Urea injection control method for SCR (Selective Catalytic Reduction) system of marine diesel engine

Abstract

The invention discloses a urea injection control method for a marine diesel engine SCR system. The urea injection control method comprises the steps that a marine diesel engine SCR system model is built; constructing a first objective function, using an improved ant colony algorithm to take three-dimensional vectors (Kp, Ki, Kd) of a PID controller as input, dispersing a search range of each dimension vector into each node, and calculating through a transition probability and a heuristic factor; iteratively updating paths and pheromones, and solving a first optimal solution; constructing a second objective function, and solving a second optimal solution by taking the first optimal solution as an initial weight of a dual-drive fruit fly optimization algorithm; and based on the second optimal solution, urea injection control is conducted in combination with an SCR system model. According to the method provided by the invention, the initial value is optimized by adopting the improved ant colony algorithm, and the improved ant colony algorithm is combined with the dual-drive fruit fly optimization algorithm, so that the dual-drive fruit fly algorithm optimized based on the improved ant colony algorithm is provided, and the urea injection amount can be accurately controlled.

Description

Technical Field

[0001] This application relates to the technical field of marine diesel engine exhaust gas denitrification and swarm intelligence optimization algorithm combination, specifically to a urea injection control method for a marine diesel engine SCR system. Background Technology

[0002] Selective catalytic reduction (SCR) is an advanced active emission control technology system. Its core function is to convert nitrogen oxides (NOx), the main pollutant in the exhaust gas emitted by equipment such as diesel engines, gas turbines, and industrial boilers, into harmless nitrogen (N2) and water (H2O) through a chemical reaction.

[0003] The most critical issue in a selective catalytic reduction (SCR) system is determining the optimal injection rate of the urea solution. Excessive injection rate leads to the release of some NH3, causing secondary pollution; insufficient injection rate results in the inability to release enough NH3 and NO. X A redox reaction occurs, resulting in a low system conversion rate that fails to meet the stipulated emission reduction requirements. Therefore, the SCR system aims to reduce NO emissions. X The key to achieving emission reduction targets lies in controlling the amount of urea injected, and developing a reasonable urea injection control strategy is crucial.

[0004] Currently, the urea control injection methods for SCR systems are mainly divided into two types: open-loop control and closed-loop control.

[0005] Open-loop control uses pre-calibrated urea injection pulse patterns under different operating conditions to control the urea injection quantity. Its advantages are simple structure and ease of implementation in practical engineering applications. However, its disadvantages include a large calibration workload, low control accuracy, and failure to meet emission standards. Closed-loop control, on the other hand, is based on sensor signal feedback. X The sensor detected NO downstream of the SCR reactor. X The signal is fed back to the control unit, continuously correcting the urea injection quantity, thereby improving control accuracy. However, due to NO... X Sensors can exhibit cross-sensitivity issues, frequently misdetecting NH3 as NO downstream of the SCR reactor. X This phenomenon leads to excessive urea injection. The urea injection volume needs precise control to ensure stable system operation. Inaccurate injection control of the injection system will affect the performance of the entire SCR system of the marine diesel engine. Summary of the Invention

[0006] This application addresses the technical problems existing in the prior art by providing a urea injection control method for a marine diesel engine SCR system.

[0007] The technical solution provided in this application is as follows: In a first aspect, this application provides a urea injection control method for a marine diesel engine SCR system, comprising: Construct a marine diesel engine SCR system model; the SCR system model includes nitrogen oxide concentration. ammonia concentration Ammonia coverage and average temperature inside the catalyst T Partial differential equations; Construct the first objective function, and use the improved ant colony algorithm to transform the three-dimensional vector of the PID controller ( K p , K i , K d The input is used to map the starting, path, and ending positions respectively. The search range of each dimension vector is diverged into each node, and the path of each node is... By calculating the transition probability and heuristic factor, the next node to be reached is selected, and the three-dimensional vector is updated. K p , K i , K d Iteratively update the path and pheromone to find the first optimal solution; A second objective function is constructed, using the first optimal solution as the initial weights of the dual-driven fruit fly optimization algorithm, and the second optimal solution is solved. The dual-driven fruit fly optimization algorithm introduces forward-guided multiple attractants and reverse-driven multiple repellents based on the fruit fly optimization algorithm to increase the global search probability. The second optimal solution includes the PID controller's... K p , K i and K d ; Based on the second optimal solution, urea injection control is performed using the SCR system model.

[0008] In one possible implementation, the SCR system model is as follows: For the state variables of the model, the expression is as follows:

[0009] In the formula: This indicates the concentration of nitrogen oxides inside the catalyst. express The total derivative with respect to time t; Indicates ammonia concentration. express The total derivative with respect to time t; Indicates ammonia coverage. express The derivative with respect to time t; This indicates the average temperature inside the catalyst. express The derivative with respect to time t; Indicates parameters, =1 / ( ), The void ratio, Represents the volume of an SCR unit; The values ​​represent the amount of each component; in indicates the inlet and out indicates the outlet. , The gas constant representing the exhaust gas of a diesel engine. The exhaust mass flow rate is (g / s). This refers to the amount of catalyst per unit volume. Atmospheric pressure; This indicates the overall denitrification efficiency of the SCR system; Indicates the ammonia adsorption rate; Indicates the denitrification conversion rate; Indicates the ammonia hydrolysis adsorption rate; Indicates the oxidation reaction rate; This represents the concentration of active atoms per unit volume of gas and is a constant. , Indicates parameters, , For the specific heat of exhaust gas at constant pressure, For the specific heat of the catalyst, m C For the quality of SCR catalytic converter; , This indicates the radiation coefficient of a diesel engine. σ represents the surface area of ​​the catalyst converter. sb The gas radiation constant; Indicates the exhaust temperature at the catalytic converter inlet. It represents the fourth power of the ambient temperature.

[0010] In one possible implementation, constructing the first objective function and solving for the first optimal solution using an improved ant colony algorithm includes: Construct a first objective function, which is determined by the PID controller. K p , K i, K d The NO obtained from the corresponding SCR system control simulation effect x The correlation function consisting of conversion rate, normalized ammonia slip, and system settling time; Initialize ant colony parameters, including ant colony size, maximum number of iterations, initial pheromone concentration, minimum volatility coefficient, and preset transfer probability threshold; Randomly select a node as the ant's initial node, and select the next node to reach by calculating the transition probability and heuristic factor, until all nodes have been visited; calculate the objective function for each round of the path. After each round of visits, update the transition probability and heuristic factor; After the iteration terminates, the optimal solution of the global objective function is output as the first optimal solution.

[0011] Furthermore, the expression for the first objective function is as follows:

[0012] In the formula: J is the objective function to be minimized; NO x Conversion rate NO x Non-conversion rate; the lower the value, the better. This represents the ammonia slip concentration (ppm). To allow the maximum ammonia slip, This is the normalized value for ammonia escape. Adjust the system time; For normalized weight coefficients, satisfying =1.

[0013] In one possible implementation, the heuristic factor is expressed as follows:

[0014] In the formula, Let the weight component of the k-th node in the d-th dimension be assigned a value. The current globally optimal weight vector The d-th dimension component; The transition probability The expression is as follows:

[0015] In the formula, Let represent the probability that the m-th ant chooses the k-th node in the d-th dimension. Let be the pheromone concentration of the k-th node in the d-th dimension. For pheromone importance factors, This is a heuristic factor and an importance factor.

