Dual fuel injection method for natural gas dual fuel engine based on multi-algorithm fusion

Abstract

The application discloses a natural gas dual-fuel engine dual-fuel injection method based on multi-algorithm fusion. The natural gas dual-fuel engine dual-fuel injection method based on multi-algorithm fusion fuses a probability reasoning framework of Bayesian optimization, global dynamic search of a chaotic particle swarm algorithm and a gradient updating mechanism of back propagation (BP), and thus the precision of model control is improved compared with a single-algorithm model. The natural gas dual-fuel engine dual-fuel injection method based on multi-algorithm fusion is applied to optimization and correction of a traditional MAP+PID control, and thus the safety of engine operation is ensured when the control model outputs uncertainty in extreme working conditions. The natural gas dual-fuel engine dual-fuel injection method based on multi-algorithm fusion can realize online updating of a model, improve the self-adaptability of injection control and ensure efficient operation of the natural gas dual-fuel engine in complex and changeable working conditions.

Description

Technical Field

[0001] This invention relates to the technical field of engine fuel injection control methods, specifically to a fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion. Background Technology

[0002] Natural gas dual-fuel engines are characterized by high efficiency and low emissions, making them an important means of achieving energy conservation and emission reduction in the transportation sector. Currently, the mainstream natural gas dual-fuel engines use diesel to ignite natural gas for combustion. Compared with traditional diesel engines, due to the addition of a fuel injection system and the significant differences between the two fuels in terms of physical state, response speed, and combustion characteristics, the MAP+PID control method for transmission and control presents multi-dimensional challenges in injection control, such as metering accuracy, transient response, and narrow combustion boundary. Summary of the Invention

[0003] In view of this, the present invention proposes a dual-fuel control algorithm for a natural gas dual-fuel engine based on the fusion of multiple algorithms including Bayesian optimization, chaotic particle swarm optimization, and backpropagation (BP) neural network. By optimizing the traditional MAP+PID control method, the injection control accuracy and transient response can be improved.

[0004] The specific technical solution adopted in this invention is as follows: A fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion includes the following steps: S1. Construct a dual-fuel injection control model that integrates multiple algorithms; S2. Online application control model; S3. Online upgrade control model.

[0005] Preferably, in S1, the following steps are included: S101. Data Acquisition; On a dual-fuel engine, the following data were collected at different torque and speed: intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, air-fuel ratio, diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity. Torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio are included in the control model input parameter dataset. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity are included in the control model output parameter dataset. S102. Data Acquisition; The model input parameter dataset and output parameter dataset are normalized. The processed dataset was divided into a training set and a validation set in a 4:1 ratio. S103. Determine the parameters of the control model neural network. Based on the number of model input parameters, it was determined that the neural network input layer contains 9 neurons; Based on the number of output parameters of the model, it is determined that the input layer of the neural network contains 7 neurons; The number of neurons in the hidden layer and the learning rate were chosen as hyperparameters for Bayesian optimization, with search space ranges of [5, 50] and [0.00001, 0.1], respectively. In the Bayesian optimization process, the optimal hyperparameters are determined by combining the K-fold cross method; The initial weights and thresholds of the control model neural network are determined using the chaotic particle swarm optimization algorithm. The model is trained using backpropagation to determine the weights and thresholds of the neural network controlling the model.

[0006] Preferably, in S2, the following steps are included: S201. Deploy the trained control model to the dual-fuel engine ECU; S202. When the engine is running, the ECU calculates the required torque based on the accelerator pedal signal and inputs it to the control model. At the same time, it processes the signals collected by various sensors and inputs the values ​​of speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio to the control model. S203. The control model calculates the output parameters, namely diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity, based on these input parameters, and inputs these parameter values ​​to the ECU. The S204.ECU compares the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the judgment model with the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram. S205. When their differences are within the set compensation threshold range, the ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the model into control signals, which drive the diesel injector and natural gas injector to work respectively.

