A superheated steam temperature predictive control method based on deep learning model

Abstract

The present invention relates to a superheated steam temperature predictive control method based on a deep learning model. This method establishes a deep learning model of a boiler's superheated steam temperature system. Based on this model, it calculates the future superheated steam temperature changes under different control variables. The control variable corresponding to the optimal control result is selected as the controller output to control the superheated steam temperature, thereby improving the unit's superheated steam temperature operational stability. Compared with existing technologies, this method offers advantages such as simplified engineering application calculations and further improved superheated steam temperature control performance.

Description

Technical Field

[0001] The present invention relates to the field of thermal intelligent control and protection, and in particular to a superheated steam temperature prediction and control method based on a deep learning model. Background Art

[0002] With the deepening of my country's economic restructuring and transformation and upgrading, as well as the vigorous implementation of power system reform, smart power plants, driven by policies such as energy conservation, consumption reduction, and emission reduction, have integrated technologies such as the Internet, big data, and artificial intelligence to promote smart operation management, smart maintenance safety, and intelligent control. These measures have become the main trend in the development of power plants. The commissioning of smart power plants can effectively enhance the core competitiveness of power plants and promote their sustainable development.

[0003] Traditional superheated steam temperature control typically utilizes control methods such as cascaded PID, Smith, state observers, and multi-model predictive control. Due to the large delay inertia of the superheated steam temperature object, numerous disturbance factors, and significant variations in its characteristics under different load conditions, these control methods are generally only able to address superheated steam temperature control under specific disturbances or operating conditions, and are unable to address control issues under full load conditions and a comprehensive range of disturbance factors. With the rapid development of intelligent control and data mining technologies, the application of deep learning technology to complex object modeling and prediction for thermal power units has become an important research and application area.

[0004] Because deep learning models combine low-level features into more abstract, high-level features, they are well-suited for the complex, high-dimensional, nonlinear, time-varying, and massively data-intensive applications of thermal power plants. However, due to their high dimensionality and nonlinear mathematical nature, they cannot be directly used as predictive models for optimal control solutions. Intelligent optimization algorithms are typically used, but due to the computational complexity of deep learning models and optimization algorithms, they cannot meet the demands of engineering applications. Currently, deep learning models are primarily used in the actual engineering control of complex thermal power plants using early warning or intelligent feedforward control methods, making it difficult to easily embed deep learning models into control solutions and fully utilize their predictive control capabilities. Summary of the Invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a superheated steam temperature prediction and control method based on a deep learning model.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] According to one aspect of the present invention, a superheated steam temperature prediction and control method based on a deep learning model is provided. The method establishes a deep learning model of the boiler superheated steam temperature system, calculates the changes in superheated steam temperature at future moments under different control quantities based on the model, and selects the control quantity corresponding to the optimal control result as the controller output to perform superheated steam temperature control, thereby improving the superheated steam temperature operating stability of the unit.

[0008] As a preferred technical solution, the method specifically includes the following steps:

[0009] Step S1: Using relevant parameters of the boiler pulverizing system, combustion system, and air and smoke system as feature input parameters, selecting historical operating data and using a recurrent neural network to train a deep learning model to predict future changes in the boiler superheated steam temperature;

[0010] Step S2, reading the parameter values of the characteristic variables from the DCS system in real time and establishing a dynamic data storage matrix queue;

[0011] Step S3: extracting the input data matrix of the model at the current moment from the dynamic data storage matrix queue according to the sampling period of the deep learning model training data and the model time compensation;

[0012] Step S4: Input the data matrix in step S3 into the deep learning model to calculate the change of the boiler superheated steam temperature in the next 30s, 60s, 90s, 120s, 150s and 180s respectively;

[0013] Step S5: Obtain the time series of the future superheated steam temperature setting value Compare the data with the six future superheated steam temperature change data predicted in step S4 to perform predictive control;

[0014] Step S6: determine the change limit of the cooling water flow rate p2 within a control cycle [-Δp a ,Δp b ], in the superheated steam temperature prediction control, the new set desuperheating water flow rate Instead of the actual desuperheating water flow rate p2(k) at the current moment, re-calculate the model prediction to obtain the superheated steam temperature change value at the future moment

[0015] Step S7, calculate the difference The superheated steam temperature predictive control performance index J under input i ;

[0016] Step S8, select the minimum J i Corresponding The superheated steam temperature is controlled as the feed water flow rate instruction at the next moment.

[0017] As a preferred technical solution, in step S1, six models are established to predict the changes of boiler superheated steam temperature in the next 30s, 60s, 90s, 120s, 150s and 180s.

[0018] As a preferred technical solution, the dynamic data storage matrix queue in step S2 is specifically:

[0019]

[0020] Where k represents the current moment, p j (k) is the parameter of the jth characteristic variable at time k, j = 1, 2, ..., n, data t is the sampling period, and d is the time step.

