Injection control method of diesel-methanol dual-fuel engine

Abstract

The invention provides an injection control method for a diesel-methanol dual-fuel engine, which comprises the following steps: acquiring a rotating speed signal, a load signal and a temperature signal of the engine, and determining initial diesel injection timing data and initial methanol injection timing data under the current working condition by combining preset respective injection MAP maps of dual fuels; according to the initial diesel injection timing data and the initial methanol injection timing data, the response time of each injector is fused, a dual-fuel injection time sequence model under a unified control framework is constructed, and the initial fusion injection moment is obtained; and according to the preliminary fusion injection time and the actually required injection duration, in combination with a preset injection priority rule, optimized dual-fuel injection time sequence control parameters are obtained through a self-adaptive genetic algorithm.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to an injection control method for a diesel-methanol dual-fuel engine. Background Art

[0002] In the field of diesel-methanol dual-fuel engines, the existing technology adopts a dual electronic control unit independent control strategy to drive the diesel and methanol injection systems respectively. In this architecture, complex data exchange and collaborative work are required between the two electronic control units to maintain engine stability. In order to improve efficiency, it is planned to integrate the dual electronic control units into a single electronic control unit to control diesel and methanol injection at the same time. However, this integration has caused unique technical difficulties: the original independent diesel and methanol injection systems have their own hardware response links. After integration, how to coordinate and prioritize the injection timing of the two types of injectors in a single control loop; in addition, the feedback loop of the original diesel control system only needs to focus on the diesel system, but now it needs to consider methanol supply at the same time, which increases the complexity of the mutual influence between the two types of injectors. The core of this technical problem lies in how to ensure that the two completely different injection systems of diesel and methanol can maintain their respective injection accuracy and avoid operation fluctuations caused by injection timing conflicts when sharing a control loop under a unified control framework, especially when the engine operating conditions change. How to make a single control system coordinate the injection priority and smoothness between the two original independent hardware actuators is a unique technical problem in this scenario. Summary of the invention

[0003] The present invention provides an injection control method for a diesel-methanol dual-fuel engine, which mainly includes:

[0004] The engine speed signal, load signal, and temperature signal are obtained, and combined with the preset dual-fuel respective injection MAP diagrams, the initial diesel injection timing data and the initial methanol injection timing data under the current working condition are determined; according to the initial diesel injection timing data and the initial methanol injection timing data, the respective injector response times are integrated, and a dual-fuel injection timing model under a unified control framework is constructed to obtain a preliminary fusion injection moment; the diesel injection pressure and methanol injection pressure signals are obtained, and the target diesel injection amount and the target methanol injection amount are calculated using the preset injector flow model, and then the actual required injection duration data is determined according to the preset pressure difference correction coefficient; according to the preliminary fusion injection moment and the actual required injection duration, the preset injection The optimized dual-fuel injection timing control parameters are obtained through an adaptive genetic algorithm based on the injection priority rule; the optimized dual-fuel injection timing control parameters are combined with the feedback information of the original diesel control system, and the injection timing dynamic adjustment parameters are calculated through a closed-loop PID algorithm to compensate for the difference between the actual injection amount and the expected value caused by the mutual interference generated by the two fuel supplies; the current dual-fuel injection timing is corrected according to the injection timing dynamic adjustment parameters, and the dual-fuel injection timing of the next cycle is predicted through a pre-established long short-term memory neural network model to determine whether there is a possibility that the injection time of the two fuels overlaps or is too close. If so, the injection start time and duration are fine-tuned to obtain the pre-adjusted injection timing.

[0005] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0006] The present invention discloses an injection control method for a dual-fuel engine. The method obtains an engine operating condition signal, determines the initial injection timing in combination with a preset MAP diagram, and constructs a dual-fuel injection timing model under a unified control framework. Furthermore, the target injection amount is calculated based on the injection pressure signal, and the injection timing parameters are optimized using an adaptive genetic algorithm. The present invention also combines a closed-loop PID algorithm and a long short-term memory neural network model to achieve dynamic adjustment and prediction of the injection timing, effectively solving the problem of mutual interference caused by the dual fuel supply. The method can effectively improve the combustion efficiency and emission performance of the dual-fuel engine, and provides a new technical solution for the precise control of the dual-fuel engine. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 The present invention is a flow chart of an injection control method for a diesel-methanol dual-fuel engine.

[0008] Figure 2 It is a schematic diagram of an injection control method of a diesel-methanol dual-fuel engine according to the present invention.

[0009] Figure 3 This is another schematic diagram of an injection control method for a diesel-methanol dual-fuel engine according to the present invention. DETAILED DESCRIPTION

[0010] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0011] like Figure 1-3 In this embodiment, a diesel-methanol dual-fuel engine injection control method may specifically include:

[0012] Step S101, obtaining the engine speed signal, load signal, and temperature signal, and combining the preset dual-fuel respective injection MAP diagrams to determine the initial diesel injection timing data and the initial methanol injection timing data under the current working condition.

[0013] The engine speed signal, load signal and temperature signal are obtained as input parameters; according to the preset dual-fuel injection MAP diagram, the speed, load and temperature parameters are used to look up the table to obtain the theoretical injection timing of diesel and methanol under the current working condition; the actual injection timing data collected by the injection timing sensor is obtained; if the deviation between the actual injection timing and the theoretical injection timing exceeds the preset threshold, it is judged as an injection timing abnormality; for the injection timing abnormality, the Kalman filter algorithm is used to filter the actual injection timing data to obtain the corrected injection timing; the corrected injection timing is transmitted to the electronic control unit of the injection system through the CAN bus as the initial injection timing control parameter of the diesel and methanol injectors; the engine speed, load and temperature under historical working conditions are obtained as input, and the corresponding corrected injection timing is used as output, and the BP neural network algorithm is used for training to establish a nonlinear mapping relationship between the speed, load, temperature and injection timing; during the operation of the engine, the speed, load and temperature parameters are obtained in real time, and they are input into the trained BP neural network model to predict and output the target injection timing of diesel and methanol, and transmit them to the electronic control unit of the injection system to update the injection timing parameters in real time.

