CN102622530B

CN102622530B - Improved genetic algorithm-based method for distributing and optimizing thermal and electrical load of steam extraction and heating unit

Info

Publication number: CN102622530B
Application number: CN201210121857.XA
Authority: CN
Inventors: 胡宝权; 刘金福; 刘娇; 宋崇明; 张怀鹏; 李飞; 陈增吉; 李涛; 徐扬; 李树臣; 张修君; 左世春; 于达仁
Original assignee: HARBIN NO 3 POWER PLANT HUADIAN ENERGY CO Ltd; Harbin Institute of Technology Shenzhen
Current assignee: HARBIN NO 3 POWER PLANT HUADIAN ENERGY CO Ltd; Harbin Institute of Technology Shenzhen
Priority date: 2012-04-24
Filing date: 2012-04-24
Publication date: 2015-07-01
Anticipated expiration: 2032-04-24
Also published as: CN102622530A

Abstract

The invention relates to an optimization method for thermal and electric load distribution of steam extraction and heat supply units based on an improved genetic algorithm, and belongs to the technical field of energy-saving monitoring of power plants. In order to realize the optimization of thermoelectric load distribution of multiple heat supply units in the power plant, the invention enables the electric load and heat load of the power plant to be optimally allocated while satisfying the needs of users, and reduces the total energy consumption to achieve the purpose of energy saving. Set the actual heat consumption and consumption difference curve of the unit to obtain the design heat consumption curve of the unit; obtain the electrical load and steam extraction volume of each unit; calculate the total heat consumption value of all units based on the improved genetic algorithm The electric load and steam extraction value of each unit at the minimum: By improving the genetic coding and fitness function, the selection, crossover, and mutation operations of the genetic algorithm make the optimization process satisfy the constraints of the output to meet the requirements of all units. Total heat consumption value The optimal solution of the electric load and steam extraction value of each unit at the minimum hour and the corresponding minimum total heat consumption. The method improves the speed of the optimization process and the accuracy of the optimization result.

Description

Optimization method for heat and electricity load distribution of steam extraction heating units based on improved genetic algorithm

技术领域 technical field

本发明涉及一种电厂多台供热机组热电负荷分配优化方法，属于发电厂节能监测技术领域。The invention relates to a method for optimizing thermoelectric load distribution of multiple heating units in a power plant, and belongs to the technical field of energy-saving monitoring of power plants.

背景技术 Background technique

随着经济的发展和人民生活质量的提高，城市集中供热系统得到迅速发展，其中热电联产能源转换效率具有明显优势，因此，供热抽汽机组得到了大力的发展。With the development of the economy and the improvement of the quality of life of the people, the urban central heating system has been developed rapidly, and the energy conversion efficiency of cogeneration has obvious advantages. Therefore, the heating and extraction unit has been vigorously developed.

抽汽供热机组向用户提供电力和采暖用热，电厂提供的热力和电力的多少，受控于热用户和电用户的需求，因此，电厂必须按照热用户和电用户的需求调整供热抽汽机组的热电负荷。The steam extraction heating unit provides users with electricity and heating heat. The amount of heat and electricity provided by the power plant is controlled by the needs of heat users and electricity users. Therefore, the power plant must adjust the heat supply and extraction according to the needs of heat users and electricity users. The heat and electricity load of the steam unit.

对于确定的热电负荷，电厂如何根据机组的类型以及机组效率的差异，在各机组间进行热电负荷的分配，使整个电厂的热耗率最低，使整个电厂的经济效益最好，是电厂生产运行中面临的问题。这就需要对电厂供热抽汽机组间的电负荷及热负荷进行分配优化，确定每台机组的电负荷和热负荷。For a certain thermal and electrical load, how should the power plant distribute the thermal and electrical load among the units according to the type of the unit and the difference in unit efficiency, so that the heat consumption rate of the entire power plant is the lowest, and the economic benefits of the entire power plant are the best. problems faced. This requires optimizing the distribution of electrical and thermal loads among the heating and extraction units of the power plant, and determining the electrical and thermal loads of each unit.

