A steam turbine exhaust pressure control method based on MPC

Abstract

The present invention provides a steam turbine middle row pressure control method based on MPC, comprising: acquiring past middle row pressure data; decomposing the past middle row pressure data with a history length of m1 into trend data, periodic data and noise data by using an STL decomposition method; defining a prediction length and a control length, inputting the trend data and periodic data obtained by decomposition into a trained LSTM model respectively, and obtaining a trend data prediction value and a periodic data prediction value; adding the trend data prediction value and the periodic data prediction value to obtain a middle row pressure prediction value; determining the optimal valve opening of a middle and low pressure connecting pipe valve based on a middle row pressure performance index, a constraint condition, an actual middle row pressure value and a middle row pressure prediction value, comparing the middle row pressure prediction value with the actual middle row pressure value, and adjusting the middle and low pressure connecting pipe valve to control the middle row pressure; rolling the prediction time domain forward by one step to realize rolling optimization control of the middle row pressure; the present invention effectively overcomes the defect of untimely PID control.

Description

Technical Field

[0001] The invention belongs to the technical field of steam turbine control, and in particular relates to a steam turbine exhaust pressure control method based on MPC. Background Art

[0002] During the heating period of the steam turbine, in order to ensure the safe operation of the intermediate pressure cylinder, the valve opening of the intermediate and low pressure connecting valve needs to be adjusted to maintain the intermediate pressure within the design allowable range. At present, the intermediate pressure of the steam turbine adopts PID control so that the intermediate pressure can automatically track the target value of the intermediate pressure, wherein the target value of the intermediate pressure is calculated by the intermediate pressure cylinder intake pressure, intermediate pressure cylinder intake temperature and turbine power. In this way, when performing the intermediate pressure control, the target value of the intermediate pressure is calculated according to the current intermediate pressure cylinder intake pressure, intermediate pressure cylinder intake temperature and turbine power, and then compared with the actual value of the intermediate pressure to control the valve opening of the intermediate and low pressure connecting pipe, which has the problem of untimely control. Summary of the invention

[0003] In view of the above shortcomings of the prior art, the purpose of the present invention is to provide a turbine middle row pressure control method based on MPC, which is based on the past middle row pressure of the steam turbine and adopts an LSTM neural network model to predict the middle row pressure, so as to timely adjust the valve opening of the medium and low pressure connecting pipes, effectively solving the problem of untimely original PID control.

[0004] To achieve the above object and other related objects, the present invention provides a method for controlling the exhaust pressure of a steam turbine based on MPC, the method comprising the following steps:

[0005] Obtain the past exhaust pressure data within the statistical period of the steam turbine;

[0006] Use the STL decomposition method to decompose the past mid-row pressure data with a history length of m1 into trend data, periodic data and noise data;

[0007] Define the prediction length m2 and control length m3, input the decomposed trend data and period data into the trained LSTM model respectively, and obtain the trend data prediction value and period data prediction value in the prediction time domain;

[0008] Add the trend data prediction value and the period data prediction value to obtain the middle row pressure prediction value;

[0009] Collect the actual value of the middle row pressure, determine the optimal valve opening of the middle and low pressure connecting pipe valve based on the middle row pressure performance index, constraint conditions, the actual value of the middle row pressure and the predicted value of the middle row pressure, compare the predicted value of the middle row pressure with the actual value of the middle row pressure, and adjust the middle and low pressure connecting pipe valve to control the middle row pressure;

[0010] The prediction time domain is continuously rolled forward by one step to realize the rolling optimization control of the middle row pressure.

[0011] Preferably, the middle row pressure performance index is:

[0012] ζ=abs(E ideal -E real ) / E real ; Among them, E ideal is the predicted value of the middle row pressure; E real is the actual value of the middle row pressure.

[0013] Preferably, the constraints are:

[0014] θ>f2(P2,T2,W2);

[0015] Among them, θ is the actual value of the low-pressure cylinder intake flow rate; P2, T2, W2 are the actual values ​​of the low-pressure cylinder intake pressure; the actual value of the low-pressure cylinder intake temperature; the actual value of the turbine power; f2(P2, T2, W2) is the calculation formula for the low-pressure cylinder intake flow rate limit.

[0016] Preferably, the adjustment method of the medium and low pressure connecting pipe valve is: if E ideal -E real >0, the valve opening of the medium and low pressure connecting pipe valve is controlled to decrease to the optimal valve opening at a preset rate; if E ideal -E real <0, control the valve opening of the medium and low pressure connecting pipe valve to increase to the optimal valve opening at a preset rate.

