Diesel engine air system control method and device, electronic equipment and automobile

Abstract

The invention discloses a diesel engine air system control method and device, electronic equipment and an automobile, and relates to the technical field of automobiles. According to the scheme, the method comprises the steps that an estimation disturbance matrix is obtained based on a state observer, and a hybrid model built based on a mechanism model and a radial neural network prediction model is obtained; the input of the hybrid model is target data in the engine operation process and the estimated disturbance matrix, and the output of the hybrid model is predicted TVA opening, predicted VGT opening, predicted EGR opening, predicted intake manifold pressure, predicted exhaust pressure and predicted EGR rate. And a decoupling controller is adopted to determine a control sequence in the prediction domain based on an output result of the air system hybrid model and the estimation disturbance matrix. According to the scheme, the mechanism model, the neural network compensation model and the disturbance observer are fused, a known mechanism and an unknown mechanism are predicted at the same time, rolling optimization and feedback correction are completed at the same time, decoupling control over a complex air system is achieved, and the overall performance of an engine is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of automotive technology, and more specifically to a diesel engine air system control method, device, electronic equipment, and automobile. Background Technology

[0002] The diesel engine air system is a complex nonlinear system involving the control of multiple components, including high-pressure exhaust gas recirculation (EGR), variable geometry turbocharger (VGT), and intake throttle valve (TVA). The coupling relationships between these components affect the control accuracy and robustness of the diesel engine air system manifold pressure, turbine inlet pressure, and EGR rate. This makes it difficult for deviation control algorithms based on single-input single-output to guarantee control performance. Furthermore, since the coupling strength between actuators varies under different operating conditions, it is difficult for the control algorithm to guarantee good global performance. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a diesel engine air system control method, device, electronic device, and automobile to improve the strong coupling between actuators under different operating conditions.

[0004] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0005] A diesel engine air system control method, comprising:

[0006] The estimated disturbance matrix is ​​obtained based on the state observer, and the estimated disturbance matrix includes the total disturbance of intake manifold pressure, the total disturbance of exhaust pressure, and the total disturbance of EGR rate.

[0007] The output of the air system hybrid model is obtained. The air system hybrid model is a hybrid model based on a mechanistic model and a radial neural network prediction model. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law, which includes: predicting TVA opening, predicting VGT opening, predicting EGR opening, and predicting intake manifold pressure, predicting exhaust pressure, and predicting EGR rate.

[0008] A decoupled controller is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period.

[0009] The diesel engine air system is controlled based on the control sequence.

[0010] Optionally, in the above diesel engine air system control method, the target quantity and real-time feedback quantity include:

[0011] Engine speed, fuel injection quantity, model parameters, actual values ​​of boost pressure, actual values ​​of exhaust manifold pressure, actual values ​​of EGR rate, and effective flow areas of TVA, VGT, and EGR are all determined by the current engine operating conditions.

[0012] Optionally, in the diesel engine air system control method, the mechanistic model in the air system mixing model is:

[0013]

[0014] Where A is the state matrix, The These are the model parameters;

[0015] The It is a vector state matrix. The For intake manifold pressure, This refers to the exhaust manifold pressure. EGR rate;

[0016] The For the input matrix, The These are the model parameters;

[0017] The The input vector matrix, The The effective flow cross-sectional area of ​​the TVA valve is... For the effective flow cross-sectional area of ​​VGT, the This refers to the effective flow cross-sectional area of ​​the EGR valve.

[0018] The Here is the perturbation matrix. The These are model parameters;

[0019] The For the unmodeled model matrix, The The intake manifold pressure is predicted based on a neural network model. The exhaust pressure is predicted based on a neural network. The EGR rate is predicted based on a neural network model.

[0020] F is the perturbation matrix. The This refers to the total pressure disturbance in the intake manifold. This represents the total disturbance of exhaust pressure. For the total EGR rate disturbance;

[0021] The For the output matrix, ;

[0022] The radial neural network prediction model in the hybrid model of the system is as follows: ;

[0023] in Indicates engine speed. Indicates the amount of fuel injected. = , = , = , = .

