
- CN202510661181.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-22
- Publication Date
- 2025-10-28
Smart Images
Figure CN120850470A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power engineering application technology, specifically relating to a method for optimizing the intake and exhaust phases of a small rotary engine. Background Technology
[0002] The rotary engine is a compact, low-vibration internal combustion engine widely used in small engines and aerospace applications. Compared to traditional reciprocating piston engines, the rotary engine's operation is relatively simple, primarily relying on the rotation of the rotor within the cylinder to complete the intake, compression, combustion, and exhaust processes. However, the in-cylinder flow field of a rotary engine is more complex, and the adjustment of the intake and exhaust phases and durations significantly impacts engine performance. Unlike reciprocating engines that regulate intake and exhaust through valve timing, the intake and exhaust phases of a rotary engine are determined by the position and size of the intake and exhaust ports within the cylinder, making optimization a challenging process.
[0003] Currently, optimization methods for intake and exhaust phase typically involve discretizing multiple design parameters and comparing their performance to find the optimal solution. However, this method is prone to missing the best combination of parameters and often requires extensive simulations or experimental verification, leading to high time and cost. Therefore, identifying the optimal parameter combination in the calculation, reducing the amount of experiments and simulations, and improving optimization efficiency have become key issues in improving the performance of rotary engines. Summary of the Invention
[0004] To address the aforementioned technical problems, this invention provides an optimization scheme for the intake and exhaust phases of a small rotary engine. This scheme seeks the optimal optimization direction under specified factors and horizontal ranges, and has the characteristic of being closer to the optimal parameter combination.
[0005] The first aspect of this invention proposes a method for optimizing the intake and exhaust phases of a small rotary engine, the method comprising:
[0006] Step S1: Determine the factors related to the intake and exhaust phases of the small rotary engine;
[0007] The factors mentioned include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter.
[0008] Step S2: Determine the value range of each factor related to the intake and exhaust phases of the small rotary engine;
[0009] Step S3: Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index;
[0010] Step S4: Optimize based on the multidimensional change relationship in the multidimensional mapping relationship;
[0011] Among them, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by fitting a regression model, and the synergistic change characteristics between input factors are obtained, thereby identifying key factors and extracting the basis of the active subspace.
[0012] Step S5: Determine the dimension k of the active subspace;
[0013] Step S6: Project data into the active subspace;
[0014] Step S7: Optimize the combination of intake and exhaust phase parameters.
[0015] In step S2: the intake start phase is later than TDC; the exhaust start phase is later than the phase corresponding to the completion of combustion; the intake duration angle and exhaust duration angle are constrained by the equivalent diameter of the intake port and the equivalent diameter of the exhaust port; after the duration angle is determined, the port width is determined under the current equivalent diameter, and the port width constrains the thickness between the port edge and the cylinder end face, and the port width is not greater than the cylinder width.
[0016] In step S3: The value range of each factor is discretized, several value points with equal gradients are selected, simulation calculations are performed, the target parameter index is obtained, and the variable x of each discretized factor is established. n The function f(x) with the target parameter index n The multidimensional mapping relationship between ).
[0017] Step S4 specifically includes:
[0018] Step S4.1: Use the Gaussian process regression (GPR) method to analyze f(x). n The model is fitted using GPR nonlinear modeling to capture the relationship between input variables and output parameters;
[0019] Step S4.2: Using the fitted GPR model, calculate x for each sample point. n ( i gradient vector at ) Gradient information characterizes the sensitivity of each factor to the target parameter index and is used to construct the covariance matrix;
[0020] Step S4.3: Construct the covariance matrix C based on the gradient vector to capture the cooperative variation characteristics among various input factors.
[0021]
[0022] Where m is the sample size. Let be the gradient vector of the i-th sample;
[0023] Step S4.4: Perform eigenvalue decomposition on the covariance matrix C to obtain eigenvalues and eigenvectors. Select the first k eigenvectors according to the magnitude of the eigenvalues to form the basis of the active subspace.
[0024] In step S5: Based on the cumulative contribution rate of the eigenvalues, determine the optimal dimension k of the active subspace, and select the eigenvectors corresponding to the first k eigenvalues with a cumulative contribution rate of 95% or higher as the basis of the active subspace.
[0025] In step S6: x n The data is projected into a low-dimensional active subspace to reduce its dimensionality. Each sample point is mapped to the new low-dimensional space through matrix multiplication to simplify the optimization process.
