Self-adaptive adjustment control system and method for NOx generation mechanism combustion experiment

Abstract

The invention relates to the technical field of combustion experiments, and relates to a self-adaptive adjustment control system and method for a NOx generation mechanism combustion experiment, and the method comprises the steps: inputting current experiment data, historical experiment data, a preset knowledge base and an experiment scheme inputted by a user; historical experimental data, experimental schemes and a preset knowledge base are input into a pre-constructed experimental element recognition model, key fuzzy variables, a fuzzy range set, regulation and control quantities and fuzzy rules are output, and an initialized fuzzy rule base is established; inputting current experimental data and an experimental scheme to a fuzzy optimization model output weight set and an activation sequence; determining an activated rule and an adjustment variable set according to the activation sequence; according to the initialized fuzzy rule base, the weight set and the adjustment variable set, performing demodeling to obtain an adjustment value of each adjustment variable; uncertain factors are processed through a fuzzy algorithm, a machine learning model is combined to optimize a rule base and a de-modeling method, and the adjustment precision of experimental parameters and the reliability of experimental results are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of combustion experiments, and more specifically, to an adaptive adjustment control system and method for NOx generation mechanism combustion experiments. Background Art

[0002] With the increasingly strict environmental protection requirements, the control of nitrogen oxides (NOx) during the combustion process has become particularly important. In the prior art, the control of NOx generation mostly relies on theoretical calculations and simple experiments, lacking a systematic and precise experimental control system. In this case, it is difficult to effectively simulate the actual combustion environment, thus unable to accurately study the NOx generation mechanism, affecting the development and application of pollutant reduction technologies.

[0003] The patent with the publication number CN117192022A discloses an integrated control system for an experimental device for high-voltage discharge ignition mechanism, which relates to the technical field of high-voltage ignition equipment and includes an ignition test module, a PCL controller, a data acquisition module, and a timing control module; the ignition test module is used to remotely control the test circuit and the experimental cavity, remotely operate switches, phase selection closing switches, circuit breaker closing, and the opening and closing of the test sample; the data acquisition module is used to collect the test combustion data of the experimental cavity; the PCL controller is used to calculate the combustion index based on the test combustion data and assist in determining the valve opening of the intake valve according to the combustion index to adjust the atmosphere in the experimental cavity and improve test safety; the timing control module is used to generate two mutually exclusive opening and closing signals from the high and low levels of the clock signal to control the test time; the invention remotely and precisely controls the required test circuit and sample switching, and the obtained test parameters are highly accurate; the system has good integration, high reliability, and strong practicability.

[0004] Although the above technology has achieved high accuracy of the obtained experimental parameters, there is still a problem of dealing with uncertain factors. All experiments have uncertain factors, and the ultimate effect pursued by the experiment needs to directly skip the uncertain factors and reach the experimental conclusion. Otherwise, the uncertain factors will interfere with the final experimental conclusion. Secondly, the experimental control system does not adjust the experimental parameters dynamically according to real-time data, which may affect the reliability of the experiment. Summary of the Invention

[0005] To solve the above problems, the present invention provides an adaptive adjustment control method for NOx generation mechanism combustion experiments, and the method includes:

[0006] Input the current experimental data, historical experimental data, preset knowledge base, and experimental scheme entered by the user;

[0007] Input historical experimental data, experimental schemes, and a preset knowledge base into a pre-built experimental element recognition model, output key fuzzy variables, a set of fuzzy ranges, control quantities, and fuzzy rules, and establish an initial fuzzy rule base;

[0008] Input current experimental data and experimental schemes into a fuzzy optimization model to output a weight set and an activation sequence;

[0009] Determine the activated rules and the set of adjustment variables according to the activation sequence;

[0010] Perform demodulation according to the initial fuzzy rule base, the weight set, and the set of adjustment variables to obtain the adjustment value of each adjustment variable;

[0011] Use the adjustment value of each adjustment variable to perform adaptive adjustment on the furnace.

[0012] Furthermore, the construction method of the experimental element recognition model includes:

[0013] Convert each set of historical experimental data, experimental schemes, and preset knowledge base into the form of a first feature vector;

[0014] Use the set of all first feature vectors as the input of the experimental element recognition model. The experimental element recognition model outputs the predicted key fuzzy variables, the set of fuzzy ranges, and fuzzy rules for each set of historical experimental data, experimental schemes, and preset knowledge base, and takes minimizing the sum of the first prediction accuracies of all key fuzzy variables, the set of fuzzy ranges, and fuzzy rules as the training objective; optimize the model parameters of the experimental element recognition model through the gradient descent algorithm, and obtain the model parameters of the experimental element recognition model with the minimum sum of the first prediction accuracies of all key fuzzy variables, the set of fuzzy ranges, and fuzzy rules to build the experimental element recognition model.

[0015] Furthermore, the combination of the key fuzzy variables, the set of fuzzy ranges, and the fuzzy rules is the initial fuzzy rule base; the initial fuzzy rule base is a set of language fuzzy definition rules.

