Marine ammonia gas engine ammonia gas substitution rate adjusting system based on working condition self-adaption

Abstract

According to the marine ammonia engine ammonia substitution rate adjusting system based on working condition self-adaption, whether the current working condition is the steady-state working condition or the transient working condition is judged through the rotating speed change rate and the load change rate, and then filtered multi-dimensional parameters are obtained; and a corrected transient ammonia gas replacement rate interval range and a corrected steady-state ammonia gas replacement rate interval range are obtained according to the filtered multi-dimensional parameters, then the ammonia gas valve opening degree and the diesel oil injection amount are obtained, an executing mechanism executes according to the ammonia gas valve opening degree and the diesel oil injection amount, and the problem that an existing adjusting method cannot dynamically adapt to transient load changes is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of engine regulation systems, and in particular to an ammonia replacement rate regulation system for a marine ammonia engine based on working condition self-adaptation. Background Art

[0002] With increasingly stringent environmental regulations, emissions from traditional diesel engines are becoming increasingly prominent, leading to their gradual phase-out in favor of engines powered by ammonia and diesel. While ammonia, as a carbon-free fuel, offers significant environmental advantages, its combustion characteristics differ significantly from those of diesel, making it difficult to achieve efficient combustion and low emissions in dual-fuel mode.

[0003] To solve the above problems, existing technologies mostly rely on fixed parameters or simple control strategies to adjust the ammonia replacement rate to switch fuel modes, such as only fixing parameters such as temperature and pressure thresholds, or using conventional PID and LSTM predictions to formulate simple control strategies. However, the above adjustment methods cannot dynamically adapt to transient load changes (such as ship acceleration and steering). Summary of the Invention

[0004] The technical problem to be solved by the technical solution of the present invention is that the existing regulation method cannot dynamically adapt to transient load changes (such as ship acceleration and steering).

[0005] The technical solution of the present invention provides an ammonia replacement rate adjustment system for a marine ammonia engine based on operating condition adaptation, comprising:

[0006] The sensor module is used to collect the original signals corresponding to the multi-dimensional parameters, including speed n, load L, intake pressure P in , coolant temperature T cool , exhaust temperature T exh 、NO x Concentration C NOx , ammonia escape rate R NH3 , atmospheric pressure P atm 、Ambient temperature T atm , exhaust temperature change rate

[0007] The signal conditioning circuit is used to obtain the original signal corresponding to the multi-dimensional parameters transmitted by the sensor module, trigger the ECU to start the control unit to read the original signal, and execute the filtering algorithm in parallel through the FPGA to perform matrix operations and construct the state vector;

[0008] Calculate the speed change rate and load change rate according to the preset timing time, determine whether the current working condition is a steady-state condition or a transient condition, and dynamically adjust the Kalman filter parameters according to the steady-state condition and the transient condition to obtain the filtered multi-dimensional parameters;

[0009] Call the basic MAP table calibrated by bench test to obtain the speed n table value, load L table value and basic ammonia replacement rate S in the basic MAP table NH3 , combined with the coolant temperature T cool and atmospheric pressure P atm , calculate the steady-state ammonia replacement rate S* NH3稳 ;

[0010] For the multi-dimensional parameters after filtering under steady-state working conditions, calculate the load deviation between the actual load and the set load, and calculate the actual NO x Concentration and limit NO x NO concentrations between x Concentration deviation value, according to load deviation value and NO x Concentration deviation value, load deviation value fuzzy processing value and NO calculated based on Gaussian membership function x The concentration deviation value is fuzzy processed according to the rule base extracted based on the test data of the marine engine bench, combined with the load deviation value fuzzy processed value and NO x The concentration deviation value is fuzzified to obtain the Kp adjustment fuzzy value corresponding to the current working condition. The center of gravity method is used to defuzzify the Kp adjustment fuzzy value to obtain the final output. According to the final output, the PID parameter PID is updated in combination with the speed change rate correction function. out ;

[0011] For the multi-dimensional parameters after transient working condition filtering, the predicted load is calculated based on the actual load and the LSTM prediction model. The corrected predicted load is obtained and the feedforward compensation is calculated, combined with the basic ammonia replacement rate S NH3 Calculate the transient ammonia replacement rate S* NH3瞬 ;