[0014] Exemplarily, the engine speed signal, load signal and temperature signal are obtained, and these three signals are regarded as input parameters of the control system. The speed signal can be obtained by the crankshaft position sensor. The load signal reflects the current output power of the engine, which can be obtained by measuring the throttle opening, the intake pressure or the injection amount. The temperature signal refers to the coolant temperature or the exhaust temperature of the engine, which is measured by the temperature sensor. For example, at a certain moment, the engine speed is 2000 rpm, the load is 60%, and the coolant temperature is 90 degrees Celsius. According to the pre-set dual-fuel injection MAP map, the speed, load and temperature parameters obtained previously are used for table lookup operation. The dual-fuel injection MAP map is obtained based on the bench test data calibration, which specifies the ideal injection timing of diesel and methanol under different working conditions. For example, based on the above working conditions, the table is looked up to obtain the theoretical injection timing of diesel as 15 degrees before the top dead center, and the theoretical injection timing of methanol as 18 degrees before the top dead center. At this time, diesel and methanol are injected simultaneously to ensure the atomization quality and avoid emission problems. The actual injection timing data is obtained through the injection timing sensor installed on the engine. For example, the sensor detects that the actual injection timing of diesel is 13 degrees before the top dead center, and the actual injection timing of methanol is 17 degrees before the top dead center. The actual injection timing data is compared with the theoretical injection timing data obtained by looking up the table. If the deviation between the two exceeds the preset threshold, it is determined that the injection timing is abnormal. For example, the threshold is set to 2 degrees. Since the deviation between the actual injection timing of diesel and the theoretical value is 2 degrees, and the deviation of methanol is 1 degree, only the diesel injection timing deviation is equal to the preset threshold, and the current injection timing is also determined to be abnormal. For the determined injection timing abnormality, the Kalman filter algorithm is used to filter the actual injection timing data to obtain the corrected injection timing. Kalman filtering is an optimal estimation algorithm that can estimate the signal containing noise based on the state equation and measurement equation of the system to obtain a more accurate value. This method is more reliable than taking the average value or directly using the actual injection timing, and can correct some errors caused by aging or wear. It can optimize and adjust the injection timing control in real time, and the Kalman filter does not require long data to achieve good results, and has good real-time performance. After applying the Kalman filter algorithm, the corrected diesel injection timing can be obtained as 14.5 degrees before the top dead center, and the methanol injection timing is 17.5 degrees before the top dead center, which is the final optimization result required. The corrected injection timing data is sent to the ECU via the CAN bus, and these data will be used as the initial injection timing control parameters of the diesel and methanol injectors. The ECU program can be quickly flashed through the CAN bus, and the upgrade can be quickly completed.The use of dual-fuel control will affect the original diesel engine calibration due to the change of working fluid, which may cause the original machine performance to fail under new conditions. Therefore, the calibration in dual-fuel technology needs to pay attention to the changes of two working fluids at the same time to achieve the final control goal, such as maintaining the diesel replacement rate and maintaining a better atomization effect. Obtain the engine speed, load, temperature and other parameters under historical working conditions, which will be used as the input of the neural network. At the same time, the corresponding corrected injection timing under these working conditions is used as the output of the neural network. The BP neural network algorithm is used for training to establish a nonlinear mapping relationship between speed, load, temperature and injection timing. This operation is mainly to achieve the effect of learning, lay the foundation for a better control algorithm, and avoid always executing according to the original control parameters. The BP neural network model is repeatedly trained with enough historical data. When the model converges, that is, the value of the loss function tends to be stable, a model that can more accurately predict the injection timing is obtained. During training, the network continuously updates the connection weights between layers through the back propagation of errors, which can simulate the model under the actual physical scene. Moreover, because the neural network has a large number of historical training results as the basis, it is more refined than MAP. The use of BP neural network can achieve self-adaptation and avoid the use of MAP-based mapping in actual production and life. It can avoid the problem of large deviations in the actual injection device due to various uncontrollable reasons. For example, the decrease in oxygen content in plateau areas may cause abnormal vibration of the engine, and the injection parameters need to be adjusted in time. During the operation of the engine, the speed, load and temperature parameters are obtained in real time, and these parameters are input into the trained BP neural network model. The model will predict and output the target injection timing of diesel and methanol under the current working conditions based on the established nonlinear mapping relationship, and pass these target injection timing data to the ECU for real-time updating of the injection timing control parameters. This can dynamically adjust the fuel injection strategy according to different engine states to achieve refined control and reduce emissions.

[0015] Step S102, based on the initial diesel injection timing data and the initial methanol injection timing data, the respective injector response times are integrated to construct a dual-fuel injection timing model under a unified control framework to obtain a preliminary integrated injection time.

[0016] Obtain the target injection timing signal of the diesel injection system and the target injection timing signal of the methanol injection system; obtain the diesel injector response time and the methanol injector response time; construct a dual-fuel engine injection timing model based on the diesel engine target injection timing signal, the methanol engine target injection timing signal, the diesel injector response time and the methanol injector response time; collect the dual-fuel engine operation data; use the dual-fuel engine operation data as training data, and use the sequence minimum optimization algorithm to train the dual-fuel engine injection timing model to obtain the trained dual-fuel engine injection timing model; calculate the diesel initial injection time and the methanol initial injection time based on the trained dual-fuel engine injection timing model; if the diesel initial injection If the time difference between the diesel initial injection time and the methanol initial injection time is greater than the preset value, the diesel initial injection time and the methanol initial injection time are used as the input of the Mamdani fuzzy control algorithm, and the final diesel injection time and the final methanol injection time are obtained through fine-tuning of the Mamdani fuzzy control algorithm; if the time difference between the diesel initial injection time and the methanol initial injection time is less than the preset value, the diesel initial injection time is used as the final diesel injection time, and the methanol initial injection time is used as the final methanol injection time; a diesel injector control signal is generated according to the final diesel injection time; a methanol injector control signal is generated according to the final methanol injection time; the diesel injector control signal is sent to the diesel injector; and the methanol injector control signal is sent to the methanol injector.