[0016] Furthermore, the construction method of the fuzzy optimization model includes:

[0017] Convert each set of historical experimental data and experimental schemes into the form of a second feature vector;

[0018] Take the set of all second feature vectors as the input of the fuzzy optimization model. The fuzzy optimization model outputs the weight set and activation sequence predicted for each set of historical experimental data and experimental scheme. With minimizing the sum of the second prediction accuracies of all predicted weight sets and activation sequences as the training objective, optimize the model parameters of the fuzzy optimization model through the gradient descent algorithm, and obtain the model parameters of the fuzzy optimization model with the minimum sum of the second prediction accuracies of all predicted weight sets and activation sequences to build the fuzzy optimization model.

[0019] Further, the weight set is the set of weights in the weighted average method, that is, the set of membership degrees; the activation sequence is the sequence composed of the activated rules and the order of the activated rules; the calculation method of demodulation adopts the weighted average method, and according to the set of adjustment variables, determine the adjustment values of the adjustment variables to be calculated.

[0020] The adaptive adjustment control method for NOx generation mechanism combustion experiments further includes:

[0021] Input historical experimental data and experimental scheme into the pre-constructed window decision model, and output the size and number of time windows;

[0022] Perform data calibration and delay compensation on the adjustment values according to the size and number of time windows;

[0023] Conduct a stability test. If the test passes, calculate the adjustment duration to obtain the adjustment duration;

[0024] Receive the adjustment duration, and the NOx generation mechanism test bench continuously adjusts the adjustment variables according to the adjustment values within the adjustment duration until the adjustment duration ends.

[0025] Further, the construction method of the window decision model includes:

[0026] Take historical experimental data and experimental scheme as analysis data, pre-collect the time windows corresponding to c groups of analysis data, and convert the analysis data and the corresponding time windows into a corresponding set of third feature vectors;

[0027] Take each set of third feature vectors as the input of the window decision model. The window decision model outputs a set of predicted time windows corresponding to each set of analysis data, with minimizing the sum of the prediction errors of all analysis data as the training objective; optimize the model parameters of the window decision model through natural inspiration optimization, and obtain the model parameters of the window decision model with the minimum sum of the prediction errors of all analysis data to build the window decision model; the window decision model is a recurrent neural network model.

[0028] Further, the method for data calibration is the moving average method, and the smoothed adjustment value is obtained by calculating the average value of the original adjustment values within the time window.

[0029] Further, the method for stability test includes: starting from the smoothed adjustment value, gradually checking the change rate between adjacent data points. When the change rate continuously falls below the set threshold ∈, it is considered that the smoothed adjustment value has passed the test and entered the stable state, where ∈ is a preset value; the calculation method for the adjustment duration includes: in the sequence composed of multiple smoothed adjustment values, record the time point v when the first adjustment value reaches the stable state, then the adjustment duration is v + N, where N is the time unit.

[0030] The present invention also discloses an adaptive adjustment control system for the NOx generation mechanism combustion experiment. The system includes:

[0031] Collection module: used to input the current experimental data, historical experimental data, preset knowledge base, and experimental scheme entered by the user;

[0032] Fuzzy new - building module: used to input the historical experimental data, experimental scheme, and preset knowledge base into the pre - constructed experimental element identification model, output the key fuzzy variables, fuzzy range set, regulation quantity, and fuzzy rules, and establish the initial fuzzy rule base;

[0033] Fuzzy optimization module: used to input the current experimental data and experimental scheme into the fuzzy optimization model to output the weight set and activation sequence;

[0034] Activation confirmation module: used to determine the activated rules and adjustment variable set according to the activation sequence;

[0035] Defuzzification module: used to perform defuzzification according to the initial fuzzy rule base, weight set, and adjustment variable set to obtain the adjustment value of each adjustment variable;

[0036] Adjustment module: used to perform adaptive adjustment on the furnace using the adjustment value of each adjustment variable;

[0037] Window confirmation module: used to input the historical experimental data and experimental scheme into the pre - constructed window decision model to output the size and number of the time window;

[0038] Data calibration module: used to perform data calibration and delay compensation on the adjustment value according to the size and number of the time window;

[0039] Stability test module: used to perform stability test. If the test passes, calculate the adjustment duration to obtain the adjustment duration;

[0040] Adjustment and Optimization Module: Receives the adjustment duration. The NOx generation mechanism test bench continuously adjusts variables according to the adjustment value within the adjustment duration until the adjustment duration ends.

[0041] Technical effects and advantages of the adaptive adjustment control system and method for NOx generation mechanism combustion experiments provided by the present invention:

[0042] Technical effects:

[0043] By introducing fuzzy algorithms and machine learning techniques, the present invention processes various uncertain factors in NOx generation mechanism experiments, improving the accuracy of experimental control; using machine learning models to optimize the fuzzy rule base and demodulation methods enables the experimental control system to dynamically adjust experimental parameters according to real-time data, ensuring the reliability and accuracy of the experiments; through time calibration and data smoothing techniques, experimental data is calibrated and delay compensated, effectively reducing errors caused by experimental data time delay and ensuring the accuracy and continuity of experimental data; the system has a high degree of automation, reducing manual intervention and operation errors, and improving the efficiency and accuracy of the experiments.