[0024] Optionally, in the above diesel engine air system control method, the decoupled controller determines the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix, including:

[0025] Obtain the reciprocal h of the input vector matrix;

[0026] The estimated perturbation matrix is ​​used to replace the perturbation matrix in the built-in function of the air system mixture model, and the built-in function of the transformed air system mixture model is transformed into a state equation based on h:

[0027]

[0028] in, , , , ,in This is an estimate of F. ;

[0029] The state equations are discretized using the Euler forward discretization method based on the discrete controller to obtain discrete equations. The discrete equations are then processed to obtain the control sequence in the prediction domain.

[0030] Optionally, in the above diesel engine air system control method, the discrete equations are processed to obtain a control sequence in the prediction domain, including:

[0031] Based on the relationship between time k+1 and time k, the discrete equation is iterated within the control region;

[0032] Substituting the iteration results into the cost function yields the control sequence within the prediction domain.

[0033] Optionally, in the above diesel engine air system control method, before obtaining the output results of the air system mixing model, the following steps are included:

[0034] Obtain engine status data corresponding to the current operating condition of the engine, and determine model parameters that match the engine status data.

[0035] A diesel engine air system control device includes a disturbance acquisition unit for acquiring an estimated disturbance matrix based on a state observer. The estimated disturbance matrix includes total intake manifold pressure disturbance, total exhaust pressure disturbance, and total EGR rate disturbance.

[0036] The prediction unit is used to acquire the output results of the air system hybrid model, which is a hybrid model based on a mechanistic model and a radial neural network prediction model. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law, which includes: predicting TVA opening, predicting VGT opening, predicting EGR opening, and predicting intake manifold pressure, predicting exhaust pressure, and predicting EGR rate.

[0037] The decoupling unit is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix using the decoupling controller. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period.

[0038] A control unit for controlling the diesel engine air system based on the control sequence.

[0039] An electronic device includes: at least one processing device and a storage device connected to said processing device, wherein:

[0040] The storage device is used to store computer programs;

[0041] The processing device is used to execute the computer program so that the electronic device can implement any of the above-described diesel engine air system control methods.

[0042] An automobile includes the aforementioned electronic equipment. Optionally, the engine of the automobile is a diesel engine.

[0043] Based on the above technical solution, the solution provided by this embodiment of the invention includes: obtaining an estimated disturbance matrix based on a state observer; obtaining a hybrid model built based on a mechanistic model and a radial neural network prediction model; the input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix; the target data includes target quantities and real-time feedback quantities; the output of the air system hybrid model is: predicted TVA opening, predicted VGT opening, predicted EGR opening, predicted intake manifold pressure, predicted exhaust pressure, and predicted EGR rate; and a decoupled controller is used to determine the control sequence within the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix. The control sequence within the prediction domain includes the optimal control law at each moment within a preset time period. Therefore, the above solution integrates the mechanistic model, the neural network compensation model, and the disturbance observer to achieve simultaneous prediction of known and unknown mechanistic components, while simultaneously completing rolling optimization and feedback correction, thus achieving decoupled control of the complex air system and ensuring the overall performance of the engine. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0045] Figure 1 This is a schematic flowchart of the diesel engine air system control method disclosed in the embodiments of this application;

[0046] Figure 2 This is a schematic diagram of the structure of the diesel engine air system control device disclosed in the embodiments of this application. Detailed Implementation

[0047] 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.

[0048] Feedforward control: In addition to the basic closed-loop control, an extra open-loop control quantity is added to improve the control effect of the system together with the closed-loop control quantity.

[0049] VGT: Variable Geometry Turbocharger, which can dynamically adjust the boost capacity of the engine by adjusting the tilt angle of the turbocharger blades through an electric motor. It is usually controlled by the boost pressure in the intake manifold.

[0050] EGR: Exhaust Gas Recirculation device, which can control the amount of exhaust gas introduced by adjusting the valve angle through a motor, thereby reducing NOx emissions. It usually uses the amount of fresh air intake as the control target.

[0051] TVA: Intake throttle valve, which can adjust the valve opening through a motor to achieve throttling control of the intake air volume, and can be used in conjunction with EGR rate adjustment.

[0052] Extended State Observer: A full-order state observer built on state-space equations. Its characteristic is that, in addition to the original state variables, the disturbance part that cannot be modeled is also turned into state variables and predicted at the same time.

[0053] MPC: Model Predictive Control.