[0017] As described above, the MPC-based steam turbine exhaust pressure control method of the present invention has the following beneficial effects:

[0018] The present invention is based on the past middle row pressure data of the steam turbine, and adopts the LSTM neural network model to predict the middle row pressure to improve the prediction effect; then, based on the predicted middle row pressure, the valve opening of the middle and low pressure connecting pipes is subjected to rolling optimization control, which effectively overcomes the problem of untimely control existing in the original PID control and improves the control accuracy of the middle row pressure. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the steam turbine in the present invention.

[0020] Figure 2 A flow chart of a steam turbine exhaust pressure control method based on MPC provided by the present invention.

[0021] Description of Reference Numerals

[0022] High and medium pressure cylinder 1, low pressure cylinder 2, medium and low pressure connecting pipe 3, medium and low pressure connecting valve 31, steam extraction pipe 4, safety valve 41, steam extraction check valve 42, steam extraction fast closing valve 43, steam extraction regulating valve 44, middle exhaust pressure measuring point 5, low inlet pressure measuring point 6, low inlet temperature measuring point 7. DETAILED DESCRIPTION

[0023] The following is a description of the implementation of the present invention by means of specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification.

[0024] See also Figure 1 to Figure 2 . It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings in this specification are only used to match the contents disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention, so they have no substantial technical significance. Any modification of the structure, change in the proportional relationship, or adjustment of the size should still fall within the scope of the technical content disclosed by the present invention without affecting the effects and purposes that can be achieved by the present invention. At the same time, the terms such as "upper", "lower", "left", "right", "middle" and "one" quoted in this specification are only for the convenience of description, and are not used to limit the scope of the implementation of the present invention. Changes or adjustments in their relative relationships should also be regarded as the scope of the implementation of the present invention without substantially changing the technical content.

[0025] The steam turbine structure involved in the present invention is as follows Figure 1 As shown, it includes a high- and medium-pressure cylinder 1 and a low-pressure cylinder 2, and a medium- and low-pressure connecting pipe 3 and a steam extraction pipe 4 are arranged in parallel at the exhaust port of the high- and medium-pressure cylinder 1; the medium- and low-pressure connecting pipe 3 is used to connect the high- and medium-pressure cylinder 1 with the low-pressure cylinder 2, and the steam extraction pipe 4 is used to connect the high- and medium-pressure cylinder 1 with the heat network user (not shown in the figure); a medium- and low-pressure connecting valve 31 is arranged on the medium- and low-pressure connecting pipe 3, which is used to adjust the flow of the medium- and low-pressure connecting pipe 3; a safety valve 41, a steam extraction check valve 42, a steam extraction fast closing valve 43 and a steam extraction regulating valve 44 are arranged on the steam extraction pipe 4; wherein, the safety valve 41 is used to protect the steam extraction pipe 4 and the steam turbine; the steam extraction pipe 4 The check valve 42 is used to prevent steam backflow, the extraction steam quick closing valve 43 is used for the rapid separation of the heat network user and the steam turbine under accident conditions, and the extraction steam regulating valve 44 is used to adjust the flow of the extraction steam pipe 4; a middle row pressure measuring point 5 is provided at the exhaust port of the high and medium pressure cylinders 1, and a low inlet pressure measuring point 6 and a low inlet temperature measuring point 7 are provided at the air inlet of the low pressure cylinder 2; each measuring point adopts a triple redundant configuration (that is, three measuring sensors are used for measurement) to improve the reliability of the measurement point; by controlling the valve opening of the middle and low pressure connecting valve 31, automatic control of the middle row pressure is achieved to ensure the safe operation of the unit.

[0026] like Figure 2As shown, an embodiment of the present invention provides a steam turbine exhaust pressure control method based on MPC (Model Predictive Control), and the control method includes the following steps:

[0027] S1. Obtain the past middle and exhaust pressure data within the statistical period of the steam turbine;

[0028] The past middle row pressure data can be obtained from the plant-level supervisory information system (Supervisory Information System in Plant Level, referred to as SIS).

[0029] S2, use STL decomposition method to decompose the past mid-row pressure data Y with a history length of m1 t Decompose into trend data T t , cycle data S t and the noisy data R t ;

[0030] STL (Seasonal and Trend decomposition using Loess) is a method for decomposing time series. Its decomposition model is: t =T t +S t +R t , which can be implemented based on python programming software.

[0031] S3, define the prediction length m2 and control length m3, and decompose the trend data T t and cycle data S t Input them into the trained LSTM neural network model respectively to obtain the trend data prediction value T in the prediction time domain t ′ and periodic data prediction value S t ′;

[0032] LSTM (Long Short-Term Memory) is a time-recurrent neural network. The LSTM neural network model can be trained through past mid-row pressure data. Since the training method is an existing technology, it will not be described in detail.