[0044] Advantages:

[0045] By using fuzzy algorithms to process uncertain factors and combining machine learning models to optimize the rule base and demodulation methods, the present invention improves the adjustment accuracy of experimental parameters and the reliability of experimental results; through data calibration and delay compensation techniques, error accumulation is effectively reduced, ensuring the continuity and accuracy of experimental data; by adopting dynamic adjustment and real-time optimization techniques, the system can adapt to various changes in the experiment, improving the robustness of the system; the high degree of automation reduces the dependence on manual operations, saves labor costs, and improves the experimental efficiency. Description of the Drawings

[0046] Figure 1 It is a flowchart of the adaptive adjustment control method for the NOx generation mechanism combustion experiment in Embodiment 1;

[0047] Figure 2 It is a flowchart of the adaptive adjustment control method for the NOx generation mechanism combustion experiment in Embodiment 2;

[0048] Figure 3 It is a schematic connection diagram of the adaptive adjustment control system for the NOx generation mechanism combustion experiment in Embodiment 3. Detailed Embodiments

[0049] Next, in combination with the accompanying drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment 1

[0051] Please refer to Figure 1 As shown, the adaptive adjustment control method for the NOx generation mechanism combustion experiment in this embodiment includes:

[0052] Input the current experimental data, historical experimental data, preset knowledge base, and experimental scheme entered by the user;

[0053] Input the historical experimental data, experimental scheme, and preset knowledge base into the pre-constructed experimental element recognition model, output key fuzzy variables, fuzzy range sets, regulation quantities, and fuzzy rules, and establish an initial fuzzy rule base;

[0054] Input the current experimental data and experimental scheme into the fuzzy optimization model to output a weight set and an activation sequence;

[0055] Determine the activated rules and adjustment variable sets according to the activation sequence;

[0056] Perform demodulation according to the initial fuzzy rule base, weight set, and adjustment variable set to obtain the adjustment value of each adjustment variable;

[0057] Use the adjustment value of each adjustment variable to perform adaptive adjustment on the furnace.

[0058] The current experimental data is the real-time experimental data during the current experiment, and the historical experimental data is the experimental data recorded in the experimental log. When an experiment is completed, the experimental data is recorded in the experimental log; both the current experimental data and the historical experimental data include the NOx generation amount (unit: mg / m 3 ), excess air coefficient (unitless, usually in the range of 0.8 - 1.5), temperature (unit: degree Celsius), and reducing gas concentration (unit: % volume ratio, such as CO or H2); the NOx generation amount, reducing gas concentration, and excess air coefficient are obtained by the flue gas sampling device at the furnace outlet, and the temperature is obtained by the temperature sensor.

[0059] The experimental scheme is a variable control set by humans. For example, before the experiment starts, technicians design the experimental scheme as follows:

[0060] Experiment 1, under the same conditions, the influence of the concentration of reducing gas on the NOx generation amount.

[0061] Experiment 2: The effect of temperature on NOx generation under the same conditions.

[0062] The preset knowledge base can be downloaded from the Internet or manually set by a technician, for example:

[0063] The research results show that high temperature and strong reducing atmosphere can reduce NOx generation.

[0064] When the temperature is between 800-1000℃, the amount of NOx generated will be moderate; when it exceeds 1000℃ and the space velocity is too high, the amount of NOx will increase significantly.

[0065] When a certain concentration of reducing additives is introduced, NOx generation can be effectively controlled.

[0066] In a low air velocity environment, increasing the CC cycle can improve combustion efficiency while reducing NOx production. The CC cycle refers to the catalytic combustion cycle, in which a catalyst is used to promote the combustion reaction so that the fuel can be completely burned at a lower temperature.

[0067] A new flame cooling technology was used in an experiment to demonstratively lower the temperature and achieve ideal results.

[0068] The method for constructing the experimental element identification model includes:

[0069] Each set of historical experimental data, experimental scheme and preset knowledge base is converted into the form of the first eigenvector.

[0070] The set of all first feature vectors is used as the input of the experimental factor identification model, and the experimental factor identification model takes the predicted key fuzzy variables, fuzzy range sets and fuzzy rules for each group of historical experimental data, experimental schemes and preset knowledge bases as outputs, and takes minimizing the sum of the first prediction accuracies of all key fuzzy variables, fuzzy range sets and fuzzy rules as the training target. The model parameters of the experimental factor identification model are optimized by the gradient descent algorithm, and the model parameters of the experimental factor identification model with the minimum sum of the first prediction accuracies of all key fuzzy variables, fuzzy range sets and fuzzy rules are obtained to build the experimental factor identification model; wherein the first prediction accuracy is calculated by the square difference between the predicted key fuzzy variables, fuzzy range sets and fuzzy rules corresponding to the e-th group of historical experimental data, experimental schemes and preset knowledge bases and the actual key fuzzy variables, fuzzy range sets and fuzzy rules corresponding to the e-th group of historical experimental data, experimental schemes and preset knowledge bases; such as the first prediction accuracy M e =(G e -S e ) 2 , where M eis the first prediction accuracy, G e is the predicted key fuzzy variables, fuzzy range sets, and fuzzy rules corresponding to the e-th group of historical experimental data, experimental schemes, and preset knowledge bases, S e is the actual key fuzzy variables, fuzzy range sets, and fuzzy rules corresponding to the e-th group of historical experimental data, experimental schemes, and preset knowledge bases.