[0012] According to PID parameters PID out Determine the ammonia escape rate R NH3 Constraint threshold and NO x Concentration C NOx Constraint threshold, combined with the NO in the multidimensional parameters after steady-state filtering x Concentration C NOx and ammonia escape rate R NH3 , and the steady-state ammonia replacement rate S* NH3稳 Determine the corrected steady-state ammonia replacement rate S* NH3稳 Interval range, combined with transient working condition filtering multi-dimensional parameter NO x Concentration C NOx and ammonia escape rate R NH3 , and the transient ammonia replacement rate S* NH3瞬 Determine the corrected transient ammonia replacement rate S* NH3瞬 interval range;

[0013] According to the corrected steady-state ammonia replacement rate S* NH3稳 Interval range and corrected transient ammonia replacement rate S* NH3瞬 The range of the interval is calculated based on the Bernoulli equation and the saturation function to calculate the ammonia valve opening, and the coolant temperature compensation term is introduced in combination with the coolant temperature T cool Calculate diesel injection quantity;

[0014] The actuator is used to control the execution according to the opening of the ammonia valve and the diesel injection amount.

[0015] Preferably, the control unit integrates an ARM Cortex-A53 processor.

[0016] Preferably, the actuator includes an ammonia flow control valve that is actuated according to the ammonia valve opening and a diesel high-pressure common rail injector that is actuated according to the diesel injection amount.

[0017] Preferably, the data module is based on FPGA and performs real-time data storage and learning.

[0018] Preferably, a sliding window is used to sample the historical speed and load of 10 cycles to construct the feature vector X t , using weight matrix for training, improving the attention of load fluctuation, building LSTM prediction model, combining the feature vector X t Forecast the load to get the predicted load

[0019] Preferably, the ammonia valve opening and the diesel injection amount are converted into PWM signal parameters and transmitted to the actuator.

[0020] Preferably, the sensor module includes a speed sensor, a NO x Sensor, ammonia concentration sensor.

[0021] Preferably, the sensor module transmits original signals corresponding to the multi-dimensional parameters via a CAN bus.

[0022] The technical solution of the present invention proposes an ammonia replacement rate regulation system for a marine ammonia engine based on working condition adaptation. The system determines whether the current working condition is a steady-state condition or a transient condition by using the speed change rate and the load change rate, thereby obtaining filtered multidimensional parameters. Based on the filtered multidimensional parameters, a corrected transient ammonia replacement rate interval and a corrected steady-state ammonia replacement rate interval are obtained, thereby obtaining the ammonia valve opening and the diesel injection amount. The actuator is executed according to the ammonia valve opening and the diesel injection amount, thereby solving the problem that the existing regulation method cannot dynamically adapt to transient load changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1A schematic flow chart of an ammonia replacement rate adjustment system for a marine ammonia engine based on operating condition adaptation provided by an embodiment of the present invention;

[0024] Figure 2 Schematic diagram of the logic chain of the ammonia replacement rate adjustment system for a marine ammonia engine based on operating condition adaptation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] Below in conjunction with specific embodiment, further set forth the present invention.Should be understood that these embodiments are only used to illustrate the present invention and are not used in limiting the scope of the present invention.In addition, should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall equally within the scope limited by the appended claims of the application.

[0026] like Figure 1 As shown, an embodiment of the present invention provides an ammonia replacement rate adjustment system for a marine ammonia engine based on working condition adaptation. This system realizes adaptive adjustment of the ammonia replacement rate through a closed-loop logic of "hardware acquisition - software processing - execution feedback - learning optimization". The hardware and software are tightly coupled, and the hardware list is as follows:

[0027] Sensor module: collects 12-dimensional parameters (speed n, load L, intake pressure P in , coolant temperature T cool , exhaust temperature T exh 、NO x Concentration C NOx , ammonia escape rate R NH3 , atmospheric pressure P atm 、Ambient temperature T atm , speed change rate Δn 50ms , load change rate ΔL 50ms , exhaust temperature change rate ).

[0028] Control unit: Integrates an ARM Cortex-A53 processor and runs a multi-scale PID+LSTM+reinforcement learning algorithm.