[0007] S206. When their difference is within the set compensation threshold range, the ECU will convert the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram into control signals to drive the diesel injector and natural gas injector to work respectively.

[0008] Preferably, in S3, the following steps are included: At time t, the S301.ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity into control signals to drive the diesel injectors and natural gas injectors. At time t+1, it collects data on torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio. S302. The torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio data at time t+1 are used as the new control model input parameter data. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity data at time t are used as the corresponding control model output parameter data and stored in the ECU's model training dataset buffer. S303. After the engine has run for a set time, the control model will use the data in the ECU model training dataset buffer to retrain the model and generate a new model to adapt to the changes in the engine's operating state.

[0009] Beneficial effects: 1. The dual-fuel injection method for natural gas dual-fuel engines based on multi-algorithm fusion of the present invention integrates the probabilistic inference framework of Bayesian optimization, the global dynamic search of chaotic particle swarm optimization algorithm, and the gradient update mechanism of backpropagation (BP), which improves the accuracy of model control compared with single-algorithm models; wherein: When the ECU receives the diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the control model, it will first compare them with the corresponding data in the MAP diagram in the ECU. When the difference is less than the set compensation threshold, the diesel injector and natural gas injector will be driven to work according to the injection parameters calculated by the control model. When the difference is greater than the set compensation threshold, the diesel injector and natural gas injector will be driven to work according to the injection parameters in the MAP diagram.

[0010] The ECU collects diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity at time t as new output data for the control model. At time t+1, it collects torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio data as new input data for the corresponding control model and saves them in the ECU's model training dataset buffer. After the engine has run for a set time, the model will be retrained using the training data in the buffer and a new control model will be generated.

[0011] 2. The dual-fuel injection method for natural gas dual-fuel engines based on multi-algorithm fusion of the present invention is applied as an optimization and correction of the traditional MAP+PID control, ensuring the safety of engine operation when the control model outputs uncertainty under extreme operating conditions.

[0012] 3. The dual-fuel injection method for natural gas dual-fuel engines based on multi-algorithm fusion of the present invention can realize online model updates, improve the adaptability of injection control, and ensure the efficient operation of natural gas dual-fuel engines under complex and variable operating conditions. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the process of the dual-fuel injection method for a natural gas dual-fuel engine based on multi-algorithm fusion according to the present invention. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1 A fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion includes the following steps: S1. Construct a dual-fuel injection control model that integrates multiple algorithms; S101. Data Acquisition; On a dual-fuel engine, the following data were collected at different torque and speed: intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, air-fuel ratio, diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity. Torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio are included in the control model input parameter dataset. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity are included in the control model output parameter dataset. S102. Data Acquisition; The model input parameter dataset and output parameter dataset are normalized. The processed dataset was divided into a training set and a validation set in a 4:1 ratio. S103. Determine the parameters of the control model neural network. Based on the number of model input parameters, it was determined that the neural network input layer contains 9 neurons; Based on the number of output parameters of the model, it is determined that the input layer of the neural network contains 7 neurons; The number of neurons in the hidden layer and the learning rate were chosen as hyperparameters for Bayesian optimization, with search space ranges of [5, 50] and [0.00001, 0.1], respectively. In the Bayesian optimization process, the optimal hyperparameters are determined by combining the K-fold cross method; The initial weights and thresholds of the control model neural network are determined using the chaotic particle swarm optimization algorithm. The model is trained using backpropagation to determine the weights and thresholds of the neural network controlling the model.

[0016] S2. Online application control model; S201. Deploy the trained control model to the dual-fuel engine ECU; S202. When the engine is running, the ECU calculates the required torque based on the accelerator pedal signal and inputs it to the control model. At the same time, it processes the signals collected by various sensors and inputs the values ​​of speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio to the control model. S203. The control model calculates the output parameters, namely diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity, based on these input parameters, and inputs these parameter values ​​to the ECU. The S204.ECU compares the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the judgment model with the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram. S205. When their differences are within the set compensation threshold range, the ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the model into control signals, which drive the diesel injector and natural gas injector to work respectively.