[0071] The combination of the key fuzzy variables, fuzzy range sets, and fuzzy rules is the initialized fuzzy rule base; the initialized fuzzy rule base is a set of linguistic fuzzy definition rules, for example:

[0072] Adjust the concentration of reducing gas (such as oxygen): If the temperature is "high", the oxygen concentration is "low", and the excess air coefficient is "low", then the concentration of reducing gas needs to be increased by 10%.

[0073] If the temperature is "medium", the concentration of reducing gas is "high", and the excess air coefficient is "medium", then the concentration of reducing gas needs to be decreased and the temperature needs to be reduced by 50°C.

[0074] Among them, "the concentration of reducing gas", "temperature", and "oxygen concentration" are key fuzzy variables, "low", "medium", and "high" are fuzzy range sets, "if '' then increase ''" and "if '' then decrease ''" are fuzzy rules, and "reduce the temperature by 50°C" and "increase the concentration of reducing gas by 10%" are adjustment amounts.

[0075] In the above text, "high", "medium", and "low" are preset by technicians according to the actual experimental situation and are not specifically defined here.

[0076] Example: The specific process of establishing the initialized fuzzy rule base is as follows:

[0077] 1. Input data

[0078] (1) Historical experimental data

[0079] The historical experimental data is from the experimental log and records the partial results of the past 10 combustion experiments.

[0080] The following is example data (simplified version, assuming 5 parameters are recorded for each experiment).

[0081]

[0082] (2) Experimental scheme

[0083] The experimental scheme is the variable control target set by technicians, for example:

[0084] Experimental objective: To study the effects of temperature and the concentration of reducing gas on the NOx generation amount.

[0085] Control variables: Adjust the temperature (T) and the concentration of reducing gas (R), and keep the excess air coefficient (A) relatively stable. "Stable" means that the preset value is controlled within the allowable error range.

[0086] (3) Preset knowledge base

[0087] The preset knowledge base contains the combustion experiment rules extracted from literature or the experience of technicians. For example:

[0088] "High temperature (>1000 °C) and low concentration of reducing gas (<3%) will cause a significant increase in the NOx generation amount."

[0089] "Increasing the concentration of reducing gas (>5%) at medium temperature (800 - 1000 °C) can effectively reduce the NOx generation amount."

[0090] "When the excess air coefficient is between 1.0 and 1.3, the combustion efficiency is relatively high and the NOx generation amount is moderate.

[0091] 2. Processing process of the experimental factor identification model

[0092] The experimental factor identification model is a machine learning-based model (such as a neural network or a decision tree), and its role is to extract key fuzzy variables, fuzzy range sets, and fuzzy rules from the input data. Assume that the model has been trained with historical data. The following are the processing steps:

[0093] (1) Data preprocessing: Convert historical experimental data, experimental schemes, and the preset knowledge base into feature vectors. For example:

[0094] Data of Experiment 1 is converted into a vector: [600, 950, 4, 1.1, 55].

[0095] The rules in the preset knowledge base are converted into structured text inputs.

[0096] (2) Model training objective: Minimize the error of the predicted key fuzzy variables, fuzzy range sets, and fuzzy rules (compared with the actual experimental rules).

[0097] (3) Model output: Identify key variables and generate fuzzy rules according to the input data.

[0098] 3. Output results

[0099] After being processed by the model, the following content is output:

[0100] (1) Key fuzzy variables

[0101] Temperature (T); Concentration of reducing gas (R); Excess air coefficient (A).

[0102] (2) Fuzzy range sets

[0103] Based on historical data and a preset knowledge base, the model defines a fuzzy range:

[0104] Temperature (T): Low: ≤800 °C, Medium: 800 - 1000 °C, High: >1000 °C.

[0105] Concentration of reducing gas (R): Low: ≤3%, Medium: 3 - 7%, High: >7%.

[0106] Excess air coefficient (A): Low: ≤1.0, Medium: 1.0 - 1.3, High: >1.3.

[0107] (3) Regulation amount

[0108] The regulation amount is the adjustment suggestion based on the experimental target, for example:

[0109] Temperature adjustment: ±30 °C, ±50 °C.

[0110] Concentration adjustment of reducing gas: ±1%, ±2%.

[0111] (4) Fuzzy rules

[0112] The model generates the following fuzzy rules according to the input data and the knowledge base:

[0113] If T is "High" and R is "Low" and A is "Medium", then increase the regulation amount of R by +2%.

[0114] If T is "Medium" and R is "Medium" and A is "Medium", then decrease the regulation amount of T by -30 °C.

[0115] If T is "High" and R is "Medium" and A is "High", then decrease the regulation amount of T by -50 °C.

[0116] If T is "Medium" and R is "High" and A is "Low", then increase the regulation amount of R by +1%.