[0054] This application addresses the multi-objective tracking control problem of intake manifold pressure, turbine inlet pressure, and EGR rate in diesel engines equipped with VGT, EGR, and TVA, proposing a global intelligent decoupled control algorithm. Based on mechanical, fluid, and thermodynamic principles, this invention explores and analyzes the mechanism of the diesel engine air system with EGR, VGT, and TVA components, constructing a control-oriented air system mechanism model. Furthermore, to improve the accuracy of the air system model, a data model based on radial basis function neural networks is built, forming a hybrid model composed of the mechanism model and the data model, thus improving the accuracy of the control-oriented model. Feedforward control of the dynamic process of the engine air system is performed based on the hybrid model. To compensate for the performance degradation of the MPC caused by the deviation between the hybrid model and the engine, it is equivalent to the total disturbance of the three control channels. An extended state observer is built to provide feedback compensation for the total disturbance, and the estimated value is fed back to the MPC controller for predictive control, improving the model's control effect.

[0055] First, this application pre-establishes a mechanistic model that includes the dynamic characteristics of intake manifold pressure, exhaust manifold pressure, and EGR rate, and simplifies it into a linear state equation that varies with operating conditions through parameter identification; that is, it establishes a linear equation for the dynamic characteristics of intake manifold pressure, exhaust manifold pressure, and EGR rate that varies with operating conditions.

[0056] Secondly, a radial basis function neural network (RBFNN) is introduced to construct a data model. By selecting key parameters from multiple dimensions such as engine speed and fuel injection quantity as inputs, a single hidden layer (16 neurons), tanh activation function and mean squared error loss function are used for training to deal with the problem of small datasets. The outputs of the mechanism model and the data model are fused to form a hybrid state equation, and an extended state space expression containing model perturbation terms is constructed.

[0057] Step 3: For disturbances not modeled in the model, design an extended state observer to estimate internal and external disturbances in real time using state variable extension techniques;

[0058] Step 4: Further expand the state variables to include the control input in the state space, and establish a discretized model predictive control (MPC) framework by combining Euler discretization. By rolling optimization of the cost function (including tracking error and control energy consumption weighted terms), decoupled control of the multivariable control system is achieved.

[0059] For details, see Figure 1 The diesel engine air system control method disclosed in this application includes:

[0060] Step S101: Obtain the estimated disturbance matrix based on the state observer The estimated disturbance matrix includes total intake manifold pressure disturbance, total exhaust pressure disturbance, and total EGR rate disturbance.

[0061] In this solution, the total intake manifold pressure disturbance can be obtained based on real-time monitoring and filtering of sensor signals. The core principle is that the intake manifold pressure disturbance is mainly caused by changes in throttle opening, engine speed fluctuations, and changes in the sealing of the intake system. The pressure signal can be acquired in real time using an intake manifold absolute pressure sensor (MAP), and the disturbance components can be separated by a filtering algorithm to obtain the total intake manifold pressure disturbance.

[0062] The total exhaust pressure disturbance can be calculated dynamically based on exhaust back pressure and flow velocity. The exhaust pressure disturbance is mainly caused by exhaust gas pulses, turbocharger dynamic response, and changes in exhaust system resistance. The total exhaust pressure disturbance can be calculated by monitoring these data.

[0063] EGR rate disturbances are caused by exhaust gas flow fluctuations, EGR valve response delays, and uneven mixing in the intake system. The total EGR rate disturbance can be calculated by detecting these data.

[0064] In this scheme, the estimated perturbation matrix can be obtained based on the state observer (i.e., the perturbation observer) ESO.

[0065] Step S102: Obtain the output results of the air system hybrid model. The air system hybrid model is a hybrid model based on a mechanistic model and a radial neural network prediction model. Specifically, the mechanistic model is a simplified mechanistic model of the air system constructed based on the air system flow and thermodynamic characteristics. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law, which includes: predicting the TVA opening, predicting the VGT opening, predicting the EGR opening, and predicting the intake manifold pressure p22, the exhaust pressure p3, and the EGR rate.

[0066] In this step, it is necessary to pre-construct an air system hybrid model based on EGR-VGT-TVA. The hybrid model includes a mechanistic model and a data model, and the data model is a radial neural network prediction model.