[0016] Furthermore, the next node to be reached is selected by calculating the transition probability and heuristic factor, and the three-dimensional vector is updated. K p , K i , K d ),include: Each ant is determined by its probability of being transferred. Select the next node in each dimension and update the 3D vector. K p , K i , K d ); If the transition probability > Select the node with the largest heuristic factor; if the transition probability... < Randomly select nodes; The preset transition probability; The pheromone update method includes, after all ants have completed their path updates, analyzing the globally optimal three-dimensional vector (…). K p , K i , K d The corresponding node adds pheromones, while other nodes decrease pheromones according to an adaptive evaporation factor.

[0017] In the formula, Let be the pheromone concentration of the k-th node in the d-th dimension during the (t+1)-th iteration. Let be the adaptive volatile factor for the t-th iteration. Let be the pheromone concentration of the k-th node in the d-th dimension. For pheromone increment:

[0018] In the formula, The minimum objective function value for the current iteration is given, and the pheromone increment is inversely proportional to the objective value. This is the optimal weight vector.

[0019] In one possible implementation, constructing the second objective function, using the first optimal solution as the initial weights of the dual-driven fruit fly optimization algorithm, and solving for the second optimal solution includes: Construct a second objective function; the expression of the second objective function is as follows:

[0020] In the formula: express Conversion rate; This represents the non-conversion rate. Indicates the concentration of ammonia escape; To the maximum allowable ammonia slip; Indicates overshoot time; Indicates the maximum overshoot time; Indicates steady-state error; Normalized weighting coefficients; Dual-drive FOA parameter initialization; the FOA parameters include: fruit fly population parameters, search range, dual-drive parameters, and position update step size; With the first optimal solution As the initial population center, randomly generated within the defined search range. N The initial position vector of each fruit fly, and a set of candidate solutions for PID parameters for each fruit fly: (1 ); Based on the simulation of the PID controller and SCR system model, the fitness value of individual fruit flies was calculated. Multiple fruit fly locations with optimal fitness values ​​are selected from the current population as attractant locations to guide fruit flies to gather towards the optimal individuals; the attraction of each fruit fly to each attractant is calculated; multiple fruit fly locations with the worst fitness values ​​are selected from the current population as repellent locations to push fruit flies away from the inferior solution area; the repulsion of each fruit fly to each repellent is calculated; the locations are updated based on the location update formula. The fitness value corresponding to the position of each fruit fly after the update is calculated and the individual and global optimal values ​​are updated. If the current fitness value of an individual fruit fly is less than its minimum fitness value, the position of the fruit fly is updated to the current position and the current fitness value. If the minimum fitness value in the current population is less than the minimum fitness value of the global population, the global optimal position is updated to the position corresponding to the minimum fitness value, and the minimum fitness value in the current population is updated to the minimum fitness value. The iteration terminates and the second optimal solution is output.

[0021] Furthermore, the expression for the position update formula is as follows:

[0022] in, This is the updated fruit fly position vector. Here is the fruit fly position vector before the update; step is the step size. For multiple attractant quantities, For the attractiveness of the attractant, , Indicates the attraction strength coefficient. Location of the attractant; For the quantity of expulsion agent, For the repulsion agent's repulsion degree, , Indicates the expulsion strength coefficient. Location of the repellent.

[0023] Furthermore, the second optimal solution is the current globally optimal position after the iteration terminates. .

[0024] Secondly, this application provides a PID controller for implementing the urea injection control method for a marine diesel engine SCR system according to any one of claims 1-9, comprising: Signal acquisition module: used to acquire diesel engine operating parameters and SCR system status parameters; Improved ACO module: Receives SCR system parameters and control objectives, solves for the first optimal solution using an improved ant colony algorithm, and outputs the optimal initial weight vector for the dual-drive FOA. ; Dual-drive FOA module: used for... Using the initial population position and the SCR system control objective as the fitness function, the proportional gain of the PID controller is dynamically optimized online. Integral coefficient Differential coefficients ; PID control module: used to control the system based on the optimized... , , Calculate the urea injection quantity control signal to drive the urea injection system to perform the injection action; Feedback adjustment module: used to adjust the output NO of the SCR system. x The concentrations of NH3 and NH3 are fed back to the fitness function, forming a closed-loop optimization.

[0025] This application has the following beneficial effects: 1. The method provided in this application introduces heuristic factors and adaptive volatile factors, and proposes an improved ant colony algorithm. This improved ant colony algorithm can reduce the dependency and guidance effect of the ant colony on the ants, improve the global search ability, and discover better solutions.

[0026] 2. The method provided in this application addresses the problems of high computational complexity, strong parameter dependence, and weak global optimization ability faced by traditional swarm intelligence optimization algorithms, especially the problems of strong parameter dependence, slow convergence speed, and easy getting trapped in local extrema faced by traditional fruit fly algorithms. Based on the traditional fruit fly optimization algorithm, bacterial chemotaxis theory is introduced to propose a fruit fly optimization algorithm based on dual driving.

[0027] 3. The method provided in this application uses an improved ant colony algorithm to optimize the initial value and combines it with a dual-driven fruit fly optimization algorithm to propose a dual-driven fruit fly algorithm based on the improved ant colony algorithm.

[0028] 4. The method provided in this application can realize online optimization of the PID controller to achieve more precise control of the urea injection quantity. Attached Figure Description

[0029] Figure 1 A schematic flowchart of the urea injection control method for a marine diesel engine SCR system provided in this application embodiment; Figure 2 This is a schematic diagram illustrating the ant colony algorithm used in this application to select the optimal path; wherein, Figure 2 (a) represents a random search path. Figure 2 (b) represents the local search path; Figure 3 A flowchart illustrating the improved ant colony algorithm provided in an embodiment of this application; Figure 4 A flowchart illustrating the dual-drive-based fruit fly optimization algorithm provided in this application embodiment; Figure 5 This is a schematic diagram of the structure of the PID controller provided in the embodiments of this application; Figure 6 This is a schematic diagram of the PID controller control flow provided in an embodiment of this application. Detailed Implementation

[0030] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0032] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that this application can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid unnecessarily obscuring the description of this application. Therefore, this application is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0033] The existing marine diesel engine SCR system has the following problems: 1. Due to the variable operating conditions of marine diesel engines, and the existence of phenomena such as the heating state of the diesel engine oxidation catalyst, the regeneration of the diesel particulate filter, and the aging of the SCR itself in the aftertreatment system, the urea injection control of the SCR system needs to take into account both the engine operating conditions and the working status of the entire aftertreatment system, making the SCR system control a multi-timescale coupled control problem.

[0034] 2. Due to the nonlinearity, large delay, large inertia, and constrained output characteristics of marine diesel engine SCR systems, this study aims to solve problems such as low urea injection control accuracy, low NOx conversion rate, and large ammonia slip in marine diesel engine SCR systems.

[0035] Based on the above-mentioned technical problems, this application provides a urea injection control method for a marine diesel engine SCR system, see [link to relevant documentation]. Figure 1 ,include: S101. Construct a marine diesel engine SCR system model; the SCR system model includes nitrogen oxide concentration. ammonia concentration Ammonia coverage and average temperature inside the catalyst T Partial differential equations; S102. Construct the first objective function, and use the improved ant colony algorithm to convert the three-dimensional vector of the PID controller ( Kp , K i , K d The input is used to map the starting, path, and ending positions respectively. The search range of each dimension vector is diverged into each node, and the path of each node is... The first optimal solution is obtained by calculating the transition probability and heuristic factor, and iteratively updating the path and pheromone. S103. Construct a second objective function, using the first optimal solution as the initial weights of the dual-driven fruit fly optimization algorithm, and solve for the second optimal solution; the dual-driven fruit fly optimization algorithm introduces forward-guided multiple attractants and reverse-driven multiple repellents based on the fruit fly optimization algorithm to increase the global search probability; the second optimal solution includes the PID controller's... K p , K i and K d ; S104. Based on the second optimal solution, urea injection control is performed in conjunction with the SCR system model.