[0017] Exemplarily, obtaining the target injection timing signal of the diesel injection system and the target injection timing signal of the methanol injection system is a key step in the dual-fuel engine control system. The target injection timing signal is usually calculated by the engine control unit (ECU) according to the current working condition. For example, under a certain working condition, the ECU may calculate that the diesel target injection timing is 15 degrees before the top dead center and the methanol target injection timing is 20 degrees before the top dead center. Next, the response time of the diesel injector and the methanol injector is obtained. The injector response time refers to the time delay from receiving the control signal to the actual start of injection. Assume that the response time of the diesel injector is 2 milliseconds and the response time of the methanol injector is 3 milliseconds. These data are usually measured experimentally and stored in the ECU. Based on the above data, a dual-fuel engine injection timing model is constructed. The model comprehensively considers the target injection timing and the injector response time to ensure that the actual injection time is consistent with the target injection time. For example, if the target diesel injection timing is 15 degrees before the top dead center, considering the response time of 2 milliseconds, the model will send a control signal 2 milliseconds in advance. Collecting dual-fuel engine operation data is the basis for model training. The operation data includes parameters such as engine speed, load, and temperature. Assume that 100 sets of data under different working conditions are collected, and each set of data contains speed, load, temperature and the corresponding actual injection timing. The data preprocessing method is used to clean and extract features from the operating data to obtain standardized training data. The SMO optimization objective function is constructed based on the training data, the SMO algorithm parameters are set, and the SMO iteration termination conditions, such as the KKT condition or the upper limit of the number of iterations, are determined. The SMO algorithm is used to train the timing model, and two variables are selected for optimization in each iteration. The Lagrange multiplier is updated by solving the quadratic programming subproblem until the iteration termination conditions are met, and the trained timing prediction model is obtained. The speed, load and temperature data under the current working conditions are obtained, and the predicted injection timing is calculated by the trained timing prediction model. The predicted injection timing is compared with the actual injection timing, and the prediction error is calculated. According to the prediction error, it is determined whether the model needs to be optimized. If the prediction error exceeds the preset threshold, the optimization model is retrained; if the prediction error is within the preset threshold, the predicted injection timing is used as the basis for injection control and the control instruction is output. For example, the trained model predicts that under the conditions of 1800 rpm, 75% load, and 88°C temperature, the diesel injection timing is 16 degrees and the methanol injection timing is 22 degrees. According to the trained model, the initial injection time of diesel and the initial injection time of methanol are calculated. Assume that under a certain condition, the model calculates that the initial injection time of diesel is 14.5 degrees before top dead center and the initial injection time of methanol is 21.5 degrees before top dead center. If the time difference between the initial injection time of diesel and the initial injection time of methanol is greater than the preset value (such as 5 degrees), fine-tuning is required. These two times are used as the input of the Mamdani fuzzy control algorithm.The Mamdani fuzzy control algorithm uses fuzzy rules to infer and output the fine-tuned injection timing. For example, 14.5 degrees and 21.5 degrees are input, and after fine-tuning by the fuzzy control algorithm, the final diesel injection timing is 14.8 degrees and the methanol injection timing is 21.2 degrees. If the time difference is less than the preset value, the initial injection timing is directly used. For example, if the time difference is only 3 degrees, the diesel injection timing is 14.5 degrees and the methanol injection timing is 21.5 degrees. Generate an injector control signal according to the final injection timing. The control signal includes the injection timing and the injection duration. For example, the final diesel injection timing is 14.8 degrees, and a corresponding control signal is generated and sent to the diesel injector. Sending the control signal to the injector is a key step in performing the injection. After receiving the control signal, the diesel injector injects at the predetermined time. Similarly, the methanol injector also injects according to the control signal. The advantage of this method is that through model training and fuzzy control algorithm fine-tuning, the injection timing can be controlled more accurately, thereby improving the combustion efficiency and emission performance of the engine. Especially under complex working conditions such as cold start and sudden load changes, the system can quickly adjust the injection timing to meet the dynamic needs of the engine. Through the comprehensive use of model training and fuzzy control technology, the dual-fuel engine injection timing control system achieves high-precision and adaptive injection control, significantly improving the engine's performance and environmental indicators.

[0018] Step S103, obtaining diesel injection pressure and methanol injection pressure signals, calculating target diesel injection amount and target methanol injection amount using a preset injector flow model, and then determining actual required injection duration data according to a preset pressure difference correction coefficient.

[0019] Obtain a diesel injection pressure signal, filter the diesel injection pressure signal through a low-pass filter, and obtain a filtered diesel pressure signal; perform polynomial fitting based on the filtered diesel pressure signal using the least squares method, establish a polynomial regression model between diesel injection pressure and injection amount, and calculate the target diesel injection amount; determine the first diesel injection duration data based on the target diesel injection amount, in combination with a preset corresponding relationship table between injector pressure difference and injection duration, using a linear interpolation method; if the first diesel injection duration data exceeds the allowable operating range of the injector, an alarm signal is generated, and the first diesel injection duration data is limited within the allowable range; the first diesel injection duration data is sent to the injector control unit; the injector control unit uses a PID control algorithm based on the current engine speed and load condition parameters, with the first diesel injection duration as the target value, and the actual The injection duration is used as a feedback value to adjust the opening and closing time of the injector and output the second diesel injection duration data; the data of the injection pressure sensor and the flow meter are collected to obtain the actual diesel injection pressure and the actual diesel injection amount; the actual diesel injection pressure and the actual diesel injection amount are polynomially fitted using the least squares method to obtain a corrected diesel injection pressure-injection amount polynomial model; if the deviation between the actual diesel injection amount and the target diesel injection amount exceeds a preset threshold, the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the diesel injection pressure-injection amount polynomial model are corrected through multiple iterations to minimize the objective function; according to the corrected diesel injection pressure-injection amount polynomial model, the target diesel injection amount is recalculated, and the corresponding relationship table between the injector pressure difference and the injection duration is updated; the above process is continuously repeated to make the diesel injection amount continuously approach the target value and achieve precise control.

[0020] Exemplarily, the diesel injection pressure signal directly reflects the injection state of the injector, which is crucial for the precise control of the injection amount. For example, assuming that under a certain working condition, the pressure signal measured by the diesel injection pressure sensor is 1500 kPa. Since the sensor signal may contain high-frequency noise, it is filtered by a low-pass filter to remove the high-frequency components to obtain a smooth filtered pressure signal. Assume that the filtered pressure signal is 1480 kPa. According to the filtered diesel pressure signal, a polynomial fitting is performed using the least squares method to establish a polynomial regression model between the diesel injection pressure and the injection amount. Assume that the third-order polynomial model obtained by fitting the experimental data is: injection amount = a * pressure^3 + b * pressure^2 + c * pressure + d. Substitute the filtered pressure signal 1480 kPa into the model to calculate the target diesel injection amount, assuming it is 50 ml. After the target diesel injection amount is determined, the first diesel injection duration data is determined by linear interpolation in combination with the preset corresponding relationship table between the injector pressure difference and the injection duration. Assume that in the corresponding relationship table, when the pressure difference is 100 kPa, the injection duration is 10 milliseconds; when the pressure difference is 200 kPa, the injection duration is 15 milliseconds. Through linear interpolation, when the pressure difference is 1480 kPa, the first diesel injection duration data is calculated to be 12.5 milliseconds. If the data exceeds the allowable working range of the injector (such as 10-20 milliseconds), an alarm signal is generated, and the data is limited to the allowable range, assuming the limit is 20 milliseconds. The first diesel injection duration data is sent to the injector control unit. The injector control unit uses the PID control algorithm according to the current engine speed and load condition parameters, takes the first diesel injection duration as the target value, and the actual injection duration as the feedback value, adjusts the opening and closing time of the injector, and outputs the second diesel injection duration data. Assuming that the current speed is 1800 rpm and the load is 75%, after the PID control algorithm is adjusted, the second diesel injection duration data is 12.8 milliseconds. Collect the data of the injection pressure sensor and the flow meter to obtain the actual diesel injection pressure and the actual diesel injection amount. Assume that the actual measured pressure is 1470 kPa and the injection volume is 49 ml. The least squares method is used to perform polynomial fitting on the actual data to obtain the modified diesel injection pressure-injection volume polynomial model. If the deviation between the actual injection volume and the target injection volume exceeds the preset threshold (such as 1 ml), the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the polynomial model are corrected through multiple iterations to minimize the objective function. The modified model can more accurately reflect the relationship between pressure and injection volume, recalculate the target diesel injection volume, and update the corresponding relationship table between the injector pressure difference and the injection duration. Assuming that the target injection volume calculated by the modified model is 51 ml, the corresponding relationship table is updated, and the injection duration corresponding to the pressure difference of 1480 kPa is adjusted to 13 milliseconds. The above process is continuously repeated to make the diesel injection volume continue to approach the target value and achieve precise control.The advantage of this method is that through the comprehensive application of filtering, polynomial fitting, PID control and gradient descent algorithm, it can effectively reduce noise interference, accurately control the injection amount, and improve the combustion efficiency and emission performance of the engine. Especially under complex working conditions, the system can quickly adjust the injection parameters to meet the dynamic needs of the engine. For example, under cold start conditions, the engine temperature is low and the fuel atomization effect is poor. By accurately controlling the injection amount and injection duration, the combustion condition can be effectively improved and the emission pollutants can be reduced. When the load changes suddenly, the system can respond quickly and adjust the injection parameters to ensure stable operation of the engine. By continuously optimizing the model and parameters, the dual-fuel engine injection control system achieves high-precision and adaptive injection control, significantly improving the overall performance and environmental protection indicators of the engine. This precise control can not only improve fuel economy, but also effectively reduce emissions to meet increasingly stringent environmental protection requirements.