[0029] Actuators: Ammonia flow control valve (accuracy ±1% opening), diesel high-pressure common rail injector (response time <5ms).

[0030] Data module: FPGA-based real-time data storage and learning unit, supporting 10ms-level data caching.

[0031] The specific logic chain is as follows Figure 2 As shown, the complete logic from data collection to replacement rate adjustment is as follows:

[0032] Step 1: Sensor data acquisition (hardware-software interface).

[0033] Hardware action: sensor module (including speed sensor, NO x Sensors, ammonia concentration sensors, etc.) collect 12-dimensional parameters in real time (see hardware list), transmit the original signal to the signal conditioning circuit (hardware) of the control unit through the CAN bus, and complete A / D conversion (such as converting 4-20mA current signal into digital quantity).

[0034] Software triggering: Embedded software in the control unit (ECU, hardware) (e.g., a real-time task based on an RTOS) triggers data reading every 10ms and stores it in a buffer (corresponding to the hardware's data register).

[0035] The raw data collected by the hardware is directly used as the input for software processing. For example, after the pulse signal of the speed sensor is counted by the hardware, the software directly reads the "speed value n"; NO x After the analog signal of the sensor is filtered by hardware, the software reads “CNO x Concentration value".

[0036] Step 2: Data preprocessing (Kalman filtering, software-hardware collaboration).

[0037] Hardware support: The FPGA module (hardware) of the control unit executes the matrix operations of the filtering algorithm in parallel (such as K k calculation), improve processing speed (ensure completion within 10ms).

[0038] Software processing: read the original data from the buffer and construct the state vector x=[n,L,CNO x ,RNH3].

[0039] According to the hardware timer, calculate the speed change rate Δn of the difference within 50ms 50ms and load change rate ΔL 50ms , according to the speed change rate Δn 50ms and load change rate ΔL 50ms Determine the current working condition (i.e., determine whether the speed is constant or variable), if |Δn 50ms |<50rpm / s and |ΔL 50ms If |<5%, it is determined to be a steady-state operating condition, otherwise it is determined to be a transient operating condition. The Kalman filter parameters are dynamically adjusted according to the steady-state and transient operating conditions:

[0040]

[0041]

[0042]

[0043] Output filtered data (Stored in the hardware's shared memory for subsequent algorithm calls).

[0044] The result of the working condition classification directly determines the subsequent control algorithm call (fuzzy PID for steady state and LSTM for transient state). The two share the same hardware resources (CPU core of the control unit). The filtered "clean data" is directly used as the input of the control algorithm. For example, the filtered "load L" is used to judge steady state / transient state, and the filtered "CNO x ” for emission constraint correction.

[0045] Step 3: Steady-state operating condition control (fuzzy PID, software-innovation combination).

[0046] Hardware support: The CPU of the control unit runs the fuzzy PID algorithm and calls the "basic MAP table" stored in the hardware (pre-stored in the Flash chip and calibrated through bench tests).

[0047] Software processing:

[0048] Calculate the steady-state ammonia replacement rate: S* NH3稳 =f MAP (n,L)×α(T cool )×β(P atm ).

[0049] (Definition of ammonia replacement rate: Range 0-80% (limited by emission constraints). (α and β are correction coefficients used by the software to fit the hardware temperature / pressure sensor data).

[0050] Steady-state substitution rate S* NH3稳 Directly input to step 5 (emission constraint correction). All parameters in the calculation process come from the calibration data collected by the hardware or stored in the hardware.

[0051] 1. Read the filtered "Load L" and "CNO x ", calculate the deviation ΔL=Lactual-Lsetting, ΔCNO x =CNO x Actual-CNO x limit.

[0052] 2. Fuzzy processing: The fuzzy set is divided into NB (negative large), NM (negative medium), ZE (zero), PM (positive medium), and PB (positive large), using the Gaussian membership function:

[0053]

[0054] Where x is the input variable (such as ΔL, ΔCNO x ), μA (ΔL) is the degree to which the variable x belongs to the fuzzy set A (such as “negative large NB” and “zero ZE”) (the value range is [0,1]).