[0026] In step S7: the optimal direction is identified based on the active subspace method, and the intake and exhaust phase parameters are optimized in the dimensionality-reduced space. The gradient ascent method is used to determine the parameter combination that maximizes the target parameter index.
[0027] The method further includes:
[0028] Step S8: Verify the optimization results;
[0029] The optimized parameter combination is simulated and verified to confirm its feasibility and performance in an actual engine. The impact of the optimized parameter combination on the target parameter index is evaluated through the simulation verification results to determine whether it has achieved the expected optimization effect.
[0030] Step S9: Implement the optimal parameter combination;
[0031] The validated optimal combination of intake and exhaust phase parameters was applied to a real small rotary engine for actual testing and performance evaluation. Based on the test results, the parameters were further fine-tuned to ensure the optimized effect of the engine in actual operation.
[0032] A second aspect of the present invention provides a small rotary engine intake and exhaust phase optimization system, the system comprising a processing unit configured to perform:
[0033] Factors related to the intake and exhaust phases of a small rotary engine are determined; wherein the factors include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter;
[0034] Determine the value range of various factors related to the intake and exhaust phases of a small rotary engine;
[0035] Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index;
[0036] Optimization is based on the multidimensional change relationship in the multidimensional mapping relationship; among which, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by using a fitted regression model, the synergistic change characteristics between input factors are obtained, and key factors are identified and the basis of the active subspace is extracted.
[0037] Determine the dimension k of the active subspace;
[0038] Project data into the active subspace;
[0039] Optimize the combination of intake and exhaust phase parameters.
[0040] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a method for optimizing the intake and exhaust phases of a small rotary engine according to the first aspect of this disclosure.
[0041] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for optimizing the intake and exhaust phases of a small rotary engine according to the first aspect of this disclosure.
[0042] In existing technologies, for optimization methods involving multiple parameters affecting a single objective variable, the complexity of the optimization problem increases exponentially with the increase in the dimensionality of the input variables, leading to a sharp increase in computational resources and time costs. In high-dimensional spaces, the nonlinear interactions and synergistic effects between variables are complex, and traditional optimization methods struggle to effectively identify and utilize these relationships. More importantly, in multidimensional variable optimization environments, many traditional optimization methods rely on a large amount of data to build accurate models. For rotary engines, simulation computation is substantial and time-consuming. The optimization method for the intake and exhaust phases of a small rotary engine disclosed in this invention can significantly reduce optimization computation and time costs.
[0043] This invention discloses an optimization method for the intake and exhaust phase of a small rotary engine. By analyzing the gradient information between input and output variables, it identifies the direction with the greatest impact on the output. This allows the optimization process to focus on critical directions, improving efficiency and effectiveness while reducing unnecessary searches in secondary directions. By identifying the direction with the greatest impact on the output variable, this method helps to better identify which combinations of input variables are more critical to the output, thus aiding in parameter setting optimization. Simultaneously, by focusing on the main influencing directions, variables with less impact on the output are filtered out, improving the model's adaptability. Attached Figure Description
[0044] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0045] Figure 1 This is a flowchart illustrating a method for optimizing the intake and exhaust phases of a small rotary engine according to an embodiment of the present invention.
[0046] Figure 2 This is a schematic diagram of the starting angle and the duration angle according to an embodiment of the present invention.
[0047] Figure 3 This is a schematic diagram of the air inlet dimensions according to an embodiment of the present invention. Detailed Implementation
[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0049] The first aspect of this invention proposes a method for optimizing the intake and exhaust phases of a small rotary engine, the method comprising (e.g.) Figure 1 As shown):
[0050] Step S1: Determine the factors related to the intake and exhaust phases of the small rotary engine;
[0051] The factors mentioned include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter.
[0052] Step S2: Determine the value range of each factor related to the intake and exhaust phases of the small rotary engine;
[0053] Step S3: Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index;
[0054] Step S4: Optimize based on the multidimensional change relationship in the multidimensional mapping relationship;
[0055] Among them, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by fitting a regression model, and the synergistic change characteristics between input factors are obtained, thereby identifying key factors and extracting the basis of the active subspace.
[0056] Step S5: Determine the dimension k of the active subspace;
[0057] Step S6: Project data into the active subspace;
[0058] Step S7: Optimize the combination of intake and exhaust phase parameters.