[0036] In one possible implementation, in S101, the SCR system model is... For the state variables of the model, the expression is as follows:

[0037] In the formula: This indicates the concentration of nitrogen oxides inside the catalyst. express The total derivative with respect to time t; Indicates ammonia concentration. express The total derivative with respect to time t; Indicates ammonia coverage. express The derivative with respect to time t; This indicates the average temperature inside the catalyst. express The derivative with respect to time t; Indicates parameters, =1 / ( ), The void ratio, Represents the volume of an SCR unit; The values ​​represent the amount of each component; in indicates the inlet and out indicates the outlet. , The gas constant representing the exhaust gas of a diesel engine. The exhaust mass flow rate is (g / s). This refers to the amount of catalyst per unit volume. Atmospheric pressure; This indicates the overall denitrification efficiency of the SCR system; Indicates the ammonia adsorption rate; Indicates the denitrification conversion rate; Indicates the ammonia hydrolysis adsorption rate; Indicates the oxidation reaction rate; This represents the concentration of active atoms per unit volume of gas and is a constant. , Indicates parameters, , For the specific heat of exhaust gas at constant pressure, For the specific heat of the catalyst, m C For the quality of SCR catalytic converter; , This indicates the radiation coefficient of a diesel engine. σ represents the surface area of ​​the catalyst converter. sb The gas radiation constant; Indicates the exhaust temperature at the catalytic converter inlet. It represents the fourth power of the ambient temperature.

[0038] In one possible implementation, see Figure 2 , Figure 3 S102 includes: S102a. Construct a first objective function, the first objective function being derived from the PID controller's (… K p , K i , K d The NO obtained from the corresponding SCR system control simulation effect x The correlation function consisting of conversion rate, normalized ammonia slip, and system settling time; S102b, using a PID controller ( K p , K i , K d The three-dimensional vectors are mapped to the starting, path, and ending positions of the ant, and the search range of each dimension vector is diffused into each node; S102c, Initialize ant colony parameters, including ant colony size, maximum number of iterations, initial pheromone concentration, minimum volatility coefficient, and preset transfer probability threshold; S102d: Randomly select a node as the ant's initial node, and select the next node to be reached by calculating the transition probability and heuristic factor, until all nodes have been visited; calculate the objective function of each round of the path. S102e: After each round of visits, update the transition probability and heuristic factor; S102f: After the iteration terminates, the optimal solution of the global objective function is output as the first optimal solution.

[0039] Furthermore, the expression for the objective function is as follows:

[0040] In the formula: J is the objective function to be minimized (the smaller J is, the better the initial weights). NO x Conversion rate NO x Non-conversion rate; the lower the value, the better. This represents the ammonia slip concentration (ppm). (Maximum permissible ammonia slip) This is the normalized value for ammonia escape; a smaller value is better. The system settling time (s) is NO. x The time it takes for the concentration to stabilize within the target range (±5%) after a disturbance; a smaller value indicates a faster response speed. For normalized weight coefficients, satisfying =1.

[0041] Furthermore, the expression for the heuristic factor is as follows:

[0042] In the formula, Let the weight component of the k-th node in the d-th dimension be assigned a value. The current globally optimal weight vector The d-th dimension component, The smaller, The larger the value, the better the node.

[0043] Furthermore, the transition probability The probability that the m-th ant chooses the k-th node in the d-th dimension, combined with pheromone concentration and heuristic factor, is given by the following formula:

[0044] In the formula, Let be the pheromone concentration of the k-th node in the d-th dimension. For pheromone importance factors, This is a heuristic factor and an importance factor.

[0045] Further, the method of selecting the next reached node by calculating the transfer probability and the heuristic factor includes: (1) Each ant selects the next node in each dimension according to the transfer probability and updates the weight vector ; (2) If the transfer probability > , select the node with the largest heuristic factor (greedy search, accelerating convergence); if the transfer probability < , randomly select a node (global search, avoiding local optimum); is the preset transfer probability.

[0046] Further, the pheromone update method includes: after all ants complete the path update, increase the pheromone for the node corresponding to the global optimal weight vector, and decay the pheromone for other nodes according to the adaptive evaporation factor: (4) In the formula, is the pheromone concentration of the k-th node in the d-th dimension at the (t + 1)-th iteration, is the adaptive evaporation factor at the t-th iteration, is the pheromone concentration of the k-th node in the d-th dimension, is the pheromone increment:

[0047] In the formula, is the minimum objective function value of the current iteration, and the pheromone increment is inversely proportional to the objective value; is the optimal weight vector.

[0048] Further, the method for determining the termination of iteration includes: (1) If the current iteration number t < G, then t = t + 1, and return to step 3 to repeat the iteration; G is the iteration number; (2) If t = G, the iteration terminates, and output the global optimal weight vector <00005​​​​​​​​​​​​​It should be noted that the difference between the improved ant colony algorithm described in this application and the traditional ant colony algorithm is that the path in the traditional ant colony algorithm is an "ordered connection of nodes (including their sequential order)," while the "path" in this application actually maps to... . The algorithm actually performs "three-dimensional parameter combination optimization" rather than "one-dimensional path sequence optimization". In essence, it is a reasonable adaptation of the ant colony algorithm to multi-dimensional parameter optimization scenarios.

[0052] 2. Dimensions and search scope of initial weights The initial weights of the dual-drive FOA are the position vectors of its initial population, with the same dimension as the search vector of FOA (i.e., the dimension of the PID parameters). D =3, corresponding to The initial search starting point). Considering the control characteristics of the marine diesel engine SCR system, the search range for the initial weights (FOA - initial population position) is as follows: w1 (corresponding to the initial value of Kp) is the initial weight of the PID proportional coefficient [0.5, 5.0]. w2 (corresponding to the initial value of Ki) PID integral coefficient initial weight [0.05, 0.5] w3 (corresponding to the initial value of Kd) PID derivative coefficients initial weights [0.01, 0.2] It should be noted that this application contains core logic equivalent to "converting node coordinates into a distance matrix to calculate the objective function," but it does not use the distance matrix based on physical coordinates as in traditional ant colony algorithms. Instead, it achieves "quality evaluation" of the path (initial weight vector) through weight component discretization and objective function quantization. Its core idea is: to convert PID parameters ( These are mapped to nodes along the ant path, and the control performance indicators (NO) output by the SCR system simulation are used to determine the path. x The objective function is constructed using conversion rate, ammonia slip, and settling time, replacing the traditional "distance" as a quantitative standard for evaluating the quality of a path.

[0053] The specific equivalent steps include: Step 1: Establish a "node-weight component" mapping (replacing the traditional node coordinate definition) The three parameters of the PID controller ( The search scope is discretized into several nodes, with each node corresponding to a specific value of the weight component of that dimension. The search scope is divided into three dimensions (d=1,2,3).

[0054] Each node's "identifier" is (d, k), where d is the dimension (1-3), and k is the node index (1-L) in that dimension. The core attribute of a node is the value of its corresponding weight component. .

[0055] Step 2: Construct a "path-initial weight vector" mapping (instead of the traditional path definition) 2.1 The ant's "path" corresponds to a complete set of initial weight vectors. (m is the ant number, 1≤m≤ant colony size M); 2.2 Path Generation Process: The ant selects one node in each dimension (d=1,2,3). The nodes in the three dimensions combine to form a path (i.e., a set of initial weight vectors). The "path" is equivalent to... .