[0055] The center values ​​are (-15%, -10%, 0, +10%, +15%), ΔC NOx The same is true for fuzzy sets, where the central values ​​correspond to (-50ppm, -20ppm, 0, +20ppm, +50ppm).

[0056] 3. Match 25 fuzzy rules (software logic, rules stored in hardware ROM), the contents are as follows:

[0057] The rule base is extracted based on marine engine bench test data and contains 25 rules:

[0058] If ΔL is negative (NB) and ΔCNOx is negative (NB), then ΔKp is positive (PB). This means that when the actual load is much lower than the set value and NO x When emissions are far below the limit, a substantial increase in Kp can rapidly increase the ammonia replacement rate, allowing the engine to quickly increase power to adapt to load requirements, while fully utilizing the clean combustion advantages of ammonia to further reduce emissions.

[0059] If ΔL is negative (NB) and ΔCNOx is negative (NM), then ΔKp is positive (PM). x The emission situation is relatively good. Moderately increasing Kp can not only increase the ammonia replacement rate and supplement power, but also prevent emissions from exceeding the standard due to a too rapid increase in combustion intensity.

[0060] If ΔL is negative (NB) and ΔCNOx is zero (ZE), then ΔKp is positive (PM). In this case, increasing Kp can increase the proportion of ammonia participating in combustion, compensate for insufficient load, and maintain NO x Emissions are stable to avoid negative impacts on emission indicators.

[0061] If ΔL is negative (NB) and ΔCNOx is positive (PM), then ΔKp is zero (ZE). x Emissions have shown an upward trend. At this time, it is not appropriate to increase Kp to increase the ammonia replacement rate to avoid NO x Emissions further deteriorate, keep Kp unchanged, maintain the current combustion state, and seek a balance between load and emissions.

[0062] If ΔL is negative (NB) and ΔCNOx is positive (PB), then ΔKp is negative (NM). xIn the event of excessive emissions, reduce Kp and lower the ammonia replacement rate to prioritize emission compliance and prevent the aggravation of emission problems. At the same time, the load can be gradually adjusted through other control measures.

[0063] If ΔL is negative medium (NM) and ΔCNOx is negative large (NB), then ΔKp is positive medium (PM). x When emissions are far below the limit, moderately increasing Kp will increase the ammonia replacement rate, helping the engine increase power without adversely affecting emissions.

[0064] If ΔL is negative (NM) and ΔCNOx is negative (NM), then ΔKp is positive (PM). In this case, increasing Kp appropriately will promote ammonia combustion, helping to improve engine power while maintaining good operating conditions within acceptable emissions.

[0065] If ΔL is negative (NM) and ΔCNOx is zero (ZE), then ΔKp is zero (ZE). x Emissions are all within an acceptable range, Kp remains unchanged, the existing combustion mode is maintained, stable engine operation is ensured, and fluctuations caused by unnecessary parameter adjustments are avoided.

[0066] If ΔL is negative (NM) and ΔCNOx is positive (PM), then ΔKp is negative (NM). x Emissions have increased, reduce Kp, reduce the ammonia combustion ratio, control NO x Emissions, maintain the stability and environmental protection of engine operation.

[0067] If ΔL is negative (NM) and ΔCNOx is positive (PB), then ΔKp is negative (NB). x In the case of serious emissions exceeding the standard, Kp is greatly reduced and the ammonia replacement rate is significantly reduced. The problem of emissions exceeding the standard is first solved, and the load is then adjusted and optimized.

[0068] If ΔL is zero (ZE) and ΔCNOx is negative (NB), then ΔKp is positive (PM). x When emissions are far below the limit, increase Kp and improve the ammonia replacement rate. Under the premise of ensuring good emissions, further optimize the engine combustion efficiency and improve energy utilization.

[0069] If ΔL is zero (ZE) and ΔCNOx is negative (NM), then ΔKp is zero (ZE). x Emissions are all within the ideal range, keeping Kp unchanged and maintaining stable and efficient engine operation without adjusting combustion parameters.

[0070] If ΔL is zero (ZE) and ΔCNOx is zero (ZE), then ΔKp is zero (ZE). The engine is operating stably, with load and emissions meeting standards. Maintaining the current Kp value ensures the engine remains in optimal operating condition and avoids unnecessary interference caused by parameter changes.