[0059] In step S2: the intake start phase is later than TDC; the exhaust start phase is later than the phase corresponding to the completion of combustion; the intake duration angle and exhaust duration angle are constrained by the equivalent diameter of the intake port and the equivalent diameter of the exhaust port; after the duration angle is determined, the port width is determined under the current equivalent diameter, and the port width constrains the thickness between the port edge and the cylinder end face, and the port width is not greater than the cylinder width.
[0060] In step S3: The value range of each factor is discretized, several value points with equal gradients are selected, simulation calculations are performed, the target parameter index is obtained, and the variable x of each discretized factor is established. n The function f(x) with the target parameter index n The multidimensional mapping relationship between ).
[0061] Step S4 specifically includes:
[0062] Step S4.1: Use the Gaussian process regression (GPR) method to analyze f(x). n The model is fitted using GPR nonlinear modeling to capture the relationship between input variables and output parameters;
[0063] Step S4.2: Using the fitted GPR model, calculate x for each sample point. n (i) gradient vector at point Gradient information characterizes the sensitivity of each factor to the target parameter index and is used to construct the covariance matrix;
[0064] Step S4.3: Construct the covariance matrix C based on the gradient vector to capture the cooperative variation characteristics among various input factors.
[0065]
[0066] Where m is the sample size. Let be the gradient vector of the i-th sample;
[0067] Step S4.4: Perform eigenvalue decomposition on the covariance matrix C to obtain eigenvalues and eigenvectors. Select the first k eigenvectors according to the magnitude of the eigenvalues to form the basis of the active subspace.
[0068] In step S5: Based on the cumulative contribution rate of the eigenvalues, determine the optimal dimension k of the active subspace, and select the eigenvectors corresponding to the first k eigenvalues with a cumulative contribution rate of 95% or higher as the basis of the active subspace.
[0069] In step S6: x n The data is projected into a low-dimensional active subspace to reduce its dimensionality. Each sample point is mapped to the new low-dimensional space through matrix multiplication to simplify the optimization process.
[0070] In step S7: the optimal direction is identified based on the active subspace method, and the intake and exhaust phase parameters are optimized in the dimensionality-reduced space. The gradient ascent method is used to determine the parameter combination that maximizes the target parameter index.
[0071] The method further includes:
[0072] Step S8: Verify the optimization results;
[0073] The optimized parameter combination is simulated and verified to confirm its feasibility and performance in an actual engine. The impact of the optimized parameter combination on the target parameter index is evaluated through the simulation verification results to determine whether it has achieved the expected optimization effect.
[0074] Step S9: Implement the optimal parameter combination;
[0075] The validated optimal combination of intake and exhaust phase parameters was applied to a real small rotary engine for actual testing and performance evaluation. Based on the test results, the parameters were further fine-tuned to ensure the optimized effect of the engine in actual operation.
[0076] Examples (e.g.) Figure 2-3 (As shown)
[0077] Step 1: Determine the factors related to the intake and exhaust phases.
[0078] The following six factors must be clearly defined: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter. For example, based on engine design requirements and operating principles, the intake initiation angle must be determined within a specific crankshaft angle range. This range is set based on the engine's combustion cycle and intake requirements, ensuring that intake begins at the appropriate time to avoid affecting combustion efficiency due to premature or delayed intake. The intake duration angle is related to engine speed, load, and other factors, and its magnitude affects the intake volume and the quality of the air-fuel mixture. The intake port equivalent diameter must consider the engine's intake flow requirements and cylinder structure; too small a diameter will limit the intake volume, while too large a diameter may affect cylinder strength and combustion stability. The exhaust initiation angle must be later than the phase corresponding to the completion of combustion to ensure that the exhaust gases are discharged in time and do not affect the next cycle. The exhaust duration angle and exhaust port equivalent diameter are interrelated; the exhaust port equivalent diameter is determined by engine displacement and exhaust efficiency requirements, and both together determine the smoothness of exhaust and the pressure changes within the cylinder.
[0079] The intake and exhaust angles are determined based on the eccentric shaft rotation angle corresponding to the position crossed by the rotor tip. The intake and exhaust port shapes are rectangular, and the equivalent diameter of the intake and exhaust port is calculated based on the dimensions of the rectangle.
[0080] Step 2: Determine the value range of each factor.