[0056] 3. Design of the objective function (fitness function) The goal of the ant colony algorithm is to find the "optimal initial weights," that is, initial weights that, when substituted into the dual-drive FOA, enable the SCR system to have good control performance in the early stages of iteration (while also considering NO). x (Conversion rate and ammonia slip). Therefore, the objective function needs to quantify the SCR system control effect corresponding to the initial weights, designed as follows:

[0057] In the formula: J is the objective function to be minimized (the smaller J is, the better the initial weights). NO x Conversion rate NO x Non-conversion rate; the lower the value, the better. This represents the ammonia slip concentration (ppm). (Maximum permissible ammonia slip) This is the normalized value for ammonia escape; a smaller value is better. The system settling time (s), i.e., NO x The time it takes for the concentration to stabilize within the target range (±5%) after a disturbance; a smaller value indicates a faster response speed. For normalized weight coefficients, satisfying =1.

[0058] Understandably, traditional ant colony optimization algorithms calculate path length using a distance matrix. This application quantifies the quality of a path (initial weight vector) using an objective function J; a smaller J indicates a better path (better initial weights). The steps for constructing and calculating the objective function are as follows: Objective function calculation process (from path to J value): (a) Input: The initial weight vector corresponding to a certain path （ Right now (initial value); (b) Simulation: Substitute the data into the SCR system dynamic model to simulate the urea injection control process and output three key indicators: Conversion rate ammonia escape concentration System settling time ; (c) Calculation: Substitute the three indicators into the objective function to obtain the J value of the path (i.e., "path quality").

[0059] 4. Reasoning process for solving the initial weights using the improved ant colony algorithm Core idea: Each ant represents a set of initial weight vectors for the Fundamental Factors of Account (FOA). Through a global search of the ant colony (guided by pheromones and aided by heuristic factors), the weight vector that minimizes the objective function J is found, serving as the initial population position for the dual-driven FOA. The specific reasoning steps are as follows: 4.1 Parameter Initialization: Ant colony parameters: colony size M (number of ants, default M=30), maximum number of iterations G (default G=50), initial pheromone concentration =0.5), minimum volatility coefficient Preset transition probability threshold =0.7; Weight search parameters: Initial weight vector dimension D=3 (corresponding to...) , , The search range for each dimension (see 2.4.2) is divided into L nodes (default L=20); the objective function weight coefficients are defined. .

[0060] 4.2 Ant path initialization: Each ant randomly selects an initial node for each dimension (i.e., the initial value of the weight component) to form an initial weight vector. (1 ); 4.3 Calculation of the objective function value: The initial weight vector for each ant As the initial population position for the dual-driven FOA, it was substituted into the dynamic model of the SCR system for simulation; NO was recorded during the simulation. x Conversion rate ammonia escape concentration Adjusting time Calculate the objective function value Save the current globally optimal weight vector. (correspond .

[0061] It should be noted that the SCR system model is a complex multiphysics simulation tool capable of simulating a range of key parameters and processes, from chemical reactions to fluid dynamics. Its simulation capabilities vary depending on the complexity of the model (from zero-dimensional lumped parameter models to three-dimensional computational fluid dynamics models).

[0062] In this embodiment of the application, the weight vector corresponding to the node position of each ant is... (1 ),Right now( K p , K i , K d The urea injection rate is calculated using the core calculation model of the PID controller, and then the NO is output through SCR system model simulation. x The conversion rate, ammonia slip concentration, and system settling time are used to calculate the objective function value. Therefore, each node corresponds to an objective function value.

[0063] Specifically, NO x The conversion rate, ammonia slip concentration, and system settling time are determined as follows: These three parameters are all derived from the simulation output of the SCR system model, or further derived from the core parameters output by the simulation. The specific determination logic is as follows: (i) Conversion rate ( ) a. Core basis: "Catalyst inlet NO" output by the SCR system model x concentration( ) and "Export NO x concentration( )".

[0064] b. Calculation method: Directly derived from the denitrification efficiency formula, the formula is:

[0065] c. Model support: The "concentration field distribution" module of the SCR system model will simulate NO. x During the axial consumption process inside the catalyst, NO is output at the inlet and outlet in real time. x Concentration data provides direct input for conversion rate calculation.

[0066] (ii) Ammonia escape concentration ( ) The core basis is the "NH3 concentration at the catalyst outlet" directly output by the SCR system model.

[0067] b. Model Support: The model's "Concentration Field Distribution" module simultaneously simulates the adsorption, reaction, and desorption processes of NH3. c. By solving for the ammonia coverage ratio ( The dynamic equation of the ammonia escape concentration (d) is used to output the concentration of NH3 that escaped without participating in the reaction, which is the concentration of NH3 that escaped without participating in the reaction, in ppm.

[0068] e-constraint correlation: The model will incorporate "surface chemical reaction kinetics" (such as ammonia oxidation side reaction) to ensure that the calculated ammonia escape concentration closely matches the actual reaction law.

[0069] (iii) System settling time (t) settle ) a. Core basis: The "NO" output of the SCR system model x Concentration dynamic response curve.

[0070] Definition b: From operating condition disturbances (such as sudden load changes, NO) x From the moment the inlet concentration abruptly changes until the outlet NO concentration... x The duration for which the concentration remains stable within the target range (±5%).

[0071] C Determination method: The model will record the exit NO in real time. x The concentration change data over time is used to determine the stable time by judging that "the concentration fluctuation within 3 consecutive sampling periods (sampling frequency 100Hz) is ≤±5%", and then the time difference from the disturbance to the stable time is calculated, which is the adjustment time.

[0072] 4.4 Calculation of transition probability and heuristic factor: Heuristic factor The heuristic factor for the m-th ant to select the k-th node in the d-th dimension (d=1,2,3) is defined as the potential of the weight component corresponding to that node to reduce the objective function, and is calculated as follows:

[0073] In the formula, Let the weight component of the k-th node in the d-th dimension be assigned a value. The current globally optimal weight vector The d-th dimension component, The smaller, The larger the value, the better the node.

[0074] Transition probability The probability that the m-th ant chooses the k-th node in the d-th dimension, combined with pheromone concentration and heuristic factor, is given by the following formula:

[0075] In the formula, is the pheromone concentration of the k-th node in the d-th dimension, is the pheromone importance factor, is the heuristic factor importance factor.

[0076] 4.5. Ant path update (weight vector optimization): (4.5.1) Each ant selects the next node in each dimension according to the transition probability and updates the weight vector ; (4.5.2) If the transition probability > , select the node with the largest heuristic factor (greedy search, accelerating convergence); (4.5.3) If the transition probability < , randomly select a node (global search, avoiding local optimum).

[0077] 4.6. Pheromone update: Global pheromone update: After all ants complete the path update, increase the pheromone for the nodes corresponding to the global optimal weight vector, and decay the pheromone for other nodes according to the adaptive evaporation factor:

[0078] In the formula, is the pheromone concentration of the k-th node in the d-th dimension at the (t + 1)-th iteration, is the adaptive evaporation factor at the t-th iteration, is the pheromone concentration of the k-th node in the d-th dimension, is the pheromone increment:

[0079] In the formula, is the minimum objective function value of the current iteration, and the pheromone increment is inversely proportional to the objective value.

[0080] 4.7. Iteration termination judgment: (4.7.1) If the current iteration number t < G, then t = t + 1, and return to step 3 to repeat the iteration; (4.7.2) If t = G, terminate the iteration and output the global optimal weight vector , which serves as the initial population position of the dual-driven FOA.

[0081] 5. Input, output, and constraint conditions

[0082] 6. Explanation of the core logic of the improved ant colony algorithm The output of the improved ant colony algorithm in this application (optimal initial weight vector) The core logic of the ant colony algorithm is consistent with "path planning → outputting the solution corresponding to the optimal path", with only an adaptation transformation made for the "three-dimensional PID parameter optimization" scenario.