负荷的优化分配是指，在全厂总的调度负荷下，根据各个机组的热力特性确定各机组应承带的负荷，从而使全厂的煤耗量最小的一种优化调度。The optimal distribution of load refers to an optimal dispatching that, under the total dispatching load of the whole plant, determines the load that each unit should carry according to the thermal characteristics of each unit, so as to minimize the coal consumption of the whole plant.

针对电厂的负荷优化分配，较早开展也较为成熟的是纯凝机组的电负荷分配优化研究，等微增率法得到了广泛的应用，由于抽汽供热机组热负荷也需参与优化分配，因此，无论是从热耗曲线获取、还是优化复杂性角度均较纯凝机组的电负荷分配优化复杂。目前，针对抽汽供热机组的热、电负荷分配优化，已开展了许多研究。For the load optimization distribution of power plants, the study on the optimization of electric load distribution of pure condensing units was carried out earlier and more maturely. The equal micro-increase rate method has been widely used. Since the heat load of steam extraction heating units also needs to participate in optimal distribution, Therefore, whether it is obtained from the heat consumption curve or the optimization complexity, it is more complicated than the electric load distribution optimization of the pure condensing unit. At present, many studies have been carried out on the optimization of heat and electricity load distribution of steam extraction heating units.

文献[1](魏豪；宋宝峰；赵伟东；王奕；《吉林电力》，2002年第5期，《供热汽轮机组热电负荷优化分配系统的开发与应用》)中介绍的热电负荷分配优化系统：采用“逐点法”分配的数学模型，利用等效热降理论对影响汽轮机组经济性的主要参数进行偏差分析，能够实现经济指标计算及能损分析、参数显示、查询及报警和汽机模拟量系统图显示的功能。本文献中的“逐点法”分配的数学模型稳定性差，计算速度较慢，无法进行连续优化。同时等效热降法应用于供热抽汽机组较为复杂。The thermoelectric load distribution optimization system introduced in the literature [1] (Wei Hao; Song Baofeng; Zhao Weidong; Wang Yi; "Jilin Electric Power", No. 5, 2002, "Development and Application of Thermal Power Load Optimal Distribution System for Heating Steam Turbine Units") : Adopt the mathematical model of "point-by-point method" distribution, and use the equivalent heat drop theory to analyze the deviation of the main parameters affecting the economy of the steam turbine unit, which can realize economic index calculation and energy loss analysis, parameter display, query and alarm and steam turbine simulation The functions shown in the volume system diagram. The mathematical model assigned by the "point-by-point method" in this paper has poor stability, slow calculation speed, and cannot be continuously optimized. At the same time, it is more complicated to apply the equivalent heat drop method to the heating and extraction unit.

文献[2](冉鹏，张树芳，《汽轮机技术》，2006年第48卷第1期：《基于遗传算法的热电厂负荷优化计算方法》)中应用遗传算法建立热电厂负荷优化模型的方法，解决了当问题规模扩大，变量和约束条件很多时，会很容易陷入局部最优，而使数值稳定性降低，最终导致收敛困难的问题。文献[2]虽然部分解决了文献[1]中的问题，但是遗传算法存在当初始种群过大，计算速度较慢的问题，同时该算法没有实现供热机组在线实时优化功能，在实际生产中不能广泛应用，也没有考虑机组的实际运行条件对热耗的影响。In the literature [2] (Ran Peng, Zhang Shufang, "Steam Turbine Technology", Volume 48, No. 1, 2006: "Optimization Calculation Method for Thermal Power Plant Load Based on Genetic Algorithm"), the method of using genetic algorithm to establish thermal power plant load optimization model solves the problem of When the scale of the problem is enlarged and there are many variables and constraints, it is easy to fall into local optimum, which reduces the numerical stability and finally leads to the problem of difficult convergence. Although the literature [2] partly solved the problem in the literature [1], the genetic algorithm has the problem of slow calculation speed when the initial population is too large, and the algorithm does not realize the online real-time optimization function of the heating unit. In actual production It cannot be widely used, and the influence of the actual operating conditions of the unit on the heat consumption is not considered.