[0021] Step S104, obtaining a methanol injection pressure signal, filtering the methanol injection pressure signal through a low-pass filter to obtain a filtered methanol pressure signal; performing polynomial fitting based on the filtered methanol pressure signal using the least squares method, establishing a polynomial regression model between the methanol injection pressure and the injection amount, and calculating the target methanol injection amount; determining the first methanol injection duration data based on the target methanol injection amount in combination with a preset corresponding relationship table between the injector pressure difference and the injection duration using a linear interpolation method; if the first methanol injection duration data exceeds the allowable operating range of the injector, an alarm signal is generated, and the first methanol injection duration data is limited within the allowable range; the first methanol injection duration data is sent to the injector control unit; the injector control unit uses a PID control algorithm based on the current engine speed and load condition parameters to take the first methanol injection duration as the target value , the actual injection duration is the feedback value, the opening and closing time of the injector is adjusted, and the second methanol injection duration data is output; the data of the injection pressure sensor and the flow meter are collected to obtain the actual methanol injection pressure and the actual methanol injection amount; the actual methanol injection pressure and the actual methanol injection amount are polynomially fitted by the least squares method to obtain a corrected methanol injection pressure-injection amount polynomial model; if the deviation between the actual methanol injection amount and the target methanol injection amount exceeds a preset threshold, the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the methanol injection pressure-injection amount polynomial model are corrected through multiple iterations to minimize the objective function; according to the corrected methanol injection pressure-injection amount polynomial model, the target methanol injection amount is recalculated, and the corresponding relationship table between the injector pressure difference and the injection duration is updated; the above process is continuously repeated to make the methanol injection amount continuously approach the target value and achieve precise control.

[0022] Exemplarily, the methanol injection pressure signal directly reflects the injection state of the injector, which is crucial to the precise control of the methanol injection amount. For example, assuming that under a certain working condition, the pressure signal measured by the methanol injection pressure sensor is 1400 kPa. Since the sensor signal may contain high-frequency noise, it is filtered by a low-pass filter to filter out the high-frequency components to obtain a smooth filtered pressure signal, assuming that the filtered pressure signal is 1390 kPa. According to the filtered methanol pressure signal, a polynomial fitting is performed using the least squares method to establish a polynomial regression model between the methanol injection pressure and the methanol injection amount. Assume that the third-order polynomial model obtained by fitting the experimental data is: methanol injection amount = a * pressure^3 + b* pressure^2 + c * pressure + d. Substitute the filtered pressure signal 1390 kPa into the model to calculate the target methanol injection amount, assuming it is 40 ml. After the target methanol injection amount is determined, the first methanol injection duration data is determined using a linear interpolation method in combination with the preset corresponding relationship table between the injector pressure difference and the injection duration. Assume that in the corresponding relationship table, when the pressure difference is 100 kPa, the injection duration is 8 milliseconds; when the pressure difference is 200 kPa, the injection duration is 12 milliseconds. Through linear interpolation, when the pressure difference is 1390 kPa, the first methanol injection duration data is calculated to be 11 milliseconds. If the data exceeds the allowable working range of the injector (such as 10-16 milliseconds), an alarm signal is generated, and the data is limited to the allowable range, assuming the limit is 16 milliseconds. The first methanol injection duration data is sent to the injector control unit. The injector control unit uses the PID control algorithm according to the current engine speed and load condition parameters, takes the first methanol injection duration as the target value, and the actual injection duration as the feedback value, adjusts the opening and closing time of the injector, and outputs the second methanol injection duration data. Assuming that the current speed is 1800 rpm and the load is 75%, after the PID control algorithm is adjusted, the second methanol injection duration data is 11.6 milliseconds. Collect the data of the injection pressure sensor and the flowmeter to obtain the actual methanol injection pressure and the actual methanol injection amount. Assume that the actual measured pressure is 1390 kPa and the injection amount is 38.5 ml. The least squares method is used to perform polynomial fitting on the actual data to obtain the revised methanol injection pressure-injection amount polynomial model. If the deviation between the actual injection amount and the target injection amount exceeds the preset threshold (such as 1 ml), the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the polynomial model are corrected through multiple iterations to minimize the objective function. The revised model can more accurately reflect the relationship between pressure and injection amount, recalculate the target methanol injection amount, and update the correspondence table between the injector pressure difference and the injection duration. Assuming that the target injection amount calculated by the revised model is 39 ml, the correspondence table is updated, and the injection duration corresponding to the pressure difference of 1390 kPa is adjusted to 12 milliseconds.The above process is continuously repeated to make the methanol injection amount continuously approach the target value and achieve precise control. The advantage of this method is that through the comprehensive application of filtering, polynomial fitting, PID control and gradient descent algorithm, it can effectively reduce noise interference, accurately control the methanol injection amount, and improve the combustion efficiency and emission performance of the engine. Especially under complex working conditions, the system can quickly adjust the injection parameters to meet the dynamic needs of the engine, while maximizing the methanol injection amount while ensuring engine performance and improving the methanol substitution rate. By continuously optimizing the model and parameters, the dual-fuel engine injection control system achieves high-precision and adaptive injection control, significantly improving the engine's methanol substitution rate index and environmental protection index. This precise control can not only improve fuel economy, but also effectively reduce emissions to meet increasingly stringent environmental protection requirements.

[0023] Step S105 , according to the preliminary fusion injection time and the actually required injection duration, combined with the preset injection priority rule, an optimized dual-fuel injection timing control parameter is obtained through an adaptive genetic algorithm.

