CN-121785146-B - Biomass gas coupling unit optimal control method and system based on digital twin

CN121785146BCN 121785146 BCN121785146 BCN 121785146BCN-121785146-B

Abstract

The invention relates to the technical field of biomass gas utilization, in particular to a biomass gas coupling unit optimizing control method and system based on digital twinning, wherein the method comprises the steps of obtaining operation data of a biomass gas coupling gas unit after entering a stable operation interval in a historical operation process, obtaining an original data set, based on the operation data, the collected original data set is screened, the data interference of the start-stop stage and the load mutation stage is removed, an effective operation data set is formed, effective response training data is constructed based on the effective operation data set, a training data set is formed, and a machine learning model for predicting gas supply flow is trained. According to the invention, only the operation data of the gas turbine set after entering the stable operation interval is selected, and the synchronous acquisition mechanism of the rotation speed maintaining state and the output power is introduced, so that the formed data set can truly reflect the actual response characteristics of the gas turbine set to the gas quality and the gas supply capacity under the continuous stable working condition.

Inventors

Jie Xiaochen
FAN QINSHAN
FU YIWEI
Cheng Manqiu
ZHENG SHIJIN
DING HONG
ZHOU FEI
ZHANG PENG
Ye Xingpei
LIU MINGRUI
LI CUNLEI
LIU ZHOU

Assignees

江苏省国信研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260305

Claims (9)