可以看出目前的供热机组热电负荷分配在线优化的解决办法存在一些问题，因此针对这些问题需要对供热机组热电负荷分配在线优化问题进行进一步研究，使电厂多台供热机组能够实现热电负荷最优分配，达到节能降耗的目的。It can be seen that there are some problems in the current online optimization of thermal and electrical load distribution of heating units. Therefore, further research on the online optimization of thermal and electrical load distribution of heating units is needed to solve these problems, so that multiple heating units in the power plant can achieve thermal and electrical load. Optimal allocation to achieve the purpose of saving energy and reducing consumption.

发明内容 Contents of the invention

本发明为了实现电厂多台供热机组热电负荷分配优化，使电厂的电负荷、热负荷在满足用户需求的同时能够最优分配，并减少总能耗达到节能的目的；进而提供了一种基于改进遗传算法的抽汽供热机组热电负荷分配优化方法。In order to realize the optimization of the thermoelectric load distribution of multiple heating units in the power plant, the present invention enables the power load and heat load of the power plant to be optimally distributed while meeting the needs of users, and reduces the total energy consumption to achieve the purpose of saving energy; Improved genetic algorithm optimization method for thermal and electrical load distribution of steam extraction and heating units.

本发明为解决上述技术问题采取的技术方案是：The technical scheme that the present invention takes for solving the problems of the technologies described above is:

本发明所述的基于改进遗传算法的抽汽供热机组热电负荷分配优化方法的具体过程为：The specific process of the thermal and electrical load distribution optimization method of the steam extraction heating unit based on the improved genetic algorithm described in the present invention is as follows:

步骤一、设置机组实际热耗曲线：根据试验得到每台机组的实际热耗曲线；所述实际热耗曲线是指以功率P和抽汽量为Q为自变量(横坐标)，热耗值R为因变量(纵坐标)的一族曲线，即Step 1. Set the actual heat consumption curve of the unit: obtain the actual heat consumption curve of each unit according to the test; the actual heat consumption curve refers to the independent variable (abscissa) with the power P and the extraction steam Q R is a family of curves of the dependent variable (ordinate), that is

第1台机组：R₁＝f(P₁，Q₁)；The first unit: R ₁ = f(P ₁ , Q ₁ );

第2台机组：R₂＝f(P₂，Q₂)；The second unit: R ₂ ＝f(P ₂ , Q ₂ );

……...

第n台机组：R_n＝f(P_n，Q_n)；The nth unit: R _n = f(P _n , Q _n );

步骤二、设置机组耗差修正曲线，确定机组耗差修正总系数θ_i，i为机组编号，i＝1，2，…，n，n表示机组数目：基于冷凝器背压、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度这六个因素偏离设计值时都会对热耗产生影响，然后根据厂家提供或电厂的耗差修正曲线查得每台机组的各个影响因素的热耗修正系数Δ_1iΔ_2iΔ_3i…Δ_6i，Δ_1iΔ_2iΔ_3i…Δ_6i分别为每台机组的冷凝器背压、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度的热耗修正系数；令θ_i＝Δ_1iΔ_2iΔ_3i…Δ_6i；Step 2: Set the unit consumption difference correction curve, determine the total unit consumption difference correction coefficient θ _i , i is the unit number, i=1, 2,..., n, n represents the number of units: based on the condenser back pressure, main steam pressure, When the six factors of main steam temperature, reheat pressure, reheat steam temperature and feed water temperature deviate from the design value, they will affect the heat consumption, and then find out the various influencing factors of each unit according to the consumption correction curve provided by the manufacturer or the power plant The heat consumption correction coefficient Δ _1i Δ _2i Δ _3i ... Δ _6i , Δ _1i Δ _2i Δ _3i ... Δ _6i are the condenser back pressure, main steam pressure, main steam temperature, reheat pressure, reheat steam Heat consumption correction coefficient of temperature and feed water temperature; let θ _i = Δ _1i Δ _2i Δ _3i ... Δ _6i ;