[0071] If ΔL is zero (ZE) and ΔCNOx is positive (PM), then ΔKp is negative (NM). x Emissions increase, reduce Kp, reduce ammonia combustion intensity, control NO x Emissions, ensuring that the engine meets environmental protection requirements while running stably.

[0072] If ΔL is zero (ZE) and ΔCNOx is positive (PB), then ΔKp is negative (NB). x Seriously exceeding the standard, significantly reducing Kp, lowering the ammonia replacement rate, and making every effort to solve the emission problem to ensure that engine emissions meet the standards.

[0073] If ΔL is positive (PM) and ΔCNOx is negative (NB), then ΔKp is zero (ZE). x Emissions are far below the limit. At this time, keep Kp unchanged and maintain the current combustion state to ensure that the engine maintains a good emission level while running efficiently.

[0074] If ΔL is positive (PM) and ΔCNOx is negative (NM), then ΔKp is negative (NM). x Emissions also increase to a certain extent, reduce Kp, reduce the ammonia combustion ratio, balance the load and emissions, and prevent NO x Emissions exceed the standard to ensure stable operation of the engine.

[0075] If ΔL is positive (PM) and ΔCNOx is zero (ZE), then ΔKp is negative (NM). x When emissions increase, reduce Kp, appropriately lower the ammonia replacement rate, maintain stable emissions, and ensure the environmental performance of the engine under different loads.

[0076] If ΔL is positive (PM) and ΔCNOx is positive (PM), then ΔKp is negative (NB). x Emissions are high, Kp is greatly reduced, the combustion intensity of ammonia is significantly reduced, and NO is strictly controlled. x Emissions, so that the engine operation returns to a reasonable emission range.

[0077] If ΔL is positive (PM) and ΔCNOx is positive (PB), then ΔKp is negative (NB). x When emissions are seriously exceeded, Kp will continue to be significantly reduced, with a focus on lowering the ammonia replacement rate, to fully resolve the issue of emissions exceeding the standard and ensure that the engine complies with environmental regulations.

[0078] If ΔL is positive (PB) and ΔCNOx is negative (NB), then ΔKp is negative (NM). x Emissions are far below the limit, but in order to avoid potential emission problems caused by continued load increase, Kp is reduced and the ammonia replacement rate is moderately lowered to maintain the stability and sustainability of engine operation.

[0079] If ΔL is positive (PB) and ΔCNOx is negative (NM), then ΔKp is negative (NB). x Emissions have been on an upward trend. By significantly reducing Kp and lowering the ammonia combustion ratio, we can primarily control the emission risks caused by excessive load and ensure that the engine's emissions are compliant under high load.

[0080] If ΔL is positive (PB) and ΔCNOx is zero (ZE), then ΔKp is negative (NB). x Emissions exceed the standard as the load increases, which greatly reduces Kp and the proportion of ammonia participating in combustion, ensuring emission stability when the engine is running at high load.

[0081] If ΔL is positive (PB) and ΔCNOx is positive (PM), then ΔKp is negative (NB). x The emissions are relatively high, Kp is continuously and significantly reduced, ammonia combustion is strictly controlled, and the focus is on solving the problem of excessive emissions to ensure the environmental performance of the engine under high load conditions.

[0082] If ΔL is positive (PB) and ΔCNOx is positive (PB), then ΔKp is negative (NB). x In extreme cases where emissions are seriously exceeded, every effort will be made to significantly reduce Kp, minimize the ammonia replacement rate, give priority to solving the urgent problem of emissions seriously exceeding the standard, restore the engine operating status to normal, and meet the dual requirements of environmental protection and power output.

[0083] Defuzzification: The final output is calculated using the centroid method:

[0084]

[0085] Among them, w i Discrete weights represent the corresponding positions (x ior x), is the "weighting factor" for the center of gravity calculation.

[0086] v i Output the central value of the fuzzy set for the i-th rule.

[0087] Get ΔKp and update PID parameter Kp ms (Millisecond level Kp).

[0088] in, is the basic proportional coefficient, f(Δn 50ms ) is the speed change rate correction function:

[0089]

[0090] (When the speed mutation rate is greater than 100rpm / s, the Kp correction amount is further amplified to improve the response speed).