[0081] The intake initiation phase should be strictly later than the time-to-discharge (TDC) phase. This is because before the TDC, the cylinder pressure and temperature conditions are unfavorable for intake. For example, if the intake initiation angle is too early, it will cause the intake air to mix with residual exhaust gas, affecting combustion efficiency. The exhaust initiation phase should be later than the phase corresponding to the completion of combustion, ensuring that exhaust is only performed after complete combustion and avoiding the waste of useful work by expelling unburned mixture. The intake and exhaust duration angles are intrinsically related to the equivalent diameters of the intake and exhaust ports. Once the duration angle is determined, the port width is determined for a given equivalent diameter. The intake port is rectangular, and the relationship between the equivalent diameter (d) and the length (a) and width (b) of the intake port is as follows: However, the port width must not be too wide. If the thickness between the port edge and the cylinder end face is too thin or the port width is greater than the cylinder width, it will lead to a decrease in the strength of the cylinder structure and may cause the intake or exhaust process to lose control, affecting the normal operation of the engine.
[0082] Step 3: Obtain the multidimensional mapping relationship between the discretized factor variables and the target parameter index.
[0083] The value ranges of each factor are discretized. Similar processing is applied to the intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter. Then, simulation calculations are performed, combining these discrete value points and using engine simulation software (such as AVL BOOST) to simulate engine operation and obtain power values under different combinations. A multidimensional mapping relationship between variable factors and target parameters is established, where represents different parameter combinations and represents the corresponding power values.
[0084] Step 4: Apply the active subspace method to optimize based on multidimensional change relationships.
[0085] Step 4.1: Fit the function using Gaussian process regression.
[0086] Gaussian process regression (GPR) was chosen as a surrogate model to fit the function. With a small sample size, GPR effectively captures the complex nonlinear relationship between input variables (such as six factors including the intake initiation angle) and output parameters (power). R0 was calculated based on the fitting results. 2 This is used to determine whether the fitting result accurately reflects the relationship between the factor variables and the target parameter.
[0087] Step 4.2: Calculate the gradient vector
[0088] Using the fitted GPR model, the gradient vector at each sample point is calculated. For example, for the intake initiation angle, its power gradient at different values is calculated; the same applies to other factors. This gradient information reflects the sensitivity of each factor to power changes and is a key basis for subsequently constructing the covariance matrix.
[0089] Step 4.3: Construct the covariance matrix
[0090] The covariance matrix is constructed based on the calculated gradient vectors. Taking the number of samples as an example (the actual number is determined by the number of combinations of discretized value points), the calculation is performed according to the formula. For example, for the two factors of intake start angle and intake duration angle, the gradient vectors at different sample points are multiplied and summed. Combined with similar calculations for other factors, the covariance matrix is obtained. This matrix reflects the cooperative variation characteristics among the input factors.
[0091] Step 4.4: Determine the active subspace basis
[0092] The covariance matrix is decomposed into eigenvalues (assuming six factors correspond to six eigenvalues λ1, λ2, λ3, λ4, λ5, λ6) and eigenvectors (v1, v2, v3, v4, v5, v6). The eigenvalues are sorted by magnitude, and the first few eigenvectors are selected (e.g., the first two eigenvectors correspond to a cumulative contribution rate of over 95%). These two eigenvectors form the basis of the active subspace.
[0093] Step 5: Determine the dimension of the active subspace.
[0094] The optimal dimension of the active subspace is determined based on the cumulative contribution rate of the eigenvalues. The proportion of each eigenvalue to the sum of all eigenvalues is calculated, and these proportions are accumulated from largest to smallest. It is assumed that when the accumulated proportion reaches 95% or more, the corresponding number of eigenvalues is 2. Therefore, v1 and v2 are chosen as the basis of the active subspace to ensure that most of the information affecting the output power is retained after dimensionality reduction.
[0095] Step 6: Project the data into the active subspace.
[0096] Projecting the original high-dimensional input data into a low-dimensional (6-dimensional) active subspace. Assume the original data is a six-dimensional vector x. n (Corresponding to six factors), the matrix multiplication operation z = Vx is performed with the basis vectors of the active subspace (such as [v1v2]). n This is mapped to a two-dimensional active subspace.
[0097] Step 7: Optimize the combination of intake and exhaust phase parameters.
