CN-122018460-A - Thermal power generating unit control strategy optimization method and system based on combination of DCS and AGC

CN122018460ACN 122018460 ACN122018460 ACN 122018460ACN-122018460-A

Abstract

The invention relates to the technical field of distributed power generation, which solves the technical problems that the prior art cannot accurately match the actual working condition, is difficult to capture the time sequence dynamic characteristic, and meanwhile, the energy evaluation and feedforward compensation lack comprehensive consideration, and control deviation and instability are easy to occur, in particular to a thermal power unit control strategy optimization method and system based on DCS and AGC combination, wherein the method comprises the following steps: the method and the device can rapidly and accurately subdivide the unit operation working conditions into specific categories according to the real-time state data of the unit, greatly improve the accuracy of working condition identification, effectively improve the predictive advance and accuracy, enable the unit to perceive the parameter change trend in advance, avoid control problems caused by parameter mutation, and improve the comprehensiveness and accuracy of energy assessment.

Inventors

WANG AICHENG
BI YULONG
LIU SHUJIE
WANG TAO
DOU ZHI
ZHANG DONG
JIA YUEJUN
YANG HUIJIE

Assignees

华能国际电力股份有限公司德州电厂

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The thermal power unit control strategy optimization method based on DCS and AGC is characterized by comprising the steps of collecting original data of a thermal power unit, constructing a working condition model based on the original data, and generating a working condition model parameter set, wherein the original data comprises original DCS historical operation data and an AGC instruction sequence; Determining a pressure set value based on a historical data threshold, collecting real-time state data of the thermal power generating unit, processing working condition data to obtain a working condition label, and generating a dynamic feedforward compensation quantity according to the pressure set value, the working condition label and a working condition model parameter set; Acquiring target load and emission constraint of a thermal power unit, generating an optimizing processing space, selecting a candidate control quantity group in the optimizing processing space, performing weighted adaptation processing to obtain an adaptation value, and performing global optimizing decoding based on the adaptation value to obtain an optimal instruction set; Acquiring a real-time data set of a thermal power generating unit, performing energy storage fuzzy processing on real-time state data and the real-time data set to generate a coefficient set and a component set, and performing instruction synthesis processing based on the coefficient set, the component set and an optimal instruction set to obtain a main control instruction set; and acquiring current unit operation data based on the main control instruction set, evaluating and updating the model according to the current unit operation data to obtain a model updating mark, and retraining the model according to the model updating mark.
2. The thermal power generating unit control strategy optimization method based on DCS and AGC combination according to claim 1, wherein the working condition model construction is performed based on the original data, and the method comprises the following steps: The method comprises the steps of obtaining an alignment variable data set through filtering and time stamp alignment processing based on original DCS historical operation data; performing key performance data processing on the AGC instruction sequence and the alignment variable data set to obtain KPI label historical data and a historical KPI label; And obtaining a working condition model parameter set through working condition clustering processing according to the KPI label historical data.
3. The thermal power generating unit control strategy optimization method based on combination of DCS and AGC according to claim 1, wherein the working condition label is obtained by working condition data processing, and the method comprises the following steps: performing feature integration processing based on the real-time state data to obtain a real-time feature vector; generating a standardized feature vector through feature standardization processing according to the real-time feature vector and the working condition model parameter set; and carrying out working condition matching treatment on the standardized feature vector and the working condition model parameter set to obtain a working condition label.
4. The method for optimizing control strategy of thermal power generating unit based on combination of DCS and AGC according to claim 1, wherein generating the dynamic feedforward compensation amount according to the pressure set value, the condition label and the condition model parameter set comprises: Performing multi-step dynamic prediction processing based on the real-time state data and the working condition model parameter set to obtain a main steam pressure predicted value and a load predicted value; According to the main steam pressure predicted value, the load predicted value, the AGC instruction sequence and the pressure set value, the estimated energy gap is generated through energy gap estimation; and carrying out dynamic feedforward processing on the estimated energy gap and the working condition label to obtain dynamic feedforward compensation quantity.
5. The thermal power generating unit control strategy optimization method based on combination of DCS and AGC according to claim 1, wherein the weighted adaptation process is performed to obtain a fitness value, and the method comprises the following steps: performing upper and lower limit processing according to the target load and emission constraint to generate an optimizing processing space; randomly selecting a plurality of candidate control quantity groups in the optimizing processing space, and performing forward processing on the candidate control quantity groups and the working condition model parameter set to obtain a key performance predicted value; And obtaining the fitness value through weighted fitness processing according to the key performance predicted value.
6. The thermal power generating unit control strategy optimization method based on combination of DCS and AGC according to claim 1, wherein the global optimizing decoding based on the fitness value comprises the following steps: Carrying out population updating processing based on the optimizing processing space to generate updated population positions; obtaining a current global optimal solution through elite reservation processing according to the updated population position; and obtaining all the current global optimal solutions to perform optimal solution decoding processing to obtain an optimal instruction set.
7. The thermal power generating unit control strategy optimization method based on combination of DCS and AGC according to claim 1, wherein the generating of coefficient sets by energy storage fuzzy processing of real-time state data and real-time data sets comprises the following steps: obtaining a component set through digital filtering processing according to the real-time data set; Performing energy storage state evaluation based on the real-time state data to generate a real-time energy storage margin; and carrying out dynamic fuzzy processing on the real-time energy storage margin and the component set to obtain a coefficient set.
8. The thermal power generating unit control strategy optimization method based on DCS and AGC combination according to claim 1, wherein the instruction synthesis processing based on the coefficient set and the optimal instruction set comprises: Based on the coefficient set and the component set, carrying out instruction dynamic generation to obtain a turbine instruction increment and a boiler instruction increment; According to the boiler instruction increment, the dynamic feedforward compensation quantity and the optimal instruction set, boiler instruction synthesis is carried out, and a boiler main control instruction is obtained; And carrying out instruction synthesis processing on the optimal instruction set and the steam turbine instruction increment to generate a steam turbine main control instruction parameter set, and combining the boiler main control instruction and the steam turbine main control instruction parameter set into a main control instruction set.
9. The thermal power unit control strategy optimization method based on DCS in combination with AGC of claim 1, wherein the evaluation and model update based on current unit operation data comprises: performing feature reconstruction processing according to the current unit operation data to generate a current feature vector and a current working condition label; performing feature consistency judgment based on the current feature vector and the current working condition label to generate a model adaptability signal; And performing model updating judgment on the model adaptability signal to generate a model updating mark.
10. A system for applying the thermal power generating unit control strategy optimization method based on DCS and AGC combination as claimed in any one of claims 1 to 9, comprising: the working condition processing module is used for collecting original data of the thermal power generating unit, constructing a working condition model based on the original data, generating a working condition model parameter set, wherein the original data comprises original DCS historical operation data and an AGC instruction sequence; The feedforward compensation module is used for determining a pressure set value based on a historical data threshold value, collecting real-time state data of the thermal power generating unit, processing working condition data to obtain a working condition label, and generating a dynamic feedforward compensation quantity according to the pressure set value, the working condition label and a working condition model parameter set; The instruction optimizing module is used for acquiring target load and emission constraint of the thermal power generating unit, generating an optimizing processing space, selecting a candidate control quantity group in the optimizing processing space, carrying out weighted adaptation processing to obtain an adaptation value, and carrying out global optimizing decoding based on the adaptation value to obtain an optimal instruction set; The instruction generation module is used for acquiring a real-time data set of the thermal power generating unit, carrying out energy storage fuzzy processing on the real-time state data and the real-time data set to generate a coefficient set and a component set, and carrying out instruction synthesis processing based on the coefficient set, the component set and an optimal instruction set to obtain a main control instruction set; The model updating module is used for acquiring current unit operation data based on the main control instruction set, evaluating and updating the model according to the current unit operation data to obtain a model updating mark, and retraining the model according to the model updating mark.