[0024] Obtain the preliminary fused injection time and the required injection duration, determine the injection amount, injection pressure and injection number of each fuel, and obtain the basic control parameters of dual-fuel injection; obtain the engine speed, load, and temperature operating parameters, query the corresponding priority from the preset dual-fuel injection priority data table, sort the dual-fuel injection control parameters, and obtain the priority-sorted parameter sequence; use the sorted dual-fuel injection control parameters as the initial population of the adaptive genetic algorithm, wherein each individual in the population contains the injection amount, injection pressure, injection number and injection time gene of each fuel; construct a fitness function, which takes the fuel consumption and emission performance of the engine as the optimization target, and iterates the optimization control through selection, crossover and mutation genetic operations. control parameters; evaluate the fitness value of the current population. If the change of the optimal fitness value of multiple consecutive iterations is less than the preset convergence threshold, the current optimal individual is output as the optimal control parameter. Otherwise, continue the iterative optimization until the preset maximum number of iterations is reached; transmit the optimal dual-fuel injection control parameters to the control unit of the injection actuator through the CAN bus; according to the received optimal dual-fuel injection control parameters, the control unit controls the opening time and duration of the injector to achieve accurate dual-fuel injection; the control unit collects the engine speed and load operating parameters in real time. If the operating parameter changes in multiple consecutive working cycles exceed the preset threshold, the re-optimization of the control parameters is triggered, and the current parameters are used as initial values ​​to smoothly transition to new optimized parameters.

[0025] Exemplarily, obtaining the initial fusion injection time and the required injection duration is a key step in the dual-fuel engine control system. For example, under a certain working condition, the initial injection time obtained by the sensor is 10 degrees of crankshaft angle before the top dead center, the diesel injection duration is 5ms, and the methanol injection duration is 4ms. These basic parameters are the starting point for subsequent optimization. Determining the injection amount, injection pressure and number of injections of each fuel is the basis for ensuring efficient operation of the engine. Assume that according to preliminary calculations, the target diesel injection amount is 50mm³, the methanol injection amount is 40mm³, the injection pressures are 1500kPa and 1200kPa respectively, and the number of injections is once per cycle. These parameters directly affect the combustion efficiency and emission performance of the engine. Obtaining the engine's operating parameters such as speed, load, and temperature is a prerequisite for dynamic adjustment. For example, the current engine speed is 3000rpm, the load is 80%, and the temperature is 90°C. From the preset dual-fuel injection priority data table, it is known that under the current working condition, the diesel injection priority is higher than the methanol. Therefore, the injection control parameters are sorted, with the diesel parameters in front and the methanol parameters in the back. The sorted dual-fuel injection control parameters are used as the initial population of the adaptive genetic algorithm. Each individual contains genes such as injection amount, injection pressure, number of injections, and injection timing. For example, an individual may be represented as [50mm³, 1500kPa, 1 time, -10°CA, 40mm³, 1200kPa, 1 time, -10°CA]. A fitness function is constructed, with the fuel consumption and emission performance of the engine as the optimization target. The fitness function may be defined as: fitness = α*(inverse of fuel consumption) + β*(emission performance score), where α and β are weight coefficients. The control parameters are iteratively optimized through genetic operations such as selection, crossover, and mutation. For example, after multiple iterations, an individual shows lower fuel consumption and better emission performance, and its fitness value is significantly improved. The fitness value of the current population is evaluated. If the change in the optimal fitness value for multiple consecutive iterations is less than the preset convergence threshold (such as 0.01), it is considered that the optimal solution has been found. For example, after 20 iterations, the optimal fitness value stabilizes at 0.95 and changes less than 0.01. At this time, the optimal individual is output as the optimal control parameter. The optimal dual-fuel injection control parameters are transmitted to the control unit of the injection actuator via the CAN bus. For example, the transmission parameters are [48mm³, 1480kPa, 1 time, -9°CA, 42mm³, 1210kPa, 1 time, -9°CA]. The control unit accurately controls the opening time and duration of the injector according to these parameters. The control unit collects the engine speed and load operating parameters in real time. If these parameters change more than the preset threshold (such as the speed change exceeds 200rpm) in multiple consecutive working cycles, the control parameters are re-optimized.For example, when the speed suddenly increases from 3000rpm to 3200rpm and the load drops from 80% to 70%, the system uses the current parameters as the initial values, re-optimizes the genetic algorithm, and smoothly transitions to the new optimized parameters to ensure that the engine always maintains efficient operation under different working conditions. In this way, the dual-fuel injection control system not only achieves high-precision injection control, but also can dynamically adjust according to real-time working conditions, significantly improving the performance and environmental protection indicators of the engine.

[0026] Step S106, combining the optimized dual-fuel injection timing control parameters with the feedback information of the original diesel control system, and calculating through a closed-loop PID algorithm to obtain the injection timing dynamic adjustment parameters to compensate for the difference between the actual injection amount and the expected value caused by the mutual interference generated by the two fuel supplies.

[0027] Obtain the optimized dual-fuel injection timing control parameters and the original diesel control system feedback information, and analyze the impact of the mutual interference of the two fuel supplies on the actual injection amount. According to the optimized control parameters and feedback information as the input of the closed-loop PID control algorithm, the dynamic adjustment parameters of the injection timing are obtained. The feedback information of the original diesel control system and the optimized control parameters produce mutual interference, and the two mutual interferences affect the actual injection amount. Limit the dynamic adjustment parameters of the injection timing within the preset safety range. Apply the limited adjustment parameters to the dual-fuel injection system to perform real-time compensation of the injection timing. Continuously obtain the feedback information of the original diesel control system during the dual-fuel injection process, and input the feedback information of the original diesel control system together with the optimized control parameters into the closed-loop PID control algorithm to obtain the injection timing after dynamic adjustment. Compare the deviation between the actual injection amount and the target value, and use the deviation data to adaptively correct the proportional, integral, and differential coefficients of the PID controller. Store the parameters of the PID controller after adaptive correction, and use the stored parameters as the initial parameters of the next dual-fuel injection timing control. Use the dual-fuel injection timing control with the initial parameters to compensate the injection amount through closed-loop adaptive control.