步骤三、得到机组设计热耗曲线：根据每台机组耗差修正总系数θ_i对机组实际热耗进行修正得到机组设计热耗曲线(即修正后的机组实际热耗曲线)，各台机组的设计热耗曲线为：Step 3. Obtain the design heat consumption curve of the unit: Correct the actual heat consumption of the unit according to the total coefficient θ _i of the consumption difference correction of each unit to obtain the design heat consumption curve of the unit (that is, the corrected actual heat consumption curve of the unit). The design heat consumption curve is:

第1台机组：R₁＝θ₁·f(P₁，Q₁)；The first unit: R ₁ = θ ₁ f(P ₁ , Q ₁ );

第2台机组：R₂＝θ₂·f(P₂，Q₂)；The second unit: R ₂ ＝θ ₂ ·f(P ₂ , Q ₂ );

……...

第n台机组：R_n＝θ_n·f(P_n，Q_n)；The nth unit: R _n = θ _n f(P _n , Q _n );

步骤四、(从系统中获取数据)获取各个机组的抽汽量Q_i(用其表征热负荷)和电负荷P_i：先测得各个机组的抽汽量Q_i和电负荷P_i，然后通过步骤三所述的机组设计热耗曲线得到相应的热耗R_i，i∈[1，n]，Step 4. (obtain data from the system) obtain the steam extraction Q _i of each unit (using it to represent the heat load) and electrical load P _i : first measure the extraction steam Q _i and electrical load P _i of each unit, and then The corresponding heat consumption R _i , i∈[1,n] is obtained through the unit design heat consumption curve described in step 3,

得到电厂n台机组的电负荷分别为P₁，P₂，…，P_n，抽汽量分别为Q₁，Q₂，…，Q_n(用其表征热负荷)，热耗值为R₁，R₂，…，R_n，n为机组数目；The electrical loads of n units in the power plant are respectively P ₁ , P ₂ , ..., P _n , the steam extraction quantities are Q ₁ , Q ₂ , ..., Q _n (which represent the heat load), and the heat consumption value is R ₁ , R ₂ ,..., R _n , n is the number of units;

优化的目的是得到使所有机组的总热耗值最小时的P₁，P₂，…，P_n，Q₁，Q₂，…，Q_n的分配方案，其中目标函数为：The purpose of optimization is to obtain the total heat consumption value of all units The distribution scheme of P ₁ , P ₂ ,..., P _n , Q ₁ , Q ₂ ,..., Q _n at the minimum, where the objective function is:

$\overset{&OverBar; &OverBar;}{R R} = = \frac{{P P}_{11} \times \times {R R}_{11} + + {P P}_{22} \times \times {R R}_{22} + + \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot; {P P}_{n no} \times \times {R R}_{n no}}{{P P}_{11} + + {P P}_{22} + + \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot; + + {P P}_{n no}} - - - - - - ((11))$

设定约束条件：Set constraints:

第一个约束条件为：Q_z＝Q₁+Q₂+…+Q_n＝const，P_z＝P₁+P₂+…+P_n＝const (2)The first constraint is: Q _z =Q ₁ +Q ₂ +...+Q _n =const, P _z =P ₁ +P ₂ +...+P _n =const (2)

即所有所有机组的总抽汽量Q_z和总电负荷P_z分别为常数；That is, the total steam extraction Q _z and the total electric load P _z of all units are constant respectively;

第二个约束条件为：Q_i∈(Q_imin，Q_imax)，P_i∈(P_imin，P_imin) (3)The second constraint condition is: Q _i ∈ (Q _imin , Q _imax ), P _i ∈ (P _imin , P _imin ) (3)