Description

Thermal power generating unit control strategy optimization method and system based on combination of DCS and AGC Technical Field The invention relates to the technical field of distributed power generation, in particular to a thermal power unit control strategy optimization method and system based on combination of DCS and AGC. Background The distributed power generation refers to a small power generation unit configured near a user side, and the distributed power generation unit is combined with a DCS (distributed control system) and an AGC (automatic generation control) to realize local control and AGC (automatic generation control) to coordinate global output through the DCS, optimize the cooperative operation of the thermal power unit and distributed energy, and mostly adopt PID (proportion integration differentiation) control, model predictive control or neural network algorithm control in the prior art to realize power distribution and dynamic adjustment. However, the working condition identification in the prior art is extensive, the actual working condition cannot be accurately matched, the prediction method is mostly a single-step or simple model, the time sequence dynamic characteristic is difficult to capture, meanwhile, the energy evaluation and the feedforward compensation lack comprehensive consideration, the control deviation and the unstable condition are easy to occur, for example, when the temperature of cooling water is extremely low and approaches 0 ℃ in severe cold, the exhaust steam pressure of a steam turbine can be reduced, the acting capacity of the steam in the steam turbine is reduced, the load output of a unit is further reduced, and at the moment, the fixed feedforward compensation cannot timely sense the load change deviation caused by the environmental temperature, so that a larger gap exists between the actual load and the set load. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a thermal power generating unit control strategy optimization method and system based on the combination of DCS and AGC, which solve the technical problems that the prior art cannot accurately match the actual working condition, the time sequence dynamic characteristic is difficult to capture, and meanwhile, the energy evaluation and the feedforward compensation lack comprehensive consideration, so that the control deviation and instability are easy to occur. The invention provides a technical scheme for solving the technical problems, which is applied to the thermal power generating unit control strategy optimization method based on the combination of DCS and AGC, and comprises the steps of collecting original data of a thermal power generating unit, constructing a working condition model based on the original data, and generating a working condition model parameter set, wherein the original data comprises original DCS historical operation data and an AGC instruction sequence; Determining a pressure set value based on a historical data threshold, collecting real-time state data of the thermal power generating unit, processing working condition data to obtain a working condition label, and generating a dynamic feedforward compensation quantity according to the pressure set value, the working condition label and a working condition model parameter set; Acquiring target load and emission constraint of a thermal power unit, generating an optimizing processing space, selecting a candidate control quantity group in the optimizing processing space, performing weighted adaptation processing to obtain an adaptation value, and performing global optimizing decoding based on the adaptation value to obtain an optimal instruction set; Acquiring a real-time data set of a thermal power generating unit, performing energy storage fuzzy processing on real-time state data and the real-time data set to generate a coefficient set and a component set, and performing instruction synthesis processing based on the coefficient set, the component set and an optimal instruction set to obtain a main control instruction set; and acquiring current unit operation data based on the main control instruction set, evaluating and updating the model according to the current unit operation data to obtain a model updating mark, and retraining the model according to the model updating mark. Preferably, the working condition model construction based on the original data comprises: The method comprises the steps of obtaining an alignment variable data set through filtering and time stamp alignment processing based on original DCS historical operation data; performing key performance data processing on the AGC instruction sequence and the alignment variable data set to obtain KPI label historical data and a historical KPI label; And obtaining a working condition model parameter set through working condition clustering processing according to the KPI label historical data. Preferably, the working conditio