[0091] Step 4: Transient operating condition control (LSTM prediction, software-innovation combination).

[0092] Hardware support: The control unit's NPU (neural network processing unit) accelerates LSTM calculations, and the data storage module (DDR memory) caches the "n" and "L" data of the past 10 cycles.

[0093] Software processing:

[0094] 1. Read historical data from DDR and construct feature vector X t =[n t-9 ...n t ,L t-9 ...L t ](Software sliding window sampling).

[0095] 2. Enhanced attention mechanism:

[0096]

[0097] Among them, the weight matrix W Q / W K / W V Through PPO reinforcement learning training, the model's attention to load fluctuations increased by 40%.

[0098] LSTM prediction load (NPU acceleration, software outputs predicted values).

[0099]

[0100] 4. Multi-step error compensation:

[0101]

[0102] 5. Calculate the feedforward compensation (k ff = 0.8, hardware bench calibration).

[0103] 6. Synthetic transient ammonia substitution rate S* NH3瞬 =S NH3 +ΔS ff (S NH3 is the basic MAP table value, same as steady state).

[0104] Logical connection: transient replacement rate S* NH3瞬 The steady-state results are transferred to step 5 to ensure the logical consistency of the replacement rate calculations for the two working conditions.

[0105] Step 5: Emission constraint correction (software and hardware collaboration, innovation).

[0106] Hardware input: Software reads the R of the ammonia escape sensor NH3 Value and NO x Sensor CNO x Value (hardware data after Kalman filtering).

[0107] Software processing:

[0108] 1. Dynamically calculate constraint thresholds:

[0109]

[0110] Among them, k1 / k2 / k3 are fitted through bench tests (the initial values ​​can be set to k1=5, k2=3, k3=0.1). When the PID output increases, the NH3 escape threshold automatically decreases.

[0111] 2. Modified replacement rate:

[0112] like

[0113] like

[0114] Logical connection:

[0115] Corrected steady-state ammonia replacement rate S* NH3稳 Interval range and transient ammonia replacement rate S* NH3瞬 The interval range is directly used as the control instruction of the executive agency to ensure the real-time linkage between emission constraints and substitution rate adjustment.

[0116] Step 6: Actuator control (software-hardware closed loop).

[0117] Hardware Action:

[0118] The actuator (ammonia valve, diesel injector) receives the PWM signal (hardware output) from the control unit and adjusts the opening / injection amount.

[0119] Software processing:

[0120] Calculate the ammonia valve opening based on the Bernoulli equation and saturation function:

[0121]

[0122] The coolant temperature compensation term is introduced to calculate the diesel injection amount:

[0123]

[0124] Low temperature start (T cool <40℃), k co It automatically increases to 0.8, increasing the diesel injection amount to assist ammonia vaporization.

[0125] The software converts the opening and injection volume into PWM signal parameters (duty cycle) and outputs them to the actuator through the hardware interface.

[0126] Step 7: Data storage and learning (combination of software and hardware, innovation).

[0127] Hardware support: The data storage module (SSD) stores the "input parameter-replacement rate instruction-emission result" triple (one every 10ms, hardware-timed storage).

[0128] Software processing: Data screening is triggered every 100 cycles (1s), valid samples are retained (software logic, sensor fault data is eliminated), and the reinforcement learning algorithm (PPO) is called every week.

[0129] Optimize the fuzzy rule base weights (stored in hardware ROM), LSTM attention weights W Q / W K / W V (Update to NPU weight register), output optimized parameters, overwrite the calibration values ​​stored in the original hardware.

[0130] Logical connection:

[0131] The learning results directly update the core parameters of the control algorithm, forming a closed loop of "run-learn-optimize" to ensure the continuous improvement of the system's adaptive capabilities.