[0083] 6.1 Core Logic Alignment: Consistency of Outputs between Traditional Ant Colony Algorithm and Algorithm in this Application The essence of the ant colony algorithm is "to find the optimal path through swarm search and then output the solution to the problem corresponding to that path." This application fully follows this logic, only making engineering adaptations to the definitions of "path" and "solution."

[0084] 6.1 Appropriateness and rationality of the scenario The core objective of this application is to "optimize the initial search starting point for PID parameters," rather than "find the movement order in physical space." Therefore, the definition of "path" has been transformed according to a specific scenario, and the output results naturally adapt to the objective. 6.2 The essence of a path is a "candidate solution vector". In traditional two-dimensional path planning, nodes are physical locations (such as cities), and the path must reflect the "sequence" (movement trajectory), so the "node order" is output; In this application, the nodes are discrete values ​​of the PID parameters (such as...). =1.0、 =0.15、 =0.08), the combination of three-dimensional nodes itself is a complete set of PID initial weights (candidate solutions).

[0085] 6.3 The output "weight vector" is a direct solution to the problem. The core requirement of this application is to provide the "optimal initial population position" for dual-drive FOA, that is, the initial search starting point for PID parameters; Weight vector corresponding to the combination of three-dimensional nodes It directly provides the initial weights required by FOA, without any additional transformation, which meets the requirement of engineering problems to "output directly usable solutions".

[0086] 6.4 Feasibility verification of the algorithm's output result: closed loop from search process to result output The complete process of improving the ant colony algorithm ensures that it can stably output the optimal weight vector. The key steps in the closed loop are as follows: 6.4.1 The mapping between nodes and solution vectors is fixed. a three-dimensional dimension ( Each node in the Kp dimension has a unique value (e.g., the third node in the Kp dimension is fixed at w1=1.0). b. Each "path search" by an ant essentially involves selecting one node from each of the three dimensions and combining them to form a unique weight vector. To clarify the mapping relationship.

[0087] 6.4.2. Pheromone and Heuristic Factors Guide the Search for the Optimal Combination Pheromones accumulate only at the nodes corresponding to the optimal weight vector (e.g., ...). corresponding node, node, node); b. Heuristic Factor By directly associating the current optimal weight component, the ants are guided to prioritize selecting "nodes that are closer to the optimal solution", ensuring that the search direction focuses on the combination that minimizes the objective function J.

[0088] 6.4.3 Output the value of the optimal combination after the iteration terminates. During the iteration process, the globally optimal weight vector is recorded in each round. and corresponding ; Once b reaches the maximum number of iterations G, directly output the current iteration. As Its three components correspond to The optimal initial value fully satisfies the requirement of "outputting the optimal three-dimensional solution vector".

[0089] In one possible implementation, the core of S103 is to improve the first optimal solution of the ACO output by "quantifying control performance through a second objective function + optimizing PID parameters through dual-drive FOA". l Using initial weights, the optimal solution for the PID controller is accurately obtained. .

[0090] See Figure 4 S103 includes: S103a. Construct the second objective function (fitness function). The expression of the second objective function is as follows:

[0091] In the formula: express The higher the conversion rate, the better; The lower the non-conversion rate, the better; This indicates the ammonia escape concentration; the lower the value, the better. =10 ppm (Maximum permissible ammonia slip); Indicates the overshoot time; a smaller value is better. Outputted from the SCR system model. Indicates the maximum overshoot time; SCR system model output; This represents the steady-state error; the smaller the value, the better. It indicates the output NO after stabilization. x Deviation between concentration and target concentration; SCR system model output; Normalized weight coefficients, satisfying Default value To balance emission reduction effects, secondary pollution, dynamic response and steady-state accuracy.

[0092] The second objective function improves the first optimal solution of the ACO output by "quantifying control performance with the second objective function + optimizing PID parameters with dual-drive FOA". lW final Using initial weights, the optimal k of the PID controller is precisely solved. p k i k d Optimization objective: Minimize ,Right now The smaller the value, the better the PID parameters ( ) is better.

[0093] S103b, Initialize dual-drive FOA parameters; the FOA parameters include: fruit fly population parameters, search range, dual-drive parameters, and position update step size.

[0094] Specifically, initialize the core FOA parameters: (1) Population parameters: Fruit fly population size N=40 (default), maximum number of iterations (default).

[0095] (2) Search scope: by Centered on this, we set dynamic search ranges for each parameter (balancing local optimization and global exploration): Search scope: ; Search scope: ; Search scope: .

[0096] (3) Dual drive parameters: Multiple attractant quantities =3 (Corresponding to the three optimization directions of "high conversion rate, low ammonia slip, and fast response"); Multiple expulsion agent quantities =2 (Corresponding to two deterioration directions: "large overshoot" and "large steady-state error"); attraction strength coefficient =0.8 (controls the attractiveness of the attractant to the fruit fly's location); Expulsion strength coefficient =0.6 (controls the repulsion force of the repellent on the fruit fly's location). (4) Position update step size: initial step size step =0.05, which decreases linearly with the number of iterations (the step size is reduced in the later stages of iteration for precise convergence).

[0097] S103c, using the first optimal solution As the initial population center, randomly generated within the search range defined in S103b. N The initial position vector of each fruit fly, and a set of candidate solutions for PID parameters for each fruit fly: (1 ).

[0098] Specifically, S103c includes: (1) Generation logic: , , , (2) Constraint verification: If the generated parameters exceed the search range, they will be automatically truncated to the range boundary.

[0099] S103d, based on PID controller and SCR system model simulation, calculate the individual fitness value (second objective function value) of fruit flies.

[0100] Specifically, the position vector of each fruit fly Substitute the SCR system model and PID controller into the simulation to calculate the fitness value. ,include: (1) Simulation input: Current diesel engine operating parameters (speed, load), SCR system status parameters (input NO) x(Concentration, catalyst temperature, etc.) (2) Simulation process: The urea injection rate is calculated by the PID controller, and the SCR system model outputs NO. x Conversion rate ammonia escape concentration C NH3,slip Overshoot time Steady-state error ; (3) Fitness calculation: Substitute the four indicators into the second objective function to obtain the fitness value for each fruit fly. ; (4) Record the global optimum: save the current population Minimum fruit fly position and corresponding fitness values .

[0101] S103e: Select multiple fruit fly positions with optimal fitness from the current population as attractant positions to guide fruit flies to gather towards the optimal individuals; calculate the attraction degree of each fruit fly to each attractant; select multiple fruit fly positions with worst fitness from the current population as repellent positions to push fruit flies away from the inferior solution area; calculate the repulsion degree of each fruit fly to each repellent; update the positions based on the position update formula.

[0102] Understandably, S103e is a Drosophila position update based on multiple attractants and multiple repellents (a dual-drive core). Traditional FOA relies solely on random search, while this step guides the Drosophila to cluster towards the optimal solution through a dual drive of "forward attraction + reverse repulsion," while maintaining global search capability.

[0103] Specifically, S103e includes: (1) Construction of multiple attractants: Select from the current population The optimal fitness position for fruit flies is selected as the "attractant position". ( ). Calculate the "attraction" of each fruit fly to each attractant: (The closer the distance, the greater the attraction.) This represents the attraction strength coefficient.

[0104] (2) Construction of multiple repellent agents: Select from the current population The position of the fruit fly with the worst fitness value is used as the "repellent position". ( ). Calculate the "rejection" of each fruit fly to each repellent: (The smaller the distance, the greater the repulsion.) This represents the expulsion intensity coefficient.