[0067] The function of the mechanistic model is:

[0068] (Formula 1)

[0069] in: , , , , , , , , ;

[0070] In the above formula, This represents the temperature inside the intake manifold. This represents the intake manifold volume. It is the ideal gas constant. This represents the intake manifold pressure. The effective flow cross-sectional area of ​​the TVA valve. This is the temperature after intercooling. For intercooling pressure, The coefficients to be calibrated are . The effective flow cross-sectional area of ​​the EGR valve. The temperature at the vortex inlet. This refers to the exhaust manifold pressure. The coefficients to be calibrated are . For inflation efficiency, For cylinder volume, This refers to the engine speed. Represents the exhaust manifold temperature. Represents the exhaust manifold volume. The effective flow cross-sectional area of ​​VGT. The temperature at the vortex inlet. The pressure behind the vortex. The coefficients to be calibrated are . EGR rate This represents the density of the intake gas.

[0071] In this scheme, after identifying the engine state, the above parameters are extracted from the engine state data, and the values ​​of each coefficient in the function of the mechanism model are obtained by processing each of the above parameters.

[0072] To facilitate parameter identification, this application further simplifies the function of the above-mentioned mechanism model, simplifying it as follows:

[0073] (Formula 2)

[0074] Each of the above functions For model parameters, model parameters The engine state data corresponding to the current operating condition of the engine can be obtained through the aforementioned process, and model parameters matching the engine state data can be determined based on the aforementioned formulas. . The representative model has known perturbations.

[0075] The radial neural network prediction model is expressed as follows:

[0076] in This is the output value of the radial neural network prediction model. Indicates engine speed. Indicates the amount of fuel injected. = , = , = , = That is, the left side of the above formula represents the output value of the radial neural network prediction model, and the right side represents the input data of the radial neural network prediction model. After inputting the parameters on the right side into the radial neural network prediction model, the model can obtain and output the prediction result based on the input data. To prevent overfitting, the radial neural network prediction model includes a hidden layer with 16 neurons, and uses mean squared error as the loss function. The activation function of the radial neural network prediction model is the tanh function.

[0077] After constructing and training the data model, the output value of the data model and the state equation (Formula 1) of the mechanistic model are fused to establish the hybrid model formula:

[0078] (Formula 3)

[0079] As shown in the above equation, the hybrid model consists of differential equations for intake manifold pressure, exhaust manifold pressure, and EGR rate, with the mechanistic model responsible for the output. , , Parameters, and data model output , , parameter.

[0080] A practical air system mixing model can be expressed in the form of the following state equations:

[0081] (Formula 4)

[0082] in:

[0083] Where A is the state matrix, The These are the model parameters;

[0084] The It is a vector state matrix. The For intake manifold pressure, This refers to the exhaust manifold pressure. EGR rate;

[0085] The For the input matrix, The These are the model parameters;

[0086] The The input vector matrix, The The effective flow cross-sectional area of ​​the TVA valve is... For the effective flow cross-sectional area of ​​VGT, the This refers to the effective flow cross-sectional area of ​​the EGR valve.

[0087] The Here is the perturbation matrix. The These are model parameters;

[0088] The For the unmodeled model matrix, The The intake manifold pressure is predicted based on a neural network model. The exhaust pressure is predicted based on a neural network. The EGR rate is predicted based on a neural network model.

[0089] F is the perturbation matrix for models that have not yet been built. The This refers to the total pressure disturbance in the intake manifold. This represents the total disturbance of exhaust pressure. For the total EGR rate disturbance;

[0090] The For the output matrix, .

[0091] In the above formula, F is an unknown quantity, which needs to be estimated using the disturbance matrix determined in step S101. Replace it. The state observer determines the estimated perturbation matrix. Since the counting of the state observer is relatively mature, this application can use existing observers to determine the estimated perturbation matrix.

[0092] Specifically: The state extension observer is an observer based on the air system hybrid model. The state observer expands the disturbance matrix F into state variables. To simplify calculations, the derivative of the disturbance is set to 0, and the state is defined. The estimated value is Output The estimated value is Establish an extended state observer:

[0093] (Formula 5)

[0094] in: The estimated perturbation matrix for the extended state observer includes both internal and external perturbations of the model. , The observer gain matrix;

[0095] .

[0096] Solving the extended state observer allows us to determine the estimated perturbation matrix. .