[0028] Exemplarily, obtaining the optimized dual-fuel injection timing control parameters and the original diesel control system feedback information is the key to ensuring the efficient operation of the dual-fuel engine. For example, under a certain working condition, the optimized control parameters are a diesel injection amount of 48mm³, an injection pressure of 1480kPa, an injection time of 9 degrees crankshaft angle before the top dead center, a methanol injection amount of 42mm³, an injection pressure of 1210kPa, and an injection time of 9 degrees crankshaft angle before the top dead center. The original diesel control system feedback information includes the actual injection amount, injection pressure, and injection time, etc. Assuming that the actual diesel injection amount is 47mm³, the injection pressure is 1475kPa, and the injection time is 8.5 degrees crankshaft angle before the top dead center. Analyzing the impact of the mutual interference of the two fuel supplies on the actual injection amount is to ensure the precise control of the injection amount. For example, during the injection of diesel and methanol, due to the slight difference in injection pressure and injection time, the actual injection amount may deviate from the target value. Assuming that the diesel injection pressure is low, the actual injection amount is reduced, and the methanol injection time is early, the actual injection amount may increase. This mutual interference needs to be dynamically adjusted by the closed-loop PID control algorithm. The optimized control parameters and feedback information are used as inputs to the closed-loop PID control algorithm in order to obtain the dynamic adjustment parameters of the injection timing. For example, the PID controller calculates the adjustment of the injection timing and injection pressure based on the deviation between the actual injection amount and the target value. Assuming that the actual diesel injection amount is 1mm³ less than the target value, the PID controller may calculate that the injection timing needs to be advanced by 0.5 crankshaft angle and the injection pressure needs to be increased by 10kPa. The feedback information of the original diesel control system and the optimized control parameters produce mutual interference, and the two mutual interferences affect the actual injection amount. For example, the actual injection amount fed back by the original diesel control system is low, while the optimized control parameters require an increase in the injection amount. This interference may cause the actual injection amount to be unstable. In order to ensure safety, the dynamic adjustment parameters of the injection timing are limited to a preset safety range. For example, the adjustment range of the injection timing is limited to ±1 crankshaft angle, and the adjustment range of the injection pressure is limited to ±50kPa. The limited adjustment parameters are applied to the dual-fuel injection system for real-time compensation of the injection timing. For example, according to the adjustment parameters calculated by the PID controller, the opening time and injection pressure of the injector are adjusted in real time to ensure that the actual injection amount is close to the target value. The continuous acquisition of feedback information from the original diesel control system during the dual-fuel injection process is to continuously optimize the injection timing. For example, the actual injection amount, injection pressure and injection time are collected every 100 milliseconds, and this information is input into the closed-loop PID control algorithm together with the optimized control parameters to obtain the injection timing after dynamic adjustment. Compare the deviation between the actual injection amount and the target value, and use the deviation data to adaptively correct the proportional, integral, and differential coefficients of the PID controller. For example, if the actual injection amount is continuously low, the PID controller may increase the proportional coefficient to increase the response speed; if the actual injection amount fluctuates greatly, the PID controller may increase the integral coefficient to reduce the steady-state error.The storage of the adaptively corrected PID controller parameters is for use in the next dual-fuel injection timing control. For example, the corrected PID parameters are stored in the memory of the control unit as the initial parameters for the next injection control. The dual-fuel injection timing control using the initial parameters performs injection amount compensation through closed-loop adaptive control. For example, under new working conditions, the control unit adjusts the injection timing and injection pressure in real time according to the stored PID parameters to ensure that the actual injection amount is consistent with the target value. This closed-loop adaptive control can not only improve the accuracy of the injection amount, but also dynamically adjust according to the real-time working conditions, significantly improving the performance and environmental protection indicators of the engine. In this way, the dual-fuel injection control system not only achieves high-precision injection control, but also dynamically adjusts according to the real-time working conditions to ensure that the engine always maintains efficient operation under different working conditions. For example, when the engine speed suddenly increases, the control system quickly adjusts the injection parameters to ensure that the injection amount matches the new working conditions, avoiding performance degradation and emission deterioration caused by improper injection amount. This dynamic adjustment mechanism enables the dual-fuel engine to maintain optimal combustion efficiency and emission performance under various complex working conditions.

[0029] Step S107, according to the injection timing dynamic adjustment parameters, correct the current dual-fuel injection timing, predict the dual-fuel injection timing of the next cycle through the pre-established long short-term memory neural network model, and determine whether there is a possibility that the injection time of the two fuels overlaps or is too close. If so, fine-tune the injection start time and duration to obtain the pre-adjusted injection timing.

[0030] The operating parameters and historical operation data of the current dual-fuel engine are obtained, and the operating parameters and historical operation data include the injection start time and duration of each fuel; a pre-trained long short-term memory neural network model is used to predict the injection timing of the dual fuel in the next cycle according to the operating parameters and historical operation data, and the predicted injection start time and duration of each fuel are obtained; it is determined whether the predicted injection times of the two fuels partially overlap or the interval is less than a preset time threshold. If so, the predicted injection start time and duration are fine-tuned according to the preset adjustment rules to obtain the pre-adjusted dual-fuel injection timing; the current The actual dual-fuel injection timing is compared with the pre-adjusted timing and the actual timing, the starting time and duration deviation of each fuel injection timing are calculated, and the deviation is input as a feedback signal into the long short-term memory neural network model for parameter update; when the deviation meets the preset convergence condition, the dual-fuel injection timing predicted by the optimized long short-term memory neural network model is used as the control instruction for the next cycle, and the opening and closing time and duration of the injector are controlled by the engine control unit to achieve precise dual fuel injection; the engine speed and load key parameters are monitored in real time. If they exceed the safety range, the preset emergency injection strategy is adopted.

[0031] Exemplarily, obtaining the operating parameters and historical operation data of the current dual-fuel engine is the basis for optimizing the injection timing. Operating parameters include engine speed, load, temperature, etc., and historical operation data include the injection start time and duration in past cycles. For example, under a certain operating condition, the engine speed is 3000rpm, the load is 80%, and the temperature is 90°C. Historical data show that the start time of diesel injection in the previous cycle is 10 degrees crankshaft angle before the top dead center, the duration is 2ms, and the start time of methanol injection is 8 degrees crankshaft angle before the top dead center, and the duration is 1.5ms. A pre-trained long short-term memory neural network model (LSTM) is used to use these data to predict the injection timing of the next cycle. The LSTM model can predict the injection start time and duration of future cycles by learning patterns and trends in historical data. For example, the model predicts that the start time of diesel injection in the next cycle is 9.5 degrees crankshaft angle before the top dead center, the duration is 2.1ms, and the start time of methanol injection is 7.5 degrees crankshaft angle before the top dead center, and the duration is 1.6ms. The purpose of judging whether the predicted injection time overlaps or the interval is too small is to avoid interference between fuels. For example, if the predicted injection time interval between diesel and methanol is less than the preset threshold of 1ms, it needs to be adjusted. According to the preset rules, fine-tune the injection start time and duration, such as advancing the diesel injection start time by 0.2 degrees crankshaft angle and delaying the methanol injection start time by 0.3 degrees crankshaft angle to ensure that the interval between the two is within a safe range. Obtain the current actual dual-fuel injection timing, compare it with the pre-adjusted timing, and calculate the deviation. For example, the actual diesel injection start time is 9.8 degrees crankshaft angle before top dead center, the duration is 2.0ms, and the methanol injection start time is 7.8 degrees crankshaft angle before top dead center, and the duration is 1.5ms. The calculated deviation is 0.3 degrees crankshaft angle deviation for diesel start time and 0.1ms for duration; 0.3 degrees crankshaft angle deviation for methanol start time and 0.1ms for duration. The deviation is input into the LSTM model as a feedback signal for parameter update to improve the prediction accuracy of the model. Through continuous iterative optimization, the model can more accurately predict the injection timing. For example, after multiple iterations, the injection start time and duration predicted by the model gradually approach the actual value, and the deviation is reduced to the preset convergence condition range. When the deviation meets the convergence condition, the injection timing predicted by the optimized LSTM model is used as the control instruction. The opening and closing time and duration of the injector are controlled by the engine control unit (ECU) to achieve precise injection. For example, the ECU accurately controls the opening and closing of the injector according to the parameters predicted by the model to ensure that the actual injection amount is consistent with the target value. Real-time monitoring of key parameters such as engine speed and load is to ensure safe operation. If the parameters exceed the safe range, the preset emergency injection strategy is adopted to prevent engine damage.For example, when the engine speed suddenly exceeds 3500rpm and the load exceeds 90%, the ECU automatically switches to emergency injection mode, such as switching to single fuel injection mode, or adopting conservative injection advance angle and injection amount to ensure the safe operation of the engine until the parameters return to the normal range. This injection timing control method based on the LSTM model can effectively improve the performance and environmental protection indicators of the dual-fuel engine. By combining the LSTM model, real-time monitoring and emergency strategies, the injection timing control of the dual-fuel engine not only achieves high-precision injection, but also has good adaptability and safety, significantly improving the overall performance and environmental performance of the engine.