[0033] S4. The trend data prediction value T t ′ and periodic data prediction value S t ' and add them together to obtain the predicted value of the middle row pressure E ideal ;

[0034] Specifically, the predicted value of the middle row pressure E ideal =T t ′+S t ′.

[0035] S5. Collect the actual value of the middle row pressure, determine the optimal valve opening of the middle and low pressure connecting pipe valve based on the middle row pressure performance index, constraint conditions, the actual value of the middle row pressure and the predicted value of the middle row pressure, compare the predicted value of the middle row pressure with the actual value of the middle row pressure, and adjust the middle and low pressure connecting pipe valve to control the middle row pressure;

[0036] Specifically, the middle row pressure performance index is: ζ=abs(E ideal -E real ) / E real ; Among them, E ideal is the predicted value of the middle row pressure, E real is the actual value of the middle row pressure;

[0037] The constraints are: θ>f2(P2,T2,W2); where θ is the actual value of the low-pressure cylinder intake flow rate; P2, T2, W2 are the actual values ​​of the low-pressure cylinder intake pressure; the actual value of the low-pressure cylinder intake temperature; the actual value of the turbine power; and f2(P2,T2,W2) is the calculation formula for the low-pressure cylinder intake flow rate limit.

[0038] According to the predicted value of the middle row pressure E ideal and the actual value of the middle row pressure E real The middle row pressure performance index is obtained, and on the basis of ensuring that the low-pressure cylinder intake flow meets the requirements, the optimal valve opening of the middle and low pressure connecting pipe valve 31 is determined with the minimization of the middle row pressure performance index as the control target; if E ideal >E real , gradually reduce the valve opening of the medium-low pressure connecting pipe valve 31 at a preset rate until the valve opening of the medium-low pressure connecting pipe valve 31 reaches the determined optimal valve opening; if E ideal <E real , gradually increase the valve opening of the medium and low pressure connecting pipe valve 31 at a preset rate until the valve opening of the medium and low pressure connecting pipe valve 31 reaches the determined optimal valve opening to prevent the sudden change of the middle row pressure from causing system disturbance.

[0039] S6. Continuously roll the prediction time domain forward by one step to achieve rolling optimization control of the middle row pressure.

[0040] The MPC-based steam turbine mid-discharge pressure control method disclosed in the present invention can effectively overcome the problem of untimely control in traditional PID, thereby improving the control accuracy of the mid-discharge pressure.

[0041] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A steam turbine exhaust pressure control method based on MPC, characterized in that: The control method comprises the following steps: Obtain the past exhaust pressure data within the statistical period of the steam turbine; Use the STL decomposition method to decompose the past mid-row pressure data with a history length of m1 into trend data, periodic data and noise data; Define the prediction length m2 and control length m3, input the decomposed trend data and period data into the trained LSTM neural network model respectively, and obtain the trend data prediction value and period data prediction value; Add the trend data prediction value and the period data prediction value to obtain the middle row pressure prediction value; Collect the actual value of the middle row pressure, determine the optimal valve opening of the middle and low pressure connecting pipe valve based on the middle row pressure performance index, constraint conditions, the actual value of the middle row pressure and the predicted value of the middle row pressure, compare the predicted value of the middle row pressure with the actual value of the middle row pressure, and adjust the middle and low pressure connecting pipe valve to control the middle row pressure; The prediction time domain is continuously rolled forward by one step to realize rolling optimization control of the middle row pressure; The middle row pressure performance index is: ζ=abs(E ideal -AND real ) / AND real Among them, E ideal is the predicted value of the middle row pressure, E real is the actual value of the middle row pressure; The constraints are: θ>f2(P2,T2,W2) Among them, θ is the actual value of the low-pressure cylinder intake flow; P2, T2, W2 are the actual values ​​of the low-pressure cylinder intake pressure; the actual value of the low-pressure cylinder intake temperature; the actual value of the turbine power; f2(P2, T2, W2) is the calculation formula for the low-pressure cylinder intake flow limit.

2. A steam turbine exhaust pressure control method based on MPC according to claim 1, characterized in that: The adjustment method of the medium and low pressure connecting pipe valve is: if E ideal -E real >0, the valve opening of the medium and low pressure connecting pipe valve is controlled to decrease to the optimal valve opening at a preset rate; if E ideal -E real <0, control the valve opening of the medium and low pressure connecting pipe valve to increase to the optimal valve opening at a preset rate.

Images