[0117] 4. Establish an initial fuzzy rule base

[0118] Integrate the above outputs into an initial fuzzy rule base in the form of a set of linguistic fuzzy definition rules. The following is a specific example:

[0119] Rule 1: If the temperature is "High" (>1000 °C), the concentration of reducing gas is "Low" (≤3%), and the excess air coefficient is "Medium" (1.0 - 1.3), then increase the concentration of reducing gas by 2%.

[0120] Rule 2: If the temperature is "Medium" (800 - 1000 °C), the concentration of reducing gas is "Medium" (3 - 7%), and the excess air coefficient is "Medium" (1.0 - 1.3), then decrease the temperature by 30 °C.

[0121] Rule 3: If the temperature is "high" (>1000 °C), the concentration of reducing gas is "medium" (3-7%), and the excess air coefficient is "high" (>1.3), then reduce the temperature by 50 °C.

[0122] Rule 4: If the temperature is "medium" (800-1000 °C), the concentration of reducing gas is "high" (>7%), and the excess air coefficient is "low" (≤1.0), then increase the concentration of reducing gas by 1%.

[0123] 5. Verification

[0124] Take Experiment 4 (T = 900 °C, R = 5%, A = 1.2, NOx = 500 mg / m 3 ) in the historical experimental data as an example:

[0125] Matching rule: Rule 2 (T = "medium", R = "medium", A = "medium").

[0126] Regulation result: Reduce the temperature by 30 °C, and T is adjusted from 900 °C to 870 °C.

[0127] Expected effect: According to the historical trend, the NOx generation amount is expected to be reduced to about 460 mg / m 3 .

[0128] The construction method of the fuzzy optimization model includes:

[0129] Convert each group of historical experimental data into the form of the experimental scheme as the second feature vector.

[0130] Take the set of all second feature vectors as the input of the fuzzy optimization model. The fuzzy optimization model takes the weight set and activation sequence predicted for each group of historical experimental data and experimental scheme as the output, and takes minimizing the sum of the second prediction accuracies of all predicted weight sets and activation sequences as the training objective. Optimize the model parameters of the fuzzy optimization model through the gradient descent algorithm, and obtain the model parameters of the fuzzy optimization model with the minimum sum of the second prediction accuracies of all predicted weight sets and activation sequences to build the fuzzy optimization model; among them, the second prediction accuracy is calculated from the square difference between the predicted weight set and activation sequence corresponding to the gth group of historical experimental data and experimental scheme, and the actual weight set and activation sequence corresponding to the gth group of historical experimental data and experimental scheme; such as the first prediction accuracy FR g =(FT g -FY g ) 2 , where FR g is the second prediction accuracy, FT g is the predicted weight set and activation sequence corresponding to the gth group of historical experimental data and experimental scheme, and FY gFor the g-th group of historical experimental data, it is the set of actual weights corresponding to the experimental scheme and the activation sequence.

[0131] The set of weights is the set of weights in the weighted average method, that is, the set of membership degrees; the activation sequence is the sequence composed of the activated rules and the order of the activated rules. In the subsequent demodulation, the adjustment variable is the key fuzzy variable to be regulated, and the set of adjustment variables is the set of all adjustment variables.

[0132] It should be noted that demodulation is an operation with an order relationship and priority. One demodulation obtains an adjustment value of the adjustment variable to be calculated according to the set of adjustment variables. Also, because the present invention calculates the adjustment value based on a computer platform and not all rules are used, an activation code in the programming language is required to activate the rule program to achieve the demodulation operation.

[0133] The calculation method of the demodulation adopts the weighted average method. According to the set of adjustment variables, the adjustment value of the adjustment variable to be calculated is determined. For example, the calculation formula for the adjustment value Gh is:

[0134]

[0135] In the formula, μ i is the membership degree of the i-th activated rule, p i is the regulation amount of the i-th activated rule; i is the number of the activated rule; B is the total number of activated rules.

[0136] It should be noted that the calculation of demodulation is still applicable when the same adjustment variable appears simultaneously in multiple activated rules, without the need for repeated calculation, and Gh is only used as the calculation expression of the adjustment value and is not unique. Because there may be multiple adjustment variables in the activated rules, the calculation formulas for the adjustment values of all adjustment variables are still applicable.

[0137] To facilitate understanding of the solution of this embodiment, the following example is provided:

[0138] 1. Input data

[0139] (1) Current experimental data

[0140] Assume that the current experiment is in progress, and the real-time collected data is as follows:

[0141] NOx generation amount: 520mg / m 3 ; Temperature (T): 920 °C, concentration of reducing gas (R): 4.5%; excess air coefficient (A): 1.15.

[0142] (2) Experimental scheme

[0143] The experimental scheme is:

[0144] Objective: Reduce the NOx generation amount to 450 mg / m 3 or less.

[0145] Controlled variables: Adjust the temperature (T) and the concentration of reducing gas (R), and keep the excess air coefficient (A) stable.

[0146] 2. Initialize the fuzzy rule base (refer to the previous example)

[0147] Rule 1: If T is "high" (>1000 °C), R is "low" (≤3%), and A is "medium", then increase the regulation amount of R by +2%.

[0148] Rule 2: If T is "medium" (800 - 1000 °C), R is "medium" (3 - 7%), and A is "medium" (1.0 - 1.3), then decrease the regulation amount of T by -30 °C.