[0017] S206. When their difference is within the set compensation threshold range, the ECU will convert the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram into control signals to drive the diesel injector and natural gas injector to work respectively.

[0018] S3. Online upgrade of the control model; At time t, the S301.ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity into control signals to drive the diesel injectors and natural gas injectors. At time t+1, it collects data on torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio. S302. The torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio data at time t+1 are used as the new control model input parameter data. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity data at time t are used as the corresponding control model output parameter data and stored in the ECU's model training dataset buffer. S303. After the engine has run for a set time, the control model will use the data in the ECU model training dataset buffer to retrain the model and generate a new model to adapt to the changes in the engine's operating state.

[0019] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion, characterized in that, The steps include the following: S1. Construct a dual-fuel injection control model that integrates multiple algorithms; S2. Online application control model; S3. Online upgrade control model.

2. The fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion according to claim 1, characterized in that, In S1, the following steps are taken: S101. Data Acquisition; On a dual-fuel engine, the following data were collected at different torque and speed: intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, air-fuel ratio, diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity. Torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio are included in the control model input parameter dataset. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity are included in the control model output parameter dataset. S102. Data Acquisition; The model input parameter dataset and output parameter dataset are normalized. The processed dataset was divided into a training set and a validation set in a 4:1 ratio. S103. Determine the parameters of the control model neural network. Based on the number of model input parameters, it was determined that the neural network input layer contains 9 neurons; Based on the number of output parameters of the model, it is determined that the input layer of the neural network contains 7 neurons; The number of neurons in the hidden layer and the learning rate were chosen as hyperparameters for Bayesian optimization, with search space ranges of [5, 50] and [0.00001, 0.1], respectively. In the Bayesian optimization process, the optimal hyperparameters are determined by combining the K-fold cross method; The initial weights and thresholds of the control model neural network are determined using the chaotic particle swarm optimization algorithm. The model is trained using backpropagation to determine the weights and thresholds of the neural network controlling the model.

3. The fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion according to claim 1, characterized in that, In S2, the following steps are taken: S201. Deploy the trained control model to the dual-fuel engine ECU; S202. When the engine is running, the ECU calculates the required torque based on the accelerator pedal signal and inputs it to the control model. At the same time, it processes the signals collected by various sensors and inputs the values ​​of speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio to the control model. S203. The control model calculates the output parameters, namely diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity, based on these input parameters, and inputs these parameter values ​​to the ECU. The S204.ECU compares the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the judgment model with the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram. S205. When their differences are within the set compensation threshold range, the ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity calculated by the model into control signals, which drive the diesel injector and natural gas injector to work respectively. S206. When their difference is within the set compensation threshold range, the ECU will convert the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity in the MAP diagram into control signals to drive the diesel injector and natural gas injector to work respectively.

4. The fuel injection control method for a natural gas dual-fuel engine based on multi-algorithm fusion according to claim 1, characterized in that, In S3, the following steps are involved: At time t, the S301.ECU converts the values ​​of diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity into control signals to drive the diesel injectors and natural gas injectors. At time t+1, it collects data on torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio. S302. The torque, speed, intake air temperature, intake air pressure, boost pressure, EGR valve opening, coolant temperature, exhaust temperature, and air-fuel ratio data at time t+1 are used as the new control model input parameter data. The diesel pre-injection timing, diesel pre-injection pulse width, diesel main injection timing, diesel main injection pulse width, diesel injection system rail pressure, natural gas injection timing, and natural gas injection quantity data at time t are used as the corresponding control model output parameter data and stored in the ECU's model training dataset buffer. S303. After the engine has run for a set time, the control model will use the data in the ECU model training dataset buffer to retrain the model and generate a new model to adapt to the changes in the engine's operating state.

Images