[0098] Based on the optimal direction identified by the active subspace method (determined by the basis vectors of the active subspace), the intake and exhaust phase parameters are optimized in the dimensionality-reduced two-dimensional space. Different x values can be obtained from simulation data. n (i) (i = number of simulation calculation groups). Each x n (i) It can correspond to a P(z1) (u) z2 (u) ), that is, power pair (z1) (u) z2 (i) The mapping relationship is established. Within the two-dimensional subspace, optimization algorithms are used to find the optimal solution.
[0099] Step 8: Verify the optimization results.
[0100] For the optimized parameter combination z opt Simulation verification was performed. This was done by inverse projection (i.e., projecting it back from the active subspace to the original high-dimensional space x). opt =Vz opt TThe specific values of six parameters, including the intake initiation angle and intake duration angle, are obtained. These parameter values are then input into the engine simulation model to simulate actual engine operation and obtain the power value. The power value is compared with the value before optimization. If the power is greater than the power before optimization and meets other engine performance requirements (such as emissions and stability), it indicates that the optimization result is feasible and has good performance, achieving the expected optimization effect. Otherwise, further analysis is needed to determine the cause, which may be due to limitations in the step size selection, initial value setting, or the model itself. The optimization parameters or methods need to be readjusted.
[0101] Step 9: Implement the optimal parameter combination.
[0102] The validated optimal intake and exhaust phase parameter combination was applied to a real small rotary engine. In actual testing, the corresponding intake and exhaust devices were installed, and parameters such as the phase and size of the intake and exhaust ports were adjusted to the optimized values. The engine was then started for operational testing. Performance indicators such as actual power output, torque, and fuel consumption rate were measured using equipment such as a dynamometer and compared with the data before optimization. This ensured that the engine achieved optimal optimization results in actual operation, achieving stable performance improvement and meeting practical application requirements.
[0103] A second aspect of the present invention provides a small rotary engine intake and exhaust phase optimization system, the system comprising a processing unit configured to perform:
[0104] Factors related to the intake and exhaust phases of a small rotary engine are determined; wherein the factors include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter;
[0105] Determine the value range of each factor related to the intake and exhaust phases of a small rotary engine;
[0106] Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index;
[0107] Optimization is based on the multidimensional change relationship in the multidimensional mapping relationship; among which, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by using a fitted regression model, the synergistic change characteristics between input factors are obtained, and key factors are identified and the basis of the active subspace is extracted.
[0108] Determine the dimension k of the active subspace;
[0109] Project data into the active subspace;
[0110] Optimize the combination of intake and exhaust phase parameters.
[0111] A third aspect of this invention discloses an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a method for optimizing the intake and exhaust phases of a small rotary engine according to the first aspect of this disclosure.
[0112] A fourth aspect of this invention discloses a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for optimizing the intake and exhaust phases of a small rotary engine according to the first aspect of this disclosure.
[0113] In existing technologies, for optimization methods involving multiple parameters affecting a single objective variable, the complexity of the optimization problem increases exponentially with the increase in the dimensionality of the input variables, leading to a sharp increase in computational resources and time costs. In high-dimensional spaces, the nonlinear interactions and synergistic effects between variables are complex, and traditional optimization methods struggle to effectively identify and utilize these relationships. More importantly, in multidimensional variable optimization environments, many traditional optimization methods rely on a large amount of data to build accurate models. For rotary engines, simulation computation is substantial and time-consuming. The optimization method for the intake and exhaust phases of a small rotary engine disclosed in this invention can significantly reduce optimization computation and time costs.
[0114] This invention discloses an optimization method for the intake and exhaust phase of a small rotary engine. By analyzing the gradient information between input and output variables, it identifies the direction with the greatest impact on the output. This allows the optimization process to focus on critical directions, improving efficiency and effectiveness while reducing unnecessary searches in secondary directions. By identifying the direction with the greatest impact on the output variable, this method helps to better identify which combinations of input variables are more critical to the output, thus aiding in parameter setting optimization. Simultaneously, by focusing on the main influencing directions, variables with less impact on the output are filtered out, improving the model's adaptability.
[0115] Please note that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification. The above embodiments only illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.