[0149] Rule 3: If T is "high" (>1000 °C), R is "medium" (3 - 7%), and A is "high" (>1.3), then decrease the regulation amount of T by -50 °C.

[0150] Rule 4: If T is "medium" (800 - 1000 °C), R is "high" (>7%), and A is "low" (≤1.0), then increase the regulation amount of R by +1%.

[0151] 3. Output of the fuzzy optimization model: Weight set and activation sequence

[0152] Input the current experimental data (T = 920 °C, R = 4.5%, A = 1.15) and the experimental plan into the fuzzy optimization model, and output the weight set and the activation sequence.

[0153] Fuzzy range matching:

[0154] T = 920 °C → "medium" (800 - 1000 °C);

[0155] R = 4.5% → "medium" (3 - 7%);

[0156] A = 1.15 → "medium" (1.0 - 1.3);

[0157] Weight set (the membership degree is given by a simple assignment example, and in practice, the membership function can be used for calculation, such as the triangular membership function):

[0158] Rule 1: Membership degree 0.1 (T is not "high", R is not "low", A is "medium");

[0159] Rule 2: Membership degree 0.9 (T is "medium", R is "medium", A is "medium");

[0160] Rule 3: Membership degree 0.2 (T is not "high", R is "medium", A is not "high");

[0161] Rule 4: Membership degree 0.3 (T is "medium", R is not "high", A is not "low");

[0162] Activation sequence:

[0163] Activated rule: Rule 2 (highest membership degree, highest priority).

[0164] Order: Only activate Rule 2 (the membership degrees of other rules are lower and do not reach the activation threshold, assuming the threshold is 0.5).

[0165] 4. Determine the activated rule and the set of adjusted variables according to the activation sequence

[0166] Activated rule: Rule 2 (if T is "medium", R is "medium", A is "medium", then reduce the control amount of T by -30°C).

[0167] Set of adjusted variables: {T} (Rule 2 only involves the adjustment of temperature T).

[0168] 5. Demodulate according to the initialized fuzzy rule base, weight set and set of adjusted variables

[0169] The weighted average method is used for demodulation to calculate the adjustment value of each adjusted variable:

[0170] Adjusted variable: Temperature (T);

[0171] Control amount and membership degree:

[0172] Rule 2: Control amount -30°C, membership degree 0.9

[0173] No other rules are activated, only Rule 2 is used.

[0174] Calculation of adjustment value:

[0175]

[0176] Result: The temperature adjustment value is a decrease of 30°C.

[0177] 6. Use the adjustment value to perform adaptive adjustment on the furnace

[0178] Adjustment operation: The furnace reduces the temperature from 920°C to 890°C through the cooling device.

[0179] Adjusted data (assuming real-time acquisition after adjustment):

[0180] Temperature (T): 890°C

[0181] NOx generation amount: 470 mg / m 3(Reduce by 50 mg / m 3 , approaching the target).

[0182] Adaptive verification: The NOx generation amount decreases from 520 mg / m 3 to 470 mg / m 3 , with significant effects; if the target (450 mg / m 3 ) is not reached, the next round of optimization can be entered.

[0183] Multivariable adjustment expansion example

[0184] If T and R need to be adjusted simultaneously, assuming activation of Rule 2 and another rule (such as a slightly modified Rule 1), the process is as follows:

[0185] Modify Rule 1: If T is "medium", R is "medium", and A is "medium", then increase R by 1%.

[0186] Activation sequence: Rule 2 (membership degree 0.9), Rule 1 (membership degree 0.8).

[0187] Order: Rule 2 → Rule 1.

[0188] Adjustment variable set: {T, R}.

[0189] Demodulation:

[0190]

[0191] Adjustment: T is reduced from 920 °C to 890 °C, and R is increased from 4.5% to 5.5%.

[0192] Example 2

[0193] As Figure 2As shown in the figure, this embodiment further improves the design on the basis of Embodiment 1. The difference is that for the adaptive adjustment control method of the NOx generation mechanism combustion experiment provided in Embodiment 1, during actual operation, according to the adjustment variable set, the adjustment value of the adjustment variable to be calculated is determined. The adjustment value cannot reach instantaneously and there is a time consumption. Also, due to the closed nature of the combustion experiment, it is very difficult for the outside world to directly observe whether the adjustment of the adjustment variable is completed and it can only be measured by external instruments. Therefore, there will be a time difference between the technician and the furnace. The internal adjustment has not been in place yet, but the instrument on the NOx generation mechanism experimental bench already shows that the adjustment value has been reached and the next experiment can be carried out. For example, when adjusting the concentration of reducing gas and releasing reducing gas into the furnace, at the moment when the flue gas sampling device at the furnace outlet detects that the concentration of reducing gas meets the standard, in fact, the inside of the furnace may not have fully met the experimental requirements because the spread of the gas takes time and it is necessary to wait for a certain time for the concentration of reducing gas in each space inside to fully meet the standard. And when the technician substitutes the experimental data of the entire furnace into the experiment, errors will occur. Generally, the experiments are continuous experiments and error accumulation will occur after the next experiment. When the error accumulates to a certain amount, it will ultimately lead to data distortion. Based on this, the adaptive adjustment control method of the NOx generation mechanism combustion experiment provided in this example further includes:

[0194] Input historical experimental data and experimental scheme into a pre-constructed window decision model, and output the size and number of time windows;

[0195] Perform data calibration and delay compensation on the adjustment value according to the size and number of time windows;

[0196] Conduct a stability test. If the test passes, calculate the adjustment duration to obtain the adjustment duration;

[0197] Receive the adjustment duration, and the NOx generation mechanism experimental bench continuously adjusts the adjustment variable according to the adjustment value within the adjustment duration until the adjustment duration ends.