1. The biomass gas coupling unit optimization control method based on digital twinning is characterized by comprising the following steps of: Acquiring operation data of the biomass gas coupling gas unit after entering a stable operation interval in a historical operation process, and acquiring an original data set, wherein the original data set comprises gas engine output power, gas engine rotating speed, gas supply flow, biomass gas state parameters and biomass gas flow; screening the collected original data set based on the operation data, and eliminating the data interference of the start-stop stage and the load mutation stage to form an effective operation data set; the method comprises the steps of constructing the output power of a gas engine, the state parameters of biomass gas, the biomass gas flow and the gas supply flow in an effective operation data set into effective response training data to form a training data set, and training a predicted gas supply flow machine learning model; Acquiring the current output power of the gas engine, the state parameters of the biomass gas and the biomass gas flow, predicting the gas supply flow based on the trained machine learning model, and controlling the gas supply flow of the gas unit; Under experimental conditions, acquiring gas state parameters of different biomass and gas engine stability risk indexes under participation proportion, and training a steady state discrimination machine learning model based on threshold labeling; inputting the current biomass gas state parameters, the biomass gas flow and the predicted gas supply flow into the steady state judgment machine learning model to obtain a judgment result of whether the current biomass gas participation ratio meets the steady operation requirement; when the judging result is not up to standard, reversely searching the participation proportion of the biomass gas along the reducing direction, judging the participation proportion results through the steady state judging machine learning model, obtaining the maximum participation proportion of the biomass gas when the judging result is up to standard, and obtaining the corresponding ideal biomass gas flow; Based on ideal biomass gas flow, biomass gas state parameters and current gas engine output power, invoking a predicted gas supply flow machine learning model to obtain a gas supply flow prediction result under a stability constraint condition; jointly acting the predicted gas supply flow and the corresponding biomass gas flow on a gas unit; The method for acquiring the stability risk index comprises the following steps: in a preset sliding time window, carrying out statistical analysis on the rotating speed data of the gas engine, and calculating the rotating speed fluctuation intensity characteristic quantity; Acquiring time variation data of the combustion temperature in a preset sliding time window, and calculating a characteristic quantity of a combustion response variation rate; And normalizing the rotating speed fluctuation intensity characteristic quantity and the combustion response change rate characteristic quantity, and then carrying out weighted summation to obtain the stability risk index.
2. The method for optimizing control of a digitally twinned biomass gas coupling unit according to claim 1, wherein the method for screening the collected raw data set to form an effective operational data set comprises: Calculating the change rate of the output power of the gas engine in a continuous time window by presetting the length of the time window, and setting an output power stability criterion, namely judging that the output power in the time window is in a stable state when the change rate of the output power is lower than a preset threshold value, judging that the corresponding data is not included in a stable operation interval when the change rate of the output power exceeds the preset threshold value; Calculating a rotation speed maintaining coefficient of the gas engine in a preset time window, and setting a rotation speed maintaining stability criterion to be that when the rotation speed maintaining coefficient is smaller than a preset rotation speed maintaining coefficient threshold value, the gas engine is judged to be in a stable maintaining state; And carrying out joint judgment on the output power stability criterion and the rotation speed maintenance stability criterion, and judging the data corresponding to the time window as effective operation data only when the two types of criteria are simultaneously met in the same time window.
3. The method for optimizing control of the biomass gas coupling unit based on digital twin according to claim 2 is characterized in that the method for obtaining the rotation speed maintenance coefficient comprises the steps of obtaining deviation between a time average value of actual rotation speeds of the gas engine and target rotation speeds set by the gas engine in a preset time window length, obtaining a deviation absolute value, obtaining standard deviation of rotation speeds of the gas engine in the preset time window length, obtaining a ratio of the deviation absolute value to the target rotation speeds set by the gas engine and a ratio of the standard deviation to the target rotation speeds set by the gas engine, and then carrying out weighted summation on the two ratios to obtain the rotation speed maintenance coefficient.
4. The method for optimized control of a digital twin-based biomass gas coupling unit according to claim 1, wherein the effective response training data includes gas engine output power, biomass gas state parameters, biomass gas flow rate and corresponding gas supply flow rate.
5. The optimized control method of the biomass gas coupling unit based on digital twin according to claim 4, wherein the biomass gas state parameters comprise an effective heat value coefficient and a gas component duty ratio; the method for collecting the biomass gas state parameters comprises the following steps: On-line data acquisition is carried out on gasification gas production before entering a gas unit at the gas outlet end of the biomass gasification system, and the acquired data comprise gas flow, volume fractions of carbon monoxide, hydrogen and methane, and volume fractions of carbon dioxide and nitrogen, so that complete gas component ratio data are formed; And based on the gas component duty ratio data, carrying out weighted summation according to the standard low-order heat value of each combustible component and the volume fraction thereof to obtain the effective heat value coefficient of the gas.
6. The method for optimizing control of a biomass gas coupling unit based on digital twin according to claim 1 or 5, wherein the training method of the predicted gas supply flow machine learning model comprises the following steps: Layering according to the operation working conditions of the gas engine, adopting layering random sampling, and dividing an effective operation data set into a training set, a verification set and a test set according to a preset proportion, so that the working conditions are uniformly distributed in the training set, the verification set and the test set; selecting a gradient lifting regression tree as a fuel gas supply flow prediction model, setting initial super parameters, initializing the model, setting an initial prediction value of the gradient lifting regression tree model as a mean value of fuel gas supply flows of a training set, calculating a difference value between the current model predicted fuel gas supply flow and the actual flow, namely residual error, aiming at each sample in the training set, constructing a new decision tree, taking the minimum residual error as a target, selecting optimal splitting characteristics and splitting points based on mean square error, and dividing the samples to different sub-nodes until a preset stop condition is met; Searching for an optimal super-parameter combination in a preset range by adopting a Bayes optimization method; And calculating the root mean square error of the verification set every 20 iterations, stopping training when the root mean square error reduction amplitude of the verification set is smaller than 0.001 for 3 continuous iterations, and storing the model parameter with the minimum root mean square error of the verification set in the training process.
7. The optimized control method for the biomass gas coupling unit based on digital twin according to claim 1, wherein the supply amount of the fuel gas in the fuel gas unit is controlled based on the predicted fuel gas supply flow rate when the determination result reaches the standard.
8. The optimized control method of the biomass gas coupling unit based on the digital twin system according to claim 1 is characterized in that the method based on the threshold value labeling comprises the steps of setting the participation ratio of biomass gas to reach the standard and setting a label to be 1 when the stability risk index of the gas unit is lower than a stability risk index threshold value; When the stability risk index of the gas unit is greater than or equal to the stability risk index threshold, the biomass gas participation proportion is set to be not up to the standard, and the label is set to be 0.
9. A digital twin-based biomass gas coupling unit optimization control system for implementing the digital twin-based biomass gas coupling unit optimization control method according to any one of claims 1 to 8, characterized by comprising: The system comprises an original operation data acquisition module, a biomass gas coupling gas unit and a control module, wherein the original operation data acquisition module acquires operation data of the biomass gas coupling gas unit after entering a stable operation interval in a historical operation process, and acquires an original data set, and the original data set comprises gas engine output power, gas engine rotating speed, gas supply flow, biomass gas state parameters and biomass gas flow; The effective operation data generation module is used for screening the collected original data set based on the operation data, eliminating the data interference of the start-stop stage and the load mutation stage and forming an effective operation data set; the training data construction module is used for constructing the output power of the gas engine, the state parameters of the biomass gas, the biomass gas flow and the gas supply flow in the effective operation data set into effective response training data to form a training data set, and training and predicting a gas supply flow machine learning model; The gas supplementing flow prediction module is used for obtaining the current output power of the gas engine, the state parameters of the biomass gas and the biomass gas flow, predicting the gas supply flow based on the trained machine learning model and controlling the gas supply flow of the gas unit; the model training module is used for acquiring stability risk indexes of the gas turbine unit under the experimental conditions and different biomass gas state parameters and different biomass gas participation proportions, constructing labeling rules of up-to-standard and up-to-failure of the biomass gas participation proportions based on a preset stability risk index threshold, and training to obtain a steady state judging machine learning model for judging the steady state of the gas turbine unit; the stability judging module inputs the current biomass gas state parameter, the biomass gas flow and the predicted gas supply flow into the steady state judging machine learning model to obtain a judging result of whether the current biomass gas participation ratio meets the steady operation requirement; The reverse search module is used for carrying out reverse search on the participation proportion of the biomass gas along the reducing direction when the judging result is not up to the standard, judging each participation proportion result through the steady state judging machine learning model, obtaining the maximum participation proportion of the biomass gas when the judging result is up to the standard, obtaining the biomass gas flow corresponding to the current maximum participation proportion of the biomass gas, and marking the biomass gas flow as ideal biomass gas flow; the fuel gas supply prediction module is used for calling a predicted fuel gas supply flow machine learning model based on ideal biomass gas flow, biomass gas state parameters and current fuel gas engine output power to obtain a fuel gas supply flow prediction result under a stability constraint condition; and the control execution module jointly acts the predicted gas supply flow and the corresponding biomass gas flow on the gas unit.