[0105] (3) Position update formula:

[0106] Formula Explanation: The first part in parentheses (the attraction term): guides fruit flies to gather towards the optimal individual, accelerating local convergence; The second part in parentheses (the expulsion term): pushes the fruit fly away from the region of inferior solutions to avoid getting trapped in local extrema; Random items rand Retain a small amount of random search to maintain global exploration capabilities.

[0107] (4) Boundary constraints: Updated If the search range is exceeded, the search will automatically adjust to the boundary value.

[0108] S103f: Calculate the fitness value corresponding to the position of each fruit fly after the update and perform individual and global optimal updates; wherein, if the current fitness value of an individual fruit fly is less than its minimum fitness value, the position of the fruit fly is updated to the current position and the current fitness value; if the minimum fitness value in the current population is less than the minimum fitness value of the global population, the global optimal position is updated to the position corresponding to the minimum fitness value, and the minimum fitness value in the current population is updated to the minimum fitness value.

[0109] Specifically, S103f includes: (1) Calculate the position of each fruit fly after the update. Corresponding fitness value (The calculation logic is the same as S103d); (2) Individual optimal update: if , If the optimal fitness value is represented, then the position of the fruit fly is updated to [value]. The fitness value is updated to Otherwise, maintain the original position and the appropriate value. (3) Global optimal update: If the minimum fitness value in the current population is... min , If the fitness value is the optimal value for the global population, then update the optimal global position. Update the position corresponding to the minimum fitness value. This is the minimum fitness value.

[0110] S103g, iteration termination, and output of the second optimal solution.

[0111] Specifically, S103g includes: (1) Iteration termination condition: a If the current iteration number If the iteration terminates, then the iteration ends. b. If the globally optimal fitness value is found after 10 consecutive iterations If the fluctuation is ≤0.001 (convergence accuracy threshold), the iteration is terminated early (to avoid invalid calculations). (2) Output the second optimal solution: the current global optimal position after the iteration terminates. These are the final optimized PID parameters, serving as the second optimal solution.

[0112] The reasoning process for S103 is described below.

[0113] 1. The rationale for using the first optimal solution as the initial weight. (1) Traditional FOA initializes the population with random values, which is prone to getting trapped in local extrema and has slow convergence. (2) Improve ACO output It is the globally optimal initial weight, which has passed the objective function. Verify its corresponding "better control performance" (NO) x Conversion rate ≥85%, ammonia slip ≤15ppm); (3) with Using the initial population center of FOA allows the population to start searching from a "better region," reducing ineffective exploration and significantly improving convergence speed.

[0114] 2. The innovative value of the dual-drive mechanism (solving the pain points of traditional FOA)

[0115] 3. Targeted Design of the Second Objective Function Compared to the first objective function of S102 (Focusing only on the basic control performance of the initial weights). The addition of "overshoot time" and "steady-state error" metrics better aligns with the core requirements of PID parameter optimization (dynamic response + steady-state accuracy). Weighting coefficient The proportion is higher (0.7% in total), so priority should be given to ensuring NO. x The conversion rate and low ammonia slip meet the core emission reduction targets of the SCR system; All indicators are obtained directly through SCR system model simulation, with clear quantification logic and strong operability.

[0116] 4. Co-optimization logic with S102 S102 (Improved ACO): Solves the problem of "initial search starting point for PID parameters", outputs globally optimal initial weights, and avoids FOA blindly searching from random regions; S103 (Dual Drive FOA): Solves the problem of "precise optimization of PID parameters" by using a dual drive mechanism to finely adjust parameters based on the initial weights. The two work together to form a two-level optimization system of "coarse optimization → fine optimization", which not only ensures global search capability but also improves local convergence accuracy, ultimately achieving the optimal configuration of PID parameters.

[0117] In one possible implementation, S104 includes: using the second optimal solution as the PID parameter of the core calculation model of the PID controller, and updating the PID parameter in real time through dual-drive FOA to perform urea injection control.

[0118] Understandably, the core logic for calculating urea injection volume is: based on "SCR system outlet NO..." x The control objective is to achieve a urea injection concentration that approaches the target emission concentration. Through the coordinated action of the proportional-integral-derivative (PID) stages, a dynamic output of urea injection quantity matching the operating conditions is generated. The specific calculation process is as follows: I. Define the input and target of PID control 1. Control Input The core input of a PID controller is the actual NO. x Concentration and target NO x The formula for calculating the concentration deviation is:

[0119] In the formula: This is a deviation signal. A positive value indicates that the actual emissions exceed the standard and the amount of urea injected needs to be increased; a negative value indicates that the amount of urea injected is excessive and the amount of injection needs to be reduced. Represents time t Target emission concentration; This represents the actual output of the SCR catalyst at time t. concentration.

[0120] 2. Control Objectives By adjusting the urea injection rate, the deviation e(t) is made to rapidly approach 0, while simultaneously satisfying:

[0121] ppm In the formula; express Conversion rate; Indicates the ammonia escape concentration, from Sensor-assisted feedback verification.

[0122] II. Core Calculation Model of PID Controller Since the urea injection rate needs to be a continuously adjustable physical quantity (volume flow rate), a positional PID algorithm (directly outputting the absolute value of the control quantity) is used, and the calculation formula is as follows:

[0123] In the formula: The urea injection rate control signal, representing the time, is ultimately converted into volumetric flow rate, L / h; , , This represents the optimal PID parameters obtained from the improved ACO-dual-drive FOA optimization (dynamically updated according to operating conditions).

[0124] III. Engineering conversion of control signals (from PID output to actual injection quantity) The control signal U(t) output by the PID controller needs to be converted into a physical quantity (volume flow rate) that the urea injection system can execute. The conversion logic is calibrated based on the hardware characteristics of the urea injection system, and the formula is as follows:

[0125] In the formula: This represents the final urea injection rate at time t, in L / h. This indicates the calibration coefficient of the urea injection system, which is determined by the injector flow characteristics. ; The value represents the basic urea injection rate at time t, in L / h.

[0126] IV. Dynamic Adjustment Mechanism (in conjunction with algorithm optimization) Due to the variable operating conditions (speed, load) of marine diesel engines and the dynamic characteristics of the SCR system, the PID parameters need to be updated in real time through dual-drive FOA. The dynamic adjustment process of urea injection quantity is as follows: 1. Real-time acquisition of operating condition parameters (n, P) and SCR system status parameters ( , , (T) 2. Calculate the deviation ;

[0127] 3. Dual-drive FOA based on the current e(t), , Optimization and Update , , ; 4. Substitute the new optimal parameters into the PID controller, calculate U(t), and convert it to... ; 5. Urea injector operation Meanwhile, the control effect is verified by feedback signals. If e(t) exceeds the allowable range (±5%), repeat steps 3-4 to re-optimize.

[0128] The effectiveness of this application will be further explained below based on experimental test results.

[0129] I. Description of Experimental Conditions 1. Test subjects Marine diesel engine model: WD615.44 (rated power 290 kW, rated speed 1500 rpm); SCR system parameters: catalyst volume 0.8 m³, catalyst type V2O5-WO3 / TiO2, urea injector maximum flow rate 50 L / h; Test conditions: Covering typical ship operating conditions (idle, partial load, rated load, load change), as well as dynamic scenarios such as DPF regeneration and catalyst aging (simulating 3000 hours of operation).