[0032] Although the present invention has been described in detail above by general description and specific embodiments, it is obvious to those skilled in the art that some modifications or improvements can be made on the basis of the present invention. Therefore, these modifications or improvements made on the basis of not departing from the spirit of the present invention all belong to the scope of protection claimed by the present invention.

Claims

1. A diesel-methanol dual-fuel engine injection control method, characterized in that: The method comprises: The engine speed signal, load signal, and temperature signal are obtained, and combined with the preset dual-fuel respective injection MAP diagrams, the initial diesel injection timing data and the initial methanol injection timing data under the current working condition are determined; according to the initial diesel injection timing data and the initial methanol injection timing data, the respective injector response times are integrated, and a dual-fuel injection timing model under a unified control framework is constructed to obtain a preliminary fusion injection moment; the diesel injection pressure and methanol injection pressure signals are obtained, and the target diesel injection amount and the target methanol injection amount are calculated using the preset injector flow model, and then the actual required injection duration data is determined according to the preset pressure difference correction coefficient; according to the preliminary fusion injection moment and the actual required injection duration, the preset injection The optimized dual-fuel injection timing control parameters are obtained through an adaptive genetic algorithm based on the injection priority rule; the optimized dual-fuel injection timing control parameters are combined with the feedback information of the original diesel control system, and the injection timing dynamic adjustment parameters are calculated through a closed-loop PID algorithm to compensate for the difference between the actual injection amount and the expected value caused by the mutual interference generated by the two fuel supplies; the current dual-fuel injection timing is corrected according to the injection timing dynamic adjustment parameters, and the dual-fuel injection timing of the next cycle is predicted through a pre-established long short-term memory neural network model to determine whether there is a possibility that the injection time of the two fuels overlaps or is too close. If so, the injection start time and duration are fine-tuned to obtain the pre-adjusted injection timing.

2. The method according to claim 1, characterized in that The method of acquiring the engine speed signal, the load signal, and the temperature signal, and combining the preset dual-fuel respective injection MAP diagrams to determine the initial diesel injection timing data and the initial methanol injection timing data under the current working condition includes: Obtaining engine speed signal, load signal and temperature signal as input parameters; According to the preset dual-fuel injection MAP diagram, the speed, load and temperature parameters are used to look up the table to obtain the theoretical injection timing of diesel and methanol under the current working conditions; Acquire actual injection timing data collected by the injection timing sensor; If the deviation between the actual injection timing and the theoretical injection timing exceeds a preset threshold, it is judged as an injection timing abnormality; In view of the abnormal injection timing, the Kalman filter algorithm is used to filter the actual injection timing data to obtain the corrected injection timing; The corrected injection timing is transmitted to the electronic control unit of the injection system through the CAN bus as the initial injection timing control parameter of the diesel and methanol injectors; The engine speed, load and temperature under historical working conditions are obtained as input, and the corresponding corrected injection timing is used as output. The BP neural network algorithm is used for training to establish a nonlinear mapping relationship between the speed, load and temperature and the injection timing. During engine operation, the speed, load and temperature parameters are acquired in real time and input into the trained BP neural network model to predict and output the target injection timing of diesel and methanol, which are then transmitted to the electronic control unit of the injection system to update the injection timing parameters in real time.

3. The method according to claim 1, characterized in that The method of fusing the initial diesel injection timing data and the initial methanol injection timing data, fusing the respective injector response times, constructing a dual-fuel injection timing model under a unified control framework, and obtaining a preliminary fusion injection time includes: obtaining a target injection timing signal of a diesel injection system and a target injection timing signal of a methanol injection system; Get the diesel injector response time and the methanol injector response time; According to the diesel engine target injection timing signal, the methanol engine target injection timing signal, the diesel injector response time and the methanol injector response time, a dual-fuel engine injection timing model is constructed; Collect dual-fuel engine operation data; The dual-fuel engine operation data is used as training data, and a sequence minimum optimization algorithm is used to train the dual-fuel engine injection timing model to obtain a trained dual-fuel engine injection timing model; According to the trained dual-fuel engine injection timing model, the initial injection time of diesel and the initial injection time of methanol are calculated; If the time difference between the initial injection time of diesel and the initial injection time of methanol is greater than a preset value, the initial injection time of diesel and the initial injection time of methanol are used as inputs of the Mamdani fuzzy control algorithm, and the final diesel injection time and the final methanol injection time are obtained through fine-tuning of the Mamdani fuzzy control algorithm; If the time difference between the initial diesel injection time and the initial methanol injection time is less than a preset value, the initial diesel injection time is used as the final diesel injection time, and the initial methanol injection time is used as the final methanol injection time; generating a diesel injector control signal according to a final diesel injection timing; generating a methanol injector control signal according to a final methanol injection time; sending a diesel injector control signal to the diesel injector; The methanol injector control signal is sent to the methanol injector.

4. The method according to claim 1, characterized in that The method of obtaining the diesel injection pressure and methanol injection pressure signals, calculating the target diesel injection amount and the target methanol injection amount using a preset injector flow model, and then determining the actual required injection duration data according to a preset pressure difference correction coefficient includes: Acquire a diesel injection pressure signal, filter the diesel injection pressure signal through a low-pass filter, and obtain a filtered diesel pressure signal; According to the filtered diesel pressure signal, the least square method is used to perform polynomial fitting, a polynomial regression model between diesel injection pressure and injection amount is established, and the target diesel injection amount is calculated; According to the target diesel injection amount, combined with the preset correspondence table between the injector pressure difference and the injection duration, the first diesel injection duration data is determined by using a linear interpolation method; If the first diesel injection duration data exceeds the allowable operating range of the injector, an alarm signal is generated, and the first diesel injection duration data is limited to the allowable range; sending first diesel injection duration data to an injector control unit; The injector control unit uses a PID control algorithm according to the current engine speed and load condition parameters, takes the first diesel injection duration as the target value and the actual injection duration as the feedback value, adjusts the opening and closing time of the injector, and outputs the second diesel injection duration data; Collect data from the injection pressure sensor and flow meter to obtain the actual diesel injection pressure and actual diesel injection amount; The actual diesel injection pressure and the actual diesel injection amount are fitted with polynomials by using the least square method to obtain a modified diesel injection pressure-injection amount polynomial model. If the deviation between the actual diesel injection amount and the target diesel injection amount exceeds a preset threshold, the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the diesel injection pressure-injection amount polynomial model are corrected through multiple iterations to minimize the objective function; According to the modified diesel injection pressure-injection quantity polynomial model, the target diesel injection quantity is recalculated, and the corresponding relationship table between the injector pressure difference and the injection duration is updated; The above process is repeated continuously, so that the diesel injection amount continues to approach the target value and precise control is achieved.