[0097] The engine status is monitored, and the required data for Formula 4 is extracted from the monitoring results. After determining Formula 4 and estimating the disturbance matrix, Substituting these data into Formula 4, the optimal control law can be calculated: predicting TVA opening, VGT opening, EGR opening, intake manifold pressure, exhaust pressure, and EGR rate. After obtaining these prediction results, the engine control system can control the engine based on these prediction results.

[0098] In summary, among the above models, the mechanistic model is used to provide the basic function as shown in Equation 1, and the radial neural network model is used to predict parameters. and ,use Compensation is applied to the first term of the function in Formula 1, using... To compensate for the second term in Formula 1, the following method is used: The third term in Equation 1 is compensated, and the air system hybrid model shown in Equation 4 is constructed based on the compensation result. After constructing the estimated disturbance matrix using an observer and using the estimated disturbance matrix as the disturbance matrix F in Equation 4, Equation 4 is solved to obtain the optimal control law.

[0099] This step improves the accuracy of the control-oriented model and the computational precision of the air system model by constructing a hybrid model consisting of a mechanistic model and a data model.

[0100] Furthermore, to compensate for the performance degradation of the hybrid model caused by the deviation between the hybrid model and the engine, it is equivalent to the total disturbance of the three control channels. An extended state observer is built to calculate and estimate the disturbance matrix. Feedback compensation is performed based on the disturbance matrix, and the estimated value is compensated into the hybrid model for predictive control, thereby improving the control effect of the hybrid model.

[0101] Step S103: The decoupled controller is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period.

[0102] After determining the predicted TVA opening, predicted VGT opening, predicted EGR opening, and predicted intake manifold pressure, predicted exhaust pressure, and predicted EGR rate, this application can also predict the optimal control law (predicted TVA opening, predicted VGT opening, predicted EGR opening, predicted intake manifold pressure, predicted exhaust pressure, and predicted EGR rate) for each time point within the preset time domain. In this process, the control sequence for each time point within the prediction domain, i.e., the optimal control law for each time point, can be determined by the decoupled controller based on the output results of the air system mixing model and the estimated disturbance matrix.

[0103] Specifically, in this process, the soft constraints on the control increment are simplified, and the reciprocal h of the input vector matrix u is obtained, i.e. (Formula 6)

[0104] By replacing the disturbance matrix in the built-in function of the air system mixing model with an estimated disturbance matrix, Equation 4 is transformed into: (Formula 7)

[0105] Let z represent x and u:

[0106] (Formula 8)

[0107] That is, the functions in the first two rows of Formula 7 are represented as shown in Formula 8.

[0108] in, y=[C 0], ;

[0109] Let the state matrix Input matrix perturbation matrix The output matrix y=[C 0] yields the continuity equation, the expression for which is:

[0110] (Formula 9)

[0111] After obtaining the state equation shown in Formula 9, the above state equation is discretized using the Euler forward discretization method to obtain the discrete equation, which is expressed as:

[0112] (Formula 10)

[0113] in, For discrete time steps, , , ;

[0114] Starting from time k, define , This represents the state predicted at time k+1 from time k, and so on. Next, define the state sequence matrix within the prediction step:

[0115] Control sequence matrix ; Observation perturbation sequence matrix: ; To predict the step size.

[0116] Based on the relationship between time k+1 and time k, the discrete equation is iterated over within the control region to obtain:

[0117] (Formula 11)

[0118] in, , ,

[0119] ;

[0120] Substituting Formula 11 obtained from the iteration into the cost function, we obtain the control sequence within the prediction domain;

[0121] The cost function is:

[0122] (Formula 12)

[0123] in The state variable sequence is obtained through a prediction model; The target state sequence; Let be the control state sequence, be the variable to be determined; Q is the tracking weight matrix; and R is the control weight matrix.

[0124] Substituting formula 11 into formula 12, we get:

[0125] (Formula 13)

[0126] in:

[0127] Specifically, solving formula 13 yields: (Formula 14)

[0128] The cost function shown in Formula 13 The control sequence within the prediction domain can then be obtained. Only the optimal control law of the air system mixture model is taken, i.e., the vector is obtained. The first component is then subjected to rolling optimization calculations to obtain the optimal control law at each moment within each preset time period. That is, by continuously performing rolling optimization calculations based on the optimal control law at the previous moment, the optimal control law at the next moment can be obtained. By continuously performing rolling optimization, the optimal control law at each moment within each preset time period can be obtained. Integrating these optimal control laws yields the control sequence.