[0130] 2. Testing Equipment and Specifications Sensor: NO x Sensors (measurement range 0~1000 ppm, accuracy ±2%), NH3 sensor (measurement range 0~50 ppm, accuracy ±0.5 ppm), exhaust temperature sensor (measurement range 200~500℃, accuracy ±5℃); Data acquisition instrument: sampling frequency 100 Hz, recording NO x Indicators such as conversion rate, ammonia slip concentration, system settling time, overshoot, and PID parameter iteration time; Comparison method: Method 1: Traditional PID control (ZN method for parameter tuning, Kp=1.2, Ki=0.2, Kd=0.1); Method 2: Single FOA optimized PID control (traditional FOA, population size 30, maximum number of iterations 100). Method 3: Classic ACO optimized PID control (traditional ACO, ant colony size 30, maximum number of iterations 50); Method 4: Improved ACO - Dual-drive FOA Adaptive PID Control (This invention method, parameters are the same as the initialization settings described above).

[0131] II. Core Measured Data and Performance Comparison (1) Comparison of control accuracy under steady-state conditions (rated load: 1500 rpm, 80% load) is shown in Table 1.

[0132] Table 1 Comparison of control accuracy under steady-state conditions

[0133] Conclusion: Under steady-state conditions, the NO of the method presented in this paper... x The conversion rate remained stable above 91%, and the ammonia slip concentration was controlled below 5 ppm, far exceeding the IMO Tier III emission regulations (NOx). x Conversion rate ≥85%, ammonia slip ≤10 ppm); compared with traditional PID, the ammonia slip concentration is significantly reduced by 65.8%, effectively avoiding corrosion of the post-treatment system; urea consumption rate is reduced by 16.2%, significantly reducing operating costs.

[0134] (2) Comparison of response performance under dynamic operating conditions (load change: 40%→80% load, speed 1500 rpm), see Table 2.

[0135] Table 2 Comparison of response performance under dynamic operating conditions

[0136] Conclusion: Under sudden load changes, the settling time of the proposed method is only 2.3 s, which is 72% shorter than that of traditional PID, and the overshoot is reduced by 81.1%, demonstrating extremely strong dynamic response capability. The maximum ammonia slip peak is controlled at 7.2 ppm, avoiding the problem of excessive urea injection caused by sudden load changes. This is due to the real-time parameter optimization of the dual-drive FOA and the optimal initial weights provided by the improved ACO, which improves the convergence speed of the algorithm by more than 50% (refer to the convergence characteristic improvement law of ACO-optimized PID).

[0137] (3) Comparison of adaptability to complex working conditions (diesel particulate filter DPF regeneration + catalyst aging scenario), see Table 3.

[0138] Table 3 Comparison of adaptability to complex working conditions

[0139] Conclusion: Under the complex scenario of DPF regeneration (sudden 100°C increase in exhaust temperature) and catalyst aging (30% decrease in activity), traditional PID control, with its fixed parameters, leads to increased NO emissions. x The conversion rate dropped to 76.2%, and ammonia slip exceeded the standard; the method of this invention improves the dynamic parameter optimization of ACO-dual-driven FOA, NO xThe conversion rate remained at 89.6%, the ammonia slip concentration was only 5.8 ppm, and the coefficient of variation (CV%) of control stability was only 1.6%, demonstrating strong anti-interference ability and adaptability to operating conditions. This is consistent with the improvement of the robustness of PID in nonlinear systems by intelligent algorithm optimization.

[0140] (4) Comparison of algorithm optimization efficiency (PID parameter tuning time), see Table 4.

[0141] Table 4 Comparison of Algorithm Optimization Efficiency

[0142] Conclusion: This invention improves the ACO (Automatic Coding Oscillator) by pre-solving the optimal initial weights (taking only 85 ms), reducing the parameter iteration convergence time of dual-drive FOA from 45 ms / cycle to 18 ms / cycle. The total computation time for one hour of operation is only 6.5 s, far below the computing power threshold of the controller hardware (typically ≥10 s), meeting the real-time requirements of engineering. Compared to a single FOA-PID, the algorithm optimization efficiency is improved by 60%, which is attributed to the global optimality of the initial weights reducing invalid searches, consistent with the convergence acceleration effect of ant colony optimization of initial parameters.

[0143] III. Core Conclusions from Measured Data The control accuracy is significantly improved: the method of this invention achieves higher NO control accuracy under steady-state conditions. x With a conversion rate ≥91% and ammonia slip concentration ≤4.2 ppm, the results represent an improvement of 8.8% and a reduction of 65.8% respectively compared to traditional PID control, fully meeting IMO Tier III and ultra-low emission regulations (NOx 100 ppm). x ≤50 ppm, ammonia slip ≤5 ppm); Rapid and stable dynamic response: Under sudden load changes, the settling time is only 2.3 seconds, and the overshoot is 3.5%, which are 58.9% and 65.7% shorter respectively compared to a single FOA-PID, effectively avoiding NO x Exceeding standards and a surge in ammonia escape; It has extremely strong adaptability to operating conditions: In complex scenarios such as DPF regeneration and catalyst aging, the coefficient of variation of control stability is only 1.6%, which is far better than the 8.3% of traditional PID, demonstrating extremely strong anti-interference ability; Its engineering practicality is outstanding: the total computation time of the algorithm is only 6.5 seconds / hour, without the need for additional hardware computing power, and the urea consumption rate is reduced by 16.2%. It balances control performance and economy and is easy to promote in engineering.

[0144] IV. Data Validity Statement The measured data are all from the average ± standard deviation of three repeated experiments. The sensor accuracy has been verified by calibration, and the data reliability is ≥98%. The parameters of the comparison methods are all set to their respective optimal values ​​(e.g., the population size and number of iterations for a single FOA-PID are consistent with the method in this paper) to ensure the fairness of the comparison. The test conditions cover the entire operating range of marine diesel engines, and the data are representative and can be extended to SCR systems of similar marine diesel engines.

[0145] The following describes a PID controller designed based on the aforementioned method. The PID controller is used to implement the aforementioned urea injection control method for marine diesel engine SCR systems.

[0146] See Figure 5 , Figure 6 The PID controller designed based on the aforementioned method provided in this application achieves precise control through a three-level optimization logic: "improved ACO to solve for optimal initial weights → dual-drive FOA to optimize PID parameters online → PID controller outputs urea injection quantity".

[0147] The controller includes: 1. Signal Acquisition Module 51: Used to acquire diesel engine operating parameters (speed, load) and SCR system status parameters (catalyst inlet temperature). , entrance NO x concentration Export NO x concentration (Export NH3 concentration); 2. Improved ACO module 52: This module receives SCR system parameters and control objectives, and uses an improved ant colony algorithm to solve for the initial weights of the dual-drive FOA, outputting the optimal initial weight vector for the dual-drive FOA. ; 3. Dual-drive FOA module 53: used for... Using the initial population position and the SCR system control objective as the fitness function, the proportional gain of the PID controller is dynamically optimized online. Integral coefficient Differential coefficients ; 4. PID control module 54: used to control the PID controller based on the optimized... , , Calculate the urea injection quantity control signal to drive the urea injection system to perform the injection action; 5. Feedback Adjustment Module 55: Used to adjust the output NO of the SCR system.x The concentrations of NH3 and NH3 are fed back to the fitness function, forming a closed-loop optimization.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for controlling urea injection in a marine diesel engine SCR system, characterized in that, include Construct a marine diesel engine SCR system model; the SCR system model includes nitrogen oxide concentration. ammonia concentration Ammonia coverage and average temperature inside the catalyst T Partial differential equations; Construct the first objective function, and use the improved ant colony algorithm to transform the three-dimensional vector of the PID controller ( K p , K i , K d The input is used to map the starting, path, and ending positions respectively. The search range of each dimension vector is diverged into each node, and the path of each node is... By calculating the transition probability and heuristic factor, the next node to be reached is selected, and the three-dimensional vector is updated. K p , K i , K d Iteratively update the path and pheromone to find the first optimal solution; A second objective function is constructed, using the first optimal solution as the initial weights of the dual-driven fruit fly optimization algorithm, and the second optimal solution is solved. The dual-driven fruit fly optimization algorithm introduces forward-guided multiple attractants and reverse-driven multiple repellents based on the fruit fly optimization algorithm to increase the global search probability. The second optimal solution includes the PID controller's... K p , K i and K d ; Based on the second optimal solution, urea injection control is performed using the SCR system model.

2. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The SCR system model is based on For the state variables of the model, the expression is as follows: In the formula: This indicates the concentration of nitrogen oxides inside the catalyst. express The total derivative with respect to time t; Indicates ammonia concentration. express The total derivative with respect to time t; Indicates ammonia coverage. express The derivative with respect to time t; This indicates the average temperature inside the catalyst. express The derivative with respect to time t; Indicates parameters, =1 / ( ), The void ratio, Represents the volume of an SCR unit; The values ​​represent the amount of each component; in indicates the inlet and out indicates the outlet. , The gas constant representing the exhaust gas of a diesel engine. This refers to the exhaust mass flow rate. This refers to the amount of catalyst per unit volume. Atmospheric pressure; This indicates the overall denitrification efficiency of the SCR system; Indicates the ammonia adsorption rate; Indicates the denitrification conversion rate; Indicates the ammonia hydrolysis adsorption rate; Indicates the oxidation reaction rate; This represents the concentration of active atoms per unit volume of gas and is a constant. , Indicates parameters, , For the specific heat of exhaust gas at constant pressure, For the specific heat of the catalyst, m C For the quality of SCR catalytic converter; , This indicates the radiation coefficient of a diesel engine. σ represents the surface area of ​​the catalyst converter. sb The gas radiation constant; Indicates the exhaust temperature at the catalytic converter inlet. It represents the fourth power of the ambient temperature.

3. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The construction of the first objective function and the solution of the first optimal solution using the improved ant colony algorithm include: Construct a first objective function, which is determined by the PID controller. K p , K i , K d The NO obtained from the corresponding SCR system control simulation effect x The correlation function consisting of conversion rate, normalized ammonia slip, and system settling time; Initialize ant colony parameters, including ant colony size, maximum number of iterations, initial pheromone concentration, minimum volatility coefficient, and preset transfer probability threshold; Randomly select a node as the ant's initial node, and select the next node to reach by calculating the transition probability and heuristic factor, until all nodes have been visited; calculate the objective function for each round of the path. After each round of visits, update the transition probability and heuristic factor; After the iteration terminates, the optimal solution of the global objective function is output as the first optimal solution.

4. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The expression for the first objective function is as follows: In the formula: J is the objective function to be minimized; NO x Conversion rate NO x Non-conversion rate; This represents the ammonia escape concentration. To allow the maximum ammonia slip, This is the normalized value for ammonia escape. Adjust the system time; For normalized weight coefficients, satisfying =1.

5. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The expression for the heuristic factor is as follows: In the formula, Let the weight component of the k-th node in the d-th dimension be assigned a value. The current globally optimal weight vector The d-th dimension component; The transition probability The expression is as follows: In the formula, Let represent the probability that the m-th ant chooses the k-th node in the d-th dimension. Let be the pheromone concentration of the k-th node in the d-th dimension. For pheromone importance factors, This is a heuristic factor and an importance factor.

6. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The process involves calculating the transition probability and heuristic factor to select the next node to be reached and updating the three-dimensional vector. K p , K i , K d ),include: Each ant is determined by its probability of being transferred. Select the next node in each dimension and update the 3D vector. K p , K i , K d ); If the transition probability > Select the node with the largest heuristic factor; if the transition probability... < Randomly select nodes; The preset transition probability; The pheromone update method includes, after all ants have completed their path updates, analyzing the globally optimal three-dimensional vector (…). K p , K i , K d The corresponding node adds pheromones, while other nodes decrease pheromones according to an adaptive evaporation factor. In the formula, Let be the pheromone concentration of the k-th node in the d-th dimension during the (t+1)-th iteration. Let be the adaptive volatile factor for the t-th iteration. Let be the pheromone concentration of the k-th node in the d-th dimension. For pheromone increment: In the formula, The minimum objective function value for the current iteration is given, and the pheromone increment is inversely proportional to the objective value. This is the optimal weight vector.

7. The urea injection control method for a marine diesel engine SCR system according to claim 1, characterized in that, The construction of the second objective function, using the first optimal solution as the initial weights of the dual-driven fruit fly optimization algorithm, and solving for the second optimal solution, includes: Construct a second objective function; the expression of the second objective function is as follows: In the formula: express Conversion rate; This represents the non-conversion rate. Indicates the concentration of ammonia escape; To the maximum allowable ammonia slip; Indicates overshoot time; Indicates the maximum overshoot time; Indicates steady-state error; Normalized weighting coefficients; Dual-drive FOA parameter initialization; the FOA parameters include: fruit fly population parameters, search range, dual-drive parameters, and position update step size; With the first optimal solution As the initial population center, randomly generated within the defined search range. N The initial position vector of each fruit fly, and a set of candidate solutions for PID parameters for each fruit fly: (1 ); Based on the simulation of the PID controller and SCR system model, the fitness value of individual fruit flies was calculated. Multiple fruit fly locations with optimal fitness values ​​are selected from the current population as attractant locations to guide fruit flies to gather towards the optimal individuals; the attraction of each fruit fly to each attractant is calculated; multiple fruit fly locations with the worst fitness values ​​are selected from the current population as repellent locations to push fruit flies away from the inferior solution area; the repulsion of each fruit fly to each repellent is calculated; the locations are updated based on the location update formula. The fitness value corresponding to the position of each fruit fly after the update is calculated and the individual and global optimal values ​​are updated. If the current fitness value of an individual fruit fly is less than its minimum fitness value, the position of the fruit fly is updated to the current position and the current fitness value. If the minimum fitness value in the current population is less than the minimum fitness value of the global population, the global optimal position is updated to the position corresponding to the minimum fitness value, and the minimum fitness value in the current population is updated to the minimum fitness value. The iteration terminates and the second optimal solution is output.

8. The urea injection control method for a marine diesel engine SCR system according to claim 7, characterized in that, The expression for the position update formula is as follows: in, This is the updated fruit fly position vector. Here is the fruit fly position vector before the update; step is the step size. For multiple attractant quantities, For the attractiveness of the attractant, , Indicates the attraction strength coefficient. Location of the attractant; For the quantity of expulsion agent, For the repulsion agent's repulsion degree, , Indicates the expulsion strength coefficient. Location of the repellent.

9. The urea injection control method for a marine diesel engine SCR system according to claim 7, characterized in that, The second optimal solution is the current globally optimal position after the iteration terminates. .

10. A PID controller, characterized in that, The method for implementing the urea injection control of the marine diesel engine SCR system according to any one of claims 1-9 includes: Signal acquisition module: used to acquire diesel engine operating parameters and SCR system status parameters; Improved ACO module: Receives SCR system parameters and control objectives, solves for the first optimal solution using an improved ant colony algorithm, and outputs the optimal initial weight vector for the dual-drive FOA. ; Dual-drive FOA module: used for... Using the initial population position and the SCR system control objective as the fitness function, the proportional gain of the PID controller is dynamically optimized online. Integral coefficient Differential coefficients ; PID control module: used to control the system based on the optimized... , , Calculate the urea injection quantity control signal to drive the urea injection system to perform the injection action; Feedback adjustment module: used to adjust the output NO of the SCR system. x The concentrations of NH3 and NH3 are fed back to the fitness function, forming a closed-loop optimization.

Images