Claims

1. A marine ammonia engine ammonia replacement rate adjustment system based on working condition adaptation, characterized in that: include: The sensor module is used to collect the original signals corresponding to the multi-dimensional parameters, including speed n, load L, coolant temperature T cool 、NO x Concentration C Nox , ammonia escape rate R NH3 ; The signal conditioning circuit is used to obtain the original signal corresponding to the multi-dimensional parameters transmitted by the sensor module, trigger the ECU to start the control unit to read the original signal, and execute the filtering algorithm in parallel through the FPGA to perform matrix operations and construct the state vector; Calculate the speed change rate and load change rate according to the preset timing time, determine whether the current working condition is a steady-state condition or a transient condition, and dynamically adjust the Kalman filter parameters according to the steady-state condition and the transient condition to obtain the filtered multi-dimensional parameters; Call the basic MAP table calibrated by bench test to obtain the speed n table value, load L table value and basic ammonia replacement rate S in the basic MAP table NH3 , combined with the coolant temperature T cool and atmospheric pressure P atm , calculate the steady-state ammonia replacement rate S* NH3稳 ; For the multi-dimensional parameters after filtering under steady-state working conditions, calculate the load deviation between the actual load and the set load, and calculate the actual NO x Concentration and limit NO x NO concentrations between x Concentration deviation value, according to load deviation value and NO x Concentration deviation value, load deviation value fuzzy processing value and NO calculated based on Gaussian membership function x The concentration deviation value is fuzzy processed according to the rule base extracted based on the test data of the marine engine bench, combined with the load deviation value fuzzy processed value and NO x The concentration deviation value is fuzzified to obtain the Kp adjustment fuzzy value corresponding to the current working condition. The center of gravity method is used to defuzzify the Kp adjustment fuzzy value to obtain the final output. According to the final output, the PID parameter PID is updated in combination with the speed change rate correction function. out ; For the multi-dimensional parameters after transient working condition filtering, the predicted load is calculated based on the actual load and the LSTM prediction model. The corrected predicted load is obtained and the feedforward compensation is calculated, combined with the basic ammonia replacement rate S NH3 Calculate the transient ammonia replacement rate S* NH3瞬 ; According to PID parameters PID out Determine the ammonia escape rate R NH3 Constraint threshold and NO x Concentration C NOx Constraint threshold, combined with the NO in the multidimensional parameters after steady-state filtering x Concentration C NOx and ammonia escape rate R NH3 , and the steady-state ammonia replacement rate S* NH3稳 Determine the corrected steady-state ammonia replacement rate S* NH3稳 Interval range, combined with transient working condition filtering multi-dimensional parameter NO x Concentration C NOx and ammonia escape rate R NH3 , and the transient ammonia replacement rate S* NH3瞬 Determine the corrected transient ammonia replacement rate S* NH3瞬 interval range; According to the corrected steady-state ammonia replacement rate S* NH3稳 Interval range and corrected transient ammonia replacement rate S* NH3瞬 The range of the interval is calculated based on the Bernoulli equation and the saturation function to calculate the ammonia valve opening, and the coolant temperature compensation term is introduced in combination with the coolant temperature T cool Calculate diesel injection quantity; The actuator is used to control the execution according to the opening of the ammonia valve and the diesel injection amount.

2. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition adaptation according to claim 1, characterized in that: The control unit integrates an ARM Cortex-A53 processor.

3. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition adaptation according to claim 1, characterized in that: The actuator includes an ammonia flow control valve that is actuated according to the ammonia valve opening and a diesel high-pressure common rail injector that is actuated according to the diesel injection amount.

4. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition self-adaptation according to claim 1, characterized in that: The data module is based on FPGA and performs real-time data storage and learning.

5. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition self-adaptation according to claim 1, characterized in that: The sliding window is used to sample the historical speed and load of 10 cycles to construct the feature vector X t , using weight matrix for training, improving the attention of load fluctuation, building LSTM prediction model, combining the feature vector X t Forecast the load to get the predicted load 6. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition self-adaptation according to claim 1, characterized in that: The ammonia valve opening and the diesel injection amount are converted into PWM signal parameters and transmitted to the actuator.

7. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition self-adaptation according to claim 1, characterized in that: The sensor module includes a speed sensor, a NO x Sensor, ammonia concentration sensor.

8. The ammonia replacement rate adjustment system for a marine ammonia engine based on working condition self-adaptation according to claim 1, characterized in that: The sensor module transmits original signals corresponding to multi-dimensional parameters via the CAN bus.

Images