[0129] Step S104: Control the diesel engine air system based on the control sequence.

[0130] Therefore, a global intelligent decoupled control algorithm for the air system was established, realizing decoupled control of TVA, VGT, and EGR controllers with intake pressure, vortex inlet pressure, and EGR rate as control targets. As can be seen from the above scheme, the method integrates a simplified mechanistic model with a data model constructed based on a radial basis function neural network to form a hybrid air system model, achieving feedforward decoupled control of known and unknown model information. The estimated disturbance matrix is ​​determined based on an extended state observer, and this estimated disturbance matrix is ​​substituted into the hybrid air system model, thus replacing the hybrid air system model with the estimated disturbance matrix. The disturbance amount is calculated by solving the air system mixing model using the collected engine state data. This yields the predicted TVA opening, VGT opening, EGR opening, intake manifold pressure, exhaust pressure, and EGR rate. Furthermore, by using the output of the decoupled controller based on the air system mixing model and the estimated disturbance matrix to determine the control sequence within the prediction domain, the predicted TVA opening, VGT opening, EGR opening, intake manifold pressure, exhaust pressure, and EGR rate at various times within a preset time period can be obtained. Therefore, the above scheme integrates the mechanistic model, neural network compensation model, and disturbance observer to simultaneously predict both known and unknown mechanistic components. Simultaneously, rolling optimization and feedback correction are performed, achieving decoupled control of the complex air system and ensuring the overall engine performance.

[0131] This embodiment discloses a diesel engine air system control device. For the specific working content of each unit in the device, please refer to the above method embodiment.

[0132] The diesel engine air system control device provided in the embodiments of the present invention is described below. The diesel engine air system control device described below and the diesel engine air system control method described above can be referred to in correspondence.

[0133] See Figure 2 The diesel engine air system control device includes:

[0134] The disturbance acquisition unit 10, corresponding to step S101 of the above method, is used to acquire an estimated disturbance matrix based on the state observer. The estimated disturbance matrix includes the total disturbance of intake manifold pressure, the total disturbance of exhaust pressure, and the total disturbance of EGR rate.

[0135] Prediction unit 20, corresponding to step S102 of the above method, is used to obtain the output results of the air system hybrid model. The air system hybrid model is a hybrid model based on a mechanistic model and a radial neural network prediction model. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law. The optimal control law includes: predicting TVA opening, predicting VGT opening, predicting EGR opening, and predicting intake manifold pressure, predicting exhaust pressure, and predicting EGR rate.

[0136] The decoupling unit 30, corresponding to step S103 of the above method, is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix using the decoupling controller. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period.

[0137] Control unit 40, corresponding to step S104 of the above method, is used to control the diesel engine air system based on the control sequence.

[0138] An electronic device includes: at least one processing device and a storage device connected to said processing device, wherein:

[0139] The storage device is used to store computer programs;

[0140] The processing device is used to execute the computer program so that the electronic device can implement any of the diesel engine air system control methods described above.

[0141] The electronic devices mentioned in this article can be ECU (Electronic Control Unit), VCU (Vehicle Control Unit), MCU (Micro Controller Unit), HCU (Hybrid Control Unit), etc.

[0142] An automobile includes the aforementioned electronic equipment. The engine of the automobile may be a diesel engine.

[0143] For ease of description, the above system is described by dividing it into various modules based on their functions. Of course, in implementing this invention, the functions of each module can be implemented in one or more software and / or hardware components.

[0144] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, for system or system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be referred to the descriptions in the method embodiments. The systems and system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0145] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0146] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0147] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0148] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for controlling a diesel engine air system, characterized in that, include: The estimated disturbance matrix is ​​obtained based on the state observer, and the estimated disturbance matrix includes the total disturbance of intake manifold pressure, the total disturbance of exhaust pressure, and the total disturbance of EGR rate. The output of the air system hybrid model is obtained. The air system hybrid model is a hybrid model based on a mechanistic model and a radial neural network prediction model. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law, which includes: predicting TVA opening, predicting VGT opening, predicting EGR opening, and predicting intake manifold pressure, predicting exhaust pressure, and predicting EGR rate. A decoupled controller is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period. The diesel engine air system is controlled based on the control sequence.