[0198] The construction method of the window decision model includes:

[0199] Take historical experimental data and experimental scheme as analysis data, pre-collect the time windows corresponding to c groups of analysis data, and convert the analysis data and the corresponding time windows into a corresponding set of third feature vectors.

[0200] Use each group of third feature vectors as the input of the window decision model. The window decision model outputs a group of prediction time windows corresponding to each group of analysis data. The prediction time window is the time window pre-collected corresponding to the analysis data. Minimize the sum of prediction errors of all analysis data as the training objective. Optimize the model parameters of the window decision model through natural inspiration optimization to obtain the model parameters of the window decision model with the minimum sum of prediction errors of all analysis data, and build the window decision model. The window decision model is a recurrent neural network model.

[0201] The method of data calibration is the moving average method. Obtain the smoothed adjustment value by calculating the average value of the original adjustment values within the time window. If the smoothed adjustment value In the formula, x t-a represents the (t - a)-th original adjustment value, N is the size of the moving average time window, a is the number of the original adjustment value, and t is the number of time windows.

[0202] The method of stability test includes: starting from the smoothed adjustment value, gradually check the change rate between adjacent data points. When the change rate continuously remains lower than the set threshold ∈, it is considered that the smoothed adjustment value has passed the test and entered the stable state. ∈ is a preset value, which is set by those skilled in the art according to the actual situation. For example, "if the furnace volume increases to 2m 3 , ∈ can be adjusted to 0.3℃ / s").

[0203] The calculation method of the adjustment duration is as follows: in the sequence composed of multiple smoothed adjustment values, record the time point v when the first adjustment value reaches the stable state. Then the adjustment duration is v + N, where N is the time unit. This can ensure that the data in the adjustment process has tended to be stable and is suitable for the next experimental operation.

[0204] Embodiment 3

[0205] As Figure 3 shown, the adaptive adjustment control system for the NOx generation mechanism combustion experiment. The system includes:

[0206] Collection module: used to input the current experimental data, historical experimental data, preset knowledge base and experimental scheme entered by the user;

[0207] Fuzzy new building module: used to input the historical experimental data, experimental scheme and preset knowledge base into the pre-constructed experimental element recognition model, output key fuzzy variables, fuzzy range sets, control quantities and fuzzy rules, and establish an initial fuzzy rule base;

[0208] Fuzzy optimization module: used to input the current experimental data and experimental scheme into the fuzzy optimization model to output the weight set and activation sequence;

[0209] Activation confirmation module: used to determine the activated rules and adjustment variable sets according to the activation sequence;

[0210] Defuzzification module: used to perform defuzzification based on the initialized fuzzy rule base, weight set, and adjustment variable sets to obtain the adjustment values of each adjustment variable;

[0211] Adjustment module: used to adaptively adjust the furnace using the adjustment values of each adjustment variable;

[0212] Window confirmation module: used to input historical experimental data and experimental plans into a pre-built window decision model and output the size and number of time windows;

[0213] Data calibration module: used to perform data calibration and delay compensation on the adjustment values according to the size and number of time windows;

[0214] Stability test module: used to conduct a stability test. If the test passes, calculate the adjustment duration to obtain the adjustment duration;

[0215] Adjustment optimization module: receives the adjustment duration. The NOx generation mechanism test bench continuously adjusts the adjustment variables according to the adjustment values within the adjustment duration until the adjustment duration ends.

[0216] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claimed rights.

[0217] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should all be included within the protection scope of the present invention.

Claims

1. Adaptive adjustment control method for NOx generation mechanism combustion experiments. The NOx generation mechanism combustion experiments are carried out on an experimental bench with a furnace. It is characterized in that the method Including: Input the current experimental data, historical experimental data, preset knowledge base and experimental scheme entered by the user; Input the historical experimental data, experimental scheme and preset knowledge base into the pre-constructed experimental element recognition model, output the key fuzzy variables, fuzzy range set, regulation quantity and fuzzy rules, and establish the initial fuzzy rule base; Input the current experimental data and experimental scheme into the fuzzy optimization model to output the weight set and activation sequence; Determine the activated rules and adjustment variable set according to the activation sequence; Perform demodulation according to the initial fuzzy rule base, weight set and adjustment variable set to obtain the adjustment value of each adjustment variable; Use the adjustment value of each adjustment variable to perform adaptive adjustment on the furnace.

2. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 1, wherein The construction method of the experimental element recognition model includes: Convert each group of historical experimental data, experimental scheme and preset knowledge base into the form of the first feature vector; Use the set of all the first feature vectors as the input of the experimental element recognition model. The experimental element recognition model outputs the predicted key fuzzy variables, fuzzy range set and fuzzy rules for each group of historical experimental data, experimental scheme and preset knowledge base. Taking the minimization of the sum of the first prediction accuracies of all the key fuzzy variables, fuzzy range set and fuzzy rules as the training objective, optimize the model parameters of the experimental element recognition model through the gradient descent algorithm, and obtain the model parameters of the experimental element recognition model with the minimum sum of the first prediction accuracies of all the key fuzzy variables, fuzzy range set and fuzzy rules to build the experimental element recognition model.

3. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 1, characterized in that The construction method of the fuzzy optimization model includes: Convert each group of historical experimental data and experimental scheme into the form of the second feature vector; Use the set of all the second feature vectors as the input of the fuzzy optimization model. The fuzzy optimization model outputs the predicted weight set and activation sequence for each group of historical experimental data and experimental scheme. Taking the minimization of the sum of the second prediction accuracies of all the predicted weight sets and activation sequences as the training objective, optimize the model parameters of the fuzzy optimization model through the gradient descent algorithm, and obtain the model parameters of the fuzzy optimization model with the minimum sum of the second prediction accuracies of all the predicted weight sets and activation sequences to build the fuzzy optimization model.

4. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 3, characterized in that, The weight set is the set of weights in the weighted average method, that is, the set of membership degrees; the activation sequence is the sequence composed of the activated rules and the order of the activated rules; the calculation method of demodulation adopts the weighted average method, and according to the adjustment variable set, determine the adjustment value of the adjustment variable to be calculated.

5. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 1, characterized in that, The method further includes: Input the historical experimental data and experimental scheme into the pre-constructed window decision model, and output the size and number of time windows; Perform data calibration and delay compensation on the adjustment value according to the size and number of time windows; Perform a stability test. If the test passes, calculate the adjustment duration to obtain the adjustment duration; Receive the adjustment duration, and the NOx generation mechanism test bench continuously adjusts the adjustment variable according to the adjustment value within the adjustment duration until the adjustment duration ends.

6. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 5, characterized in that The construction method of the window decision model includes: Taking historical experimental data and experimental schemes as analysis data, pre-collecting the time windows corresponding to c groups of analysis data, and converting the analysis data and the corresponding time windows into a corresponding set of third feature vectors; Taking each set of third feature vectors as the input of the window decision model, the window decision model taking a set of predicted time windows corresponding to each set of analysis data as the output, and taking minimizing the sum of the prediction errors of all analysis data as the training objective; optimizing the model parameters of the window decision model through nature-inspired optimization, obtaining the model parameters of the window decision model with the smallest sum of the prediction errors of all analysis data, and building the window decision model; the window decision model is a recurrent neural network model.

7. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 6, characterized in that, The method for data calibration is the moving average method, and the smoothed adjustment value is obtained by calculating the average value of the original adjustment values within the time window.

8. The adaptive adjustment control method for the NOx generation mechanism combustion experiment according to claim 7, characterized in that The method for stability test includes: starting from the smoothed adjustment value, gradually checking the change rate between adjacent data points. When the change rate continuously remains lower than the set threshold ∈, it is considered that the smoothed adjustment value has passed the test and entered the stable state, where ∈ is a preset value; the calculation method for the adjustment duration includes: in the sequence composed of multiple smoothed adjustment values, recording the time point v when the first adjustment value reaches the stable state, then the adjustment duration is v + N, where N is the time unit.

9. An adaptive adjustment control system for combustion experiments on NOx generation mechanism, characterized in that, Implementing the adaptive adjustment control method for the NOx generation mechanism combustion experiment according to any one of claims 1-8, the system includes: Collection module: used for inputting the current experimental data, historical experimental data, preset knowledge base and experimental scheme entered by the user; Fuzzy creation module: used for inputting historical experimental data, experimental scheme and preset knowledge base into the pre-constructed experimental element recognition model, outputting key fuzzy variables, fuzzy range sets, regulation quantities and fuzzy rules, and establishing an initial fuzzy rule base; Fuzzy optimization module: used for inputting the current experimental data and experimental scheme into the fuzzy optimization model to output a weight set and an activation sequence; Activation confirmation module: used for determining the activated rules and adjustment variable sets according to the activation sequence; Defuzzification module: used for defuzzification according to the initial fuzzy rule base, weight set and adjustment variable sets to obtain the adjustment value of each adjustment variable; Adjustment module: used for adaptively adjusting the furnace chamber by using the adjustment values of each adjustment variable; Window confirmation module: used for inputting historical experimental data and experimental scheme into the pre-constructed window decision model, and outputting the size and number of time windows; Data calibration module: used for calibrating the adjustment values and compensating for delays according to the size and number of time windows; Stability test module: used for performing a stability test, and if the test passes, calculating the adjustment duration to obtain the adjustment duration; Adjustment optimization module: receiving the adjustment duration, the NOx generation mechanism test bench continuously adjusts the adjustment variables according to the adjustment values within the adjustment duration until the adjustment duration ends.

Images