5. The method according to claim 4, characterized in that Also includes: Acquire a methanol injection pressure signal, and filter the methanol injection pressure signal through a low-pass filter to obtain a filtered methanol pressure signal; According to the filtered methanol pressure signal, a polynomial fitting is performed using the least square method to establish a polynomial regression model between the methanol injection pressure and the injection amount, and the target methanol injection amount is calculated. According to the target methanol injection amount, combined with a preset correspondence table between the injector pressure difference and the injection duration, the first methanol injection duration data is determined by using a linear interpolation method; If the first methanol injection duration data exceeds the allowable operating range of the injector, an alarm signal is generated, and the first methanol injection duration data is limited to the allowable range; sending first methanol injection duration data to an injector control unit; The injector control unit uses a PID control algorithm according to the current engine speed and load condition parameters, takes the first methanol injection duration as the target value and the actual injection duration as the feedback value, adjusts the opening and closing time of the injector, and outputs the second methanol injection duration data; Collect data from the injection pressure sensor and flow meter to obtain actual methanol injection pressure and actual methanol injection amount; The actual methanol injection pressure and the actual methanol injection amount are fitted with polynomials by using the least square method to obtain a modified methanol injection pressure-injection amount polynomial model. If the deviation between the actual methanol injection amount and the target methanol injection amount exceeds a preset threshold, the deviation value is substituted into the objective function of the gradient descent algorithm, and the coefficients of the methanol injection pressure-injection amount polynomial model are corrected through multiple iterations to minimize the objective function; According to the modified methanol injection pressure-injection amount polynomial model, the target methanol injection amount is recalculated, and the corresponding relationship table between the injector pressure difference and the injection duration is updated; The above process is repeated continuously, so that the methanol injection amount continues to approach the target value and precise control is achieved.

6. The method according to claim 1, characterized in that The optimized dual fuel injection timing control parameters are obtained by an adaptive genetic algorithm based on the preliminary fusion injection time and the actual required injection duration, combined with the preset injection priority rule, including: Obtain the initial fusion injection time and the required injection duration, determine the injection amount, injection pressure and injection number of each fuel, and obtain the basic control parameters of dual fuel injection; Obtaining the speed, load and temperature operating parameters of the engine, querying the corresponding priority from a preset dual-fuel injection priority data table, sorting the dual-fuel injection control parameters, and obtaining a priority-sorted parameter sequence; The sorted dual-fuel injection control parameters are used as the initial population of the adaptive genetic algorithm, wherein each individual in the population contains the injection amount, injection pressure, injection number and injection time gene of each fuel; Construct a fitness function that takes the fuel consumption and emission performance of the engine as the optimization target and iteratively optimizes the control parameters through selection, crossover and mutation genetic operations; Evaluate the fitness value of the current population. If the change in the optimal fitness value of multiple consecutive iterations is less than the preset convergence threshold, the current optimal individual is output as the optimal control parameter. Otherwise, continue iterative optimization until the preset maximum number of iterations is reached. Transmitting optimal dual fuel injection control parameters to a control unit of an injection actuator via a CAN bus; According to the received optimal dual-fuel injection control parameters, the control unit controls the opening time and duration of the injector to achieve accurate dual-fuel injection; The control unit collects the engine's speed and load operating parameters in real time. If the operating parameter changes exceed the preset threshold within multiple consecutive working cycles, it triggers the re-optimization of the control parameters, uses the current parameters as the initial values, and smoothly transitions to the new optimized parameters.

7. The method according to claim 6, characterized in that The optimized dual-fuel injection timing control parameters are combined with the feedback information of the original diesel control system, and the injection timing dynamic adjustment parameters are calculated by a closed-loop PID algorithm to compensate for the difference between the actual injection amount and the expected value caused by the mutual interference generated by the two fuel supplies, including: Obtain optimized dual-fuel injection timing control parameters and original diesel control system feedback information, and analyze the impact of the mutual interference of the two fuel supplies on the actual injection quantity; The optimized control parameters and feedback information are used as inputs of the closed-loop PID control algorithm to obtain the dynamic adjustment parameters of the injection timing; The original diesel control system feedback information and the optimized control parameters produce mutual interference, and the two mutual interferences affect the actual injection amount; Limit the dynamic adjustment parameters of the injection timing to a preset safety range; Applying the limited post-adjustment parameters to the dual fuel injection system to perform real-time compensation of injection timing; Continuously obtain the original diesel control system feedback information during the dual-fuel injection process, and input the original diesel control system feedback information and the optimized control parameters into the closed-loop PID control algorithm to obtain the injection timing after dynamic adjustment; Compare the deviation between the actual injection amount and the target value, and use the deviation data to adaptively correct the proportional, integral and differential coefficients of the PID controller; The PID controller parameters after the adaptive correction are stored, and the stored parameters are used as initial parameters for the next dual fuel injection timing control; The dual-fuel injection timing control with initial parameters is adopted to compensate the injection quantity through closed-loop adaptive control.

8. The method according to claim 1, characterized in that The method of dynamically adjusting the parameters according to the injection timing to correct the current dual-fuel injection timing, predicting the dual-fuel injection timing of the next cycle through the pre-established long short-term memory neural network model, and judging whether there is a possibility that the injection times of the two fuels overlap or are too close, and if so, fine-tuning the injection start time and duration to obtain the pre-adjusted injection timing, includes: Acquire the current operating parameters and historical operating data of the dual-fuel engine, wherein the operating parameters and historical operating data include the injection start time and duration of each fuel; Using a pre-trained long short-term memory neural network model, based on operating parameters and historical operating data, the injection timing of the dual fuels in the next cycle is predicted to obtain the predicted injection start time and duration of each fuel; Determine whether the predicted injection times of the two fuels partially overlap or the interval is less than a preset time threshold, and if so, fine-tune the predicted injection start time and duration according to a preset adjustment rule to obtain a pre-adjusted dual-fuel injection timing; Obtain the currently actually executed dual fuel injection timing, compare the pre-adjusted timing with the actual timing, calculate the start time and duration deviation of each fuel injection timing, and input the deviation as a feedback signal into the long short-term memory neural network model for parameter update; When the deviation meets the preset convergence condition, the dual-fuel injection timing predicted by the optimized long short-term memory neural network model is used as the control instruction for the next cycle, and the opening and closing time and duration of the injector are controlled by the engine control unit to achieve accurate dual-fuel injection; The engine speed and load key parameters are monitored in real time. If they exceed the safe range, the preset emergency injection strategy is adopted.

Images