Claims
1. A method for optimizing the intake and exhaust phases of a small rotary engine, characterized in that, The method comprises: Step S1: Determine the factors related to the intake and exhaust phases of the small rotary engine; The factors mentioned include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter. Step S2: Determine the value range of each factor related to the intake and exhaust phases of the small rotary engine; Step S3: Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index; Step S4: Optimize based on the multidimensional change relationship in the multidimensional mapping relationship; Among them, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by fitting a regression model, and the synergistic change characteristics between input factors are obtained, thereby identifying key factors and extracting the basis of the active subspace. Step S5: Determine the dimension k of the active subspace; Step S6: Project data into the active subspace; Step S7: Optimize the combination of intake and exhaust phase parameters.
2. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 1, characterized in that, In step S2: the intake start phase is later than TDC; the exhaust start phase is later than the phase corresponding to the completion of combustion; the intake duration angle and exhaust duration angle are constrained by the equivalent diameter of the intake port and the equivalent diameter of the exhaust port; after the duration angle is determined, the port width is determined under the current equivalent diameter, and the port width constrains the thickness between the port edge and the cylinder end face, and the port width is not greater than the cylinder width.
3. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 2, characterized in that, In step S3: The value range of each factor is discretized, several value points with equal gradients are selected, simulation calculations are performed, the target parameter index is obtained, and the variable x of each discretized factor is established. n The function f(x) with the target parameter index n The multidimensional mapping relationship between ).
4. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 3, characterized in that, Step S4 specifically includes: Step S4.1: Use the Gaussian process regression (GPR) method to analyze f(x). n The model is fitted using GPR nonlinear modeling to capture the relationship between input variables and output parameters; Step S4.2: Using the fitted GPR model, calculate x for each sample point. n (i) gradient vector at point Gradient information characterizes the sensitivity of each factor to the target parameter index and is used to construct the covariance matrix; Step S4.3: Construct the covariance matrix C based on the gradient vector to capture the cooperative variation characteristics among various input factors. Where m is the sample size. Let be the gradient vector of the i-th sample; Step S4.4: Perform eigenvalue decomposition on the covariance matrix C to obtain eigenvalues and eigenvectors. Select the first k eigenvectors according to the magnitude of the eigenvalues to form the basis of the active subspace.
5. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 4, characterized in that, In step S5: Based on the cumulative contribution rate of the eigenvalues, determine the optimal dimension k of the active subspace, and select the eigenvectors corresponding to the first k eigenvalues with a cumulative contribution rate of 95% or higher as the basis of the active subspace.
6. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 5, characterized in that, In step S6: x n The data is projected into a low-dimensional active subspace to reduce its dimensionality. Each sample point is mapped to the new low-dimensional space through matrix multiplication to simplify the optimization process.
7. The method for optimizing the intake and exhaust phases of a small rotary engine according to claim 6, characterized in that, In step S7: the optimal direction is identified based on the active subspace method, and the intake and exhaust phase parameters are optimized in the dimensionality-reduced space. The gradient ascent method is used to determine the parameter combination that maximizes the target parameter index. The method further comprises: Step S8: Verify the optimization results; The optimized parameter combination is simulated and verified to confirm its feasibility and performance in an actual engine. The impact of the optimized parameter combination on the target parameter index is evaluated through the simulation verification results to determine whether it has achieved the expected optimization effect. Step S9: Implement the optimal parameter combination; The validated optimal combination of intake and exhaust phase parameters was applied to a real small rotary engine for actual testing and performance evaluation. Based on the test results, the parameters were further fine-tuned to ensure the optimized effect of the engine in actual operation.
8. A small rotary engine intake and exhaust phase optimization system, characterized in that, The system includes a processing unit configured to perform: Factors related to the intake and exhaust phases of a small rotary engine are determined; wherein the factors include: intake initiation angle, intake duration angle, intake port equivalent diameter, exhaust initiation angle, exhaust duration angle, and exhaust port equivalent diameter; Determine the value range of each factor related to the intake and exhaust phases of a small rotary engine; Determine the multidimensional mapping relationship between the variables of each discretized factor and the target parameter index; Optimization is based on the multidimensional change relationship in the multidimensional mapping relationship; among which, based on a small sample dataset, the relationship between input variables and output parameters is captured, gradient information is extracted by using a fitted regression model, the synergistic change characteristics between input factors are obtained, and key factors are identified and the basis of the active subspace is extracted. Determine the dimension k of the active subspace; Project data into the active subspace; Optimize the combination of intake and exhaust phase parameters.
9. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the intake and exhaust phase optimization method for a small rotary engine as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the intake and exhaust phase optimization method for a small rotary engine as described in any one of claims 1-7.