Description

Biomass gas coupling unit optimal control method and system based on digital twin Technical Field The invention relates to the technical field of biomass gas utilization, in particular to a biomass gas coupling unit optimization control method and system based on digital twinning. Background Along with the continuous improvement of the specific gravity of biomass energy in a renewable energy system, the biomass gasification power generation and the technical path for the coupling operation of the biomass gasification power generation and a gas turbine set gradually become an important way for realizing clean and high-efficiency energy conversion. The combustible gas generated by the biomass gasification system is introduced into the gas unit for energy utilization, so that the utilization efficiency of biomass resources can be improved to a certain extent, and the dependence on fossil energy is reduced. However, in practical engineering application, the biomass gasification process is obviously influenced by the ingredient difference, the water content fluctuation and the reaction condition change of raw materials, so that the generated combustible gas has poor heat value and component stability, and the safety and stability operation of a subsequent gas unit are challenged. In the prior art, the existing biomass gas coupling gas unit control technology mostly adopts an open-loop or weak closed-loop control mode based on an empirical threshold value or a single working condition parameter, usually uses the output power of a gas engine as a core adjustment object, lacks modeling of an internal coupling relation between the quality fluctuation of the biomass gas and the operation stability of the gas engine, and training data are often mixed with data of a start-stop stage and a load mutation stage, so that the generalization capability and engineering applicability of the model are insufficient. Disclosure of Invention In order to solve the problems, the invention provides an optimized control method of a biomass gas coupling unit based on digital twinning. The method comprises the following steps of obtaining operation data of a biomass gas coupling gas unit after entering a stable operation interval in a historical operation process based on a digital twin biomass gas coupling unit optimization control method, and obtaining an original data set, wherein the original data set comprises gas engine output power, gas engine rotating speed, gas supply flow, biomass gas state parameters and biomass gas flow; screening the collected original data set based on the operation data, and eliminating the data interference of the start-stop stage and the load mutation stage to form an effective operation data set; the method comprises the steps of constructing the output power of a gas engine, the state parameters of biomass gas, the biomass gas flow and the gas supply flow in an effective operation data set into effective response training data to form a training data set, and training a predicted gas supply flow machine learning model; and acquiring the current gas engine output power, biomass gas state parameters and biomass gas flow, predicting gas supply flow based on the trained machine learning model, and controlling the gas supply flow of the gas unit. As further description of the technical scheme, the biomass gas coupling unit optimization control method based on digital twinning further comprises the following steps: Under experimental conditions, acquiring gas state parameters of different biomass and gas engine stability risk indexes under participation proportion, and training a steady state discrimination machine learning model based on threshold labeling; inputting the current biomass gas state parameters, the biomass gas flow and the predicted gas supply flow into the steady state judgment machine learning model to obtain a judgment result of whether the current biomass gas participation ratio meets the steady operation requirement; when the judging result is not up to standard, reversely searching the participation proportion of the biomass gas along the reducing direction, judging the participation proportion results through the steady state judging machine learning model, obtaining the maximum participation proportion of the biomass gas when the judging result is up to standard, and obtaining the corresponding ideal biomass gas flow; Based on ideal biomass gas flow, biomass gas state parameters and current gas engine output power, invoking a predicted gas supply flow machine learning model to obtain a gas supply flow prediction result under a stability constraint condition; And jointly acting the predicted gas supply flow and the corresponding biomass gas flow on the gas unit. As a further description of the above technical solution, the method for screening the collected original data set to form the effective operation data set includes: Calculating the change rate of the output power of the gas en