[0021] As a preferred technical solution, in step S2, after reading the real-time data of the characteristic variables, a shift operation is performed on the dynamic data storage matrix, and the real-time data is assigned to the parameters corresponding to the current moment.

[0022] As a preferred technical solution, in step S3, the input data matrix of the model at the current moment is specifically:

[0023]

[0024] Where k represents the current moment, p j (k) is the parameter of the jth characteristic variable at time k, j = 1, 2, ..., n, data t is the sampling period, and d is the time step.

[0025] As a preferred technical solution, in step S6, the newly set cooling water flow rate The specific calculation is as follows:

[0026]

[0027] Δp i =min(-Δp a +i·Δp,Δp b ),i=0,1,…,m

[0028]

[0029] Where Δp is the desuperheating water flow control range, i is the number of calculations, and Ceiling is the rounding-up operation.

[0030] As a preferred technical solution, in step S7, the superheated steam temperature prediction control performance index J i The specific calculation is as follows:

[0031]

[0032] Where ω is the calculation coefficient of the change in the desuperheating water flow rate; for The predicted value of the wall temperature at time l in the future under the action of It is the superheated steam temperature set value at the next moment l.

[0033] As a preferred technical solution, the prediction model in step S1 can be adjusted according to actual control requirements.

[0034] As an optimal technical solution, the relevant parameters in step S1 include unit power, feed water flow, AF ground coal amount and total coal amount, total air volume, AF ground primary air volume, primary air pressure, secondary air damper opening of each layer of burners, feed water enthalpy value, swing valve position of each corner burner, main steam pressure, mid-point temperature, furnace area soot blowing signal, secondary A side desuperheating water flow, secondary A side desuperheater outlet steam temperature and secondary A side inlet superheated steam temperature.

[0035] Compared with the existing technology, the present invention addresses the control difficulties of the superheated steam temperature control object, such as large delay inertia, many disturbance factors and nonlinear characteristics, and proposes a deep learning predictive control method that is convenient for practical engineering applications. By establishing a deep learning model of the boiler superheated steam temperature system, the changes in superheated steam temperature at future moments under different control quantities are calculated based on the model, and the control quantity corresponding to the optimal control result is selected as the controller output to control the superheated steam temperature, thereby improving the operating stability of the unit's superheated steam temperature, thereby further improving the unit's safe and stable operation level. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of the superheated steam temperature prediction and control method based on the deep learning model of the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0038] This paper proposes a superheated steam temperature predictive control method based on a deep learning model. Aiming at the control difficulties such as large delay inertia, many disturbance factors and nonlinear characteristics of the superheated steam temperature control object, a deep learning predictive control method that is convenient for practical engineering application is proposed. By establishing a deep learning model of the boiler superheated steam temperature system, the changes in superheated steam temperature at future moments under different control quantities are calculated based on the model, and the control quantity corresponding to the optimal control result is selected as the controller output to control the superheated steam temperature, thereby improving the operating stability of the unit's superheated steam temperature and improving the safe and stable operation level of the unit.

[0039] like Figure 1 As shown, the present invention takes the predictive control of the superheated steam temperature on the A side of the secondary stage of a 1000MW ultra-supercritical unit boiler as an example, and the specific process is as follows:

[0040] 1) Parameters related to the boiler's pulverizing system, combustion system, and air and smoke system were used as feature input parameters. Historical operating data was selected and a recurrent neural network was used to train a deep learning model to predict future changes in the superheated steam temperature on the boiler's secondary A side. The data sampling period was 20 seconds, and the time step was 10 steps. Six models were developed to predict changes in the boiler wall temperature over the next 30 seconds, 60 seconds, 90 seconds, 120 seconds, 150 seconds, and 180 seconds. The model input feature parameters are shown in Table 1.

[0041] Table 1

[0042]

[0043] 2) Read the parameter values of the characteristic variables in Table 1 from the DCS system in real time. Considering the large difference between the model sampling period and the DCS data period, a dynamic data storage matrix queue is established to drive the model in real time.

[0044]

[0045] In the formula, k represents the current moment, p j (k) (j = 1, 2, ..., n) is the parameter of the jth characteristic variable at time k. After reading the real-time data of the characteristic variable, a shift operation is performed on the dynamic data storage matrix, and the real-time data is assigned to the parameter corresponding to the current time (time k).