2. The diesel engine air system control method according to claim 1, characterized in that, The target quantity and real-time feedback quantity include: Engine speed, fuel injection quantity, model parameters, actual values ​​of boost pressure, actual values ​​of exhaust manifold pressure, actual values ​​of EGR rate, and effective flow areas of TVA, VGT, and EGR are all determined by the current engine operating conditions.

3. The diesel engine air system control method according to claim 2, characterized in that, The mechanistic model in the air system mixing model is as follows: Where A is the state matrix. The These are the model parameters; The It is a vector state matrix. The For intake manifold pressure, This refers to the exhaust manifold pressure. EGR rate; The For the input matrix, The These are the model parameters; The The input vector matrix, The The effective flow cross-sectional area of ​​the TVA valve is... For the effective flow cross-sectional area of ​​VGT, the This refers to the effective flow cross-sectional area of ​​the EGR valve. The Here is the perturbation matrix. The These are model parameters; The For the unmodeled model matrix, The The intake manifold pressure is predicted based on a neural network model. The exhaust pressure is predicted based on a neural network. The EGR rate is predicted based on a neural network model. F is the perturbation matrix. The This refers to the total pressure disturbance in the intake manifold. This represents the total disturbance of exhaust pressure. For the total EGR rate disturbance; The For the output matrix, ; The radial neural network prediction model in the hybrid model of the system is as follows: in Indicates engine speed. Indicates the amount of fuel injected. = , = , = , = .

4. The diesel engine air system control method according to claim 3, characterized in that, The control sequence in the prediction domain is determined using a decoupled controller based on the output of the air system hybrid model and the estimated disturbance matrix, including: Obtain the reciprocal h of the input vector matrix; The estimated perturbation matrix is ​​used to replace the perturbation matrix in the built-in function of the air system mixture model, and the built-in function of the transformed air system mixture model is transformed into a state equation based on h: in, , , , ,in This is an estimate of F. ; The state equations are discretized using the Euler forward discretization method based on the discrete controller to obtain discrete equations. The discrete equations are then processed to obtain the control sequence in the prediction domain.

5. The diesel engine air system control method according to claim 4, characterized in that, The discrete equations are processed to obtain the control sequence in the prediction domain, including: Based on the relationship between time k+1 and time k, the discrete equation is iterated within the control region; Substituting the iteration results into the cost function yields the control sequence within the prediction domain.

6. The diesel engine air system control method according to claim 4, characterized in that, Before obtaining the output results of the air system hybrid model, the following should be included: Obtain engine status data corresponding to the current operating condition of the engine, and determine model parameters that match the engine status data.

7. A diesel engine air system control device, characterized in that, include: The disturbance acquisition unit is used to acquire an estimated disturbance matrix based on the state observer. The estimated disturbance matrix includes the total disturbance of intake manifold pressure, the total disturbance of exhaust pressure, and the total disturbance of EGR rate. The prediction unit is used to acquire the output results of the air system hybrid model, which is a hybrid model based on a mechanistic model and a radial neural network prediction model. The input of the air system hybrid model is the target data during engine operation and the estimated disturbance matrix. The target data includes the target quantity and the real-time feedback quantity. The output is the optimal control law, which includes: predicting TVA opening, predicting VGT opening, predicting EGR opening, and predicting intake manifold pressure, predicting exhaust pressure, and predicting EGR rate. The decoupling unit is used to determine the control sequence in the prediction domain based on the output of the air system hybrid model and the estimated disturbance matrix using the decoupling controller. The control sequence in the prediction domain includes the optimal control law at each time point within a preset time period. A control unit for controlling the diesel engine air system based on the control sequence.

8. An electronic device, characterized in that, include: At least one processing device and a storage device connected to the processing device, wherein: The storage device is used to store computer programs; The processing device is used to execute the computer program so that the electronic device can implement the diesel engine air system control method as described in any one of claims 1 to 6.

9. A car, characterized in that, Includes the electronic device as described in claim 8.

10. The automobile according to claim 9, characterized in that, The car's engine is a diesel engine.

Images