[0046] 3) Extract the input data matrix of the model at the current moment from the dynamic data storage matrix queue according to the sampling period of the deep learning model training data and the model time compensation:

[0047]

[0048] 4) Input the data matrix in step 3) into the deep learning model to calculate the changes in the boiler superheated steam temperature in the next 30s, 60s, 90s, 120s, 150s and 180s respectively;

[0049] 5) Obtain the time series of the future superheated steam temperature set value Compare the data with the 6 future superheated steam temperature change data predicted in step 4) to perform predictive control;

[0050] 6) Determine the change limit of the cooling water flow rate p2 within a control cycle [-Δp a ,Δp b ], in the superheated steam temperature prediction control, the new set desuperheating water flow rate Instead of the actual desuperheating water flow rate p2(k) at the current moment, re-calculate the model prediction to obtain the superheated steam temperature change value at the future moment

[0051] Δp i =min(-Δp a +i·Δp,Δp b ),i=0,1,…,m

[0052]

[0053] Where Δp is the cooling water flow control interval, which is 2t / h; i is the number of calculations; Ceiling is the rounding operation; Δp a and Δp b Take -10t / h and 10t / h respectively.

[0054] 7) Calculation difference Superheated steam temperature predictive control performance indicators under input:

[0055]

[0056] Where ω is the calculation coefficient of the change in the cooling water flow rate, which is taken as 0.01; for The predicted value of the wall temperature at time l in the future under the action of It is the superheated steam temperature set value at the next moment l.

[0057] 8) Select the minimum J i Corresponding The superheated steam temperature is controlled as the feed water flow rate instruction at the next moment.

[0058] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A superheated steam temperature prediction and control method based on a deep learning model, characterized in that: This method establishes a deep learning model of the boiler superheated steam temperature system. Based on this model, the changes in superheated steam temperature at future moments under different control variables are calculated. The control variable corresponding to the optimal control result is selected as the controller output to control the superheated steam temperature, thereby improving the superheated steam temperature operation stability of the unit. The method specifically comprises the following steps: Step S1: Using relevant parameters of the boiler pulverizing system, combustion system, and air and smoke system as feature input parameters, selecting historical operating data and using a recurrent neural network to train a deep learning model to predict future changes in the boiler superheated steam temperature; Step S2, reading the parameter values of the characteristic variables from the DCS system in real time and establishing a dynamic data storage matrix queue; Step S3: extracting the input data matrix of the model at the current moment from the dynamic data storage matrix queue according to the sampling period of the deep learning model training data and the model time compensation; Step S4: Input the data matrix in step S3 into the deep learning model to calculate the change of the boiler superheated steam temperature in the next 30s, 60s, 90s, 120s, 150s and 180s respectively; Step S5: Obtain the time series of the future superheated steam temperature setting value Compare the data with the six future superheated steam temperature change data predicted in step S4 to perform predictive control; Step S6: determine the change limit of the cooling water flow rate p2 within a control cycle [-Δp a ,Δp b ], in the superheated steam temperature prediction control, the new set desuperheating water flow rate Instead of the actual desuperheating water flow rate p2(k) at the current moment, re-calculate the model prediction to obtain the superheated steam temperature change value at the future moment Step S7, calculate the difference The superheated steam temperature predictive control performance index J under input i ; Step S8, select the minimum J i Corresponding The superheated steam temperature is controlled as the feed water flow rate instruction at the next moment.

2. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: In step S1, six models are established to predict the changes of boiler superheated steam temperature in the next 30s, 60s, 90s, 120s, 150s and 180s.

3. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: The dynamic data storage matrix queue in step S2 is specifically: Where k represents the current moment, p j (k) is the parameter of the jth characteristic variable at time k, j = 1, 2, ..., n, data t is the sampling period, and d is the time step.

4. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: In step S2, after reading the real-time data of the characteristic variables, a shift operation is performed on the dynamic data storage matrix, and the real-time data is assigned to the parameters corresponding to the current moment.

5. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: In step S3, the input data matrix of the model at the current moment is specifically: Where k represents the current moment, p j (k) is the parameter of the jth characteristic variable at time k, j = 1, 2, ..., n, data t is the sampling period, and d is the time step.

6. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: In step S6, the newly set cooling water flow rate The specific calculation is as follows: Δp i =min(-Δp a +i·Δp,Δp b ),i=0,1,…,m Where Δp is the desuperheating water flow control range, i is the number of calculations, and Ceiling is the rounding-up operation.

7. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: In step S7, the superheated steam temperature prediction control performance index J i The specific calculation is as follows: J i =∑|T l i -T l s |+ω·|Δp i |,l=30,60,90,120,150,180 Where ω is the calculation coefficient of the change in cooling water flow rate; T l i for The predicted value of the wall temperature at time l in the future under the action of T l s It is the superheated steam temperature set value at the next moment l.

8. The superheated steam temperature prediction and control method based on a deep learning model according to claim 1, characterized in that: The prediction model in step S1 can be adjusted according to actual control requirements.

Images