即每台机组的最大最小电负荷分别为：P_1min，P_1max；P_2min，P_2max；…；P_nmin，P_nmax；最大最小抽汽量分别为Q_1min，Q_1max；Q_2min，Q_2max；…；Q_nmin，Q_nnax)；That is, the maximum and minimum electrical loads of each unit are: P _1min , P _1max ; P _2min , P _2max ; ...; P _nmin , P _nmax ; the maximum and minimum steam extraction volumes are Q _1min , Q _1max ; Q _2min , Q _2max ;...; Q _nmin , Q _nmax );

步骤五、基于改进遗传算法求出满足所有机组的总热耗值最小时的各台机组的电负荷和抽汽量值：具体过程如下，Step 5. Calculate the total heat consumption value of all units based on the improved genetic algorithm Electric load and steam extraction value of each unit at the minimum hour: the specific process is as follows,

1、初始种群设定1. Initial population setting

用2n×m的矩阵则能表示初始种群：The initial population can be represented by a 2n×m matrix:

$(\begin{matrix} {P P}_{1111} {P P}_{1212} . . . . . . . . . . . . {P P}_{11 n no} {Q Q}_{1111} {Q Q}_{1212} . . . . . . . . . . . . {Q Q}_{11 n no} \\ {P P}_{21 twenty one} {P P}_{22 twenty two} . . . . . . . . . . . . {P P}_{22 n no} {Q Q}_{21 twenty one} {Q Q}_{22 twenty two} . . . . . . . . . . . . {Q Q}_{22 n no} \\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \\ {P P}_{m m 11} {P P}_{m m 22} . . . . . . . . . . . . {P P}_{mn mn} {Q Q}_{m m 11} {Q Q}_{m m 22} . . . . . . . . . . . . {Q Q}_{mn mn} \end{matrix})$

m为设定的个体数目，抽汽量Q_i和电负荷P_i均为满足第二个约束条件的随机数；上述初始种群采用满足第二个约束条件的约束编码的形式构造；m is the set number of individuals, the steam extraction Q _i and the electric load P _i are random numbers satisfying the second constraint condition; the above-mentioned initial population is constructed in the form of constraint coding satisfying the second constraint condition;

对上述初始种群中前(n-1)个机组的电负荷和抽汽量进行满足第二个约束条件的约束编码，而最后一个机组通过下式计算：The electric load and steam extraction capacity of the first (n-1) units in the above initial population are constrained to be encoded to satisfy the second constraint condition, and the last unit is calculated by the following formula:

$\{\begin{matrix} {P P}_{n no} = = {P P}_{z z} - - {P P}_{11} - - {P P}_{22} - - \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot; - - {P P}_{n no - - 11} \\ {Q Q}_{n no} = = {Q Q}_{z z} - - {Q Q}_{11} - - {Q Q}_{22} - - \cdot &Center Dot; \cdot \cdot \cdot &Center Dot; - - {Q Q}_{n no - - 11} \end{matrix} - - - - - - ((44))$

这样即可得到所有机组满足第一个约束条件和前(n-1)个机组满足第二约束条件的初始种群：In this way, the initial population of all units satisfying the first constraint condition and the first (n-1) units satisfying the second constraint condition can be obtained:

$(\begin{matrix} {P P}_{1111} {P P}_{1212} . . . . . . . . . . . . {P P}_{11 ((n no - - 11))} {Q Q}_{1111} {Q Q}_{1212} . . . . . . . . . . . . {Q Q}_{11 ((n no - - 11))} \\ {P P}_{21 twenty one} {P P}_{22 twenty two} . . . . . . . . . . . . {P P}_{22 ((n no - - 11))} {Q Q}_{21 twenty one} {Q Q}_{22 twenty two} . . . . . . . . . . . . {Q Q}_{22 ((n no - - 11))} \\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \\ {P P}_{m m 11} {P P}_{m m 22} . . . . . . . . . . . . {P P}_{m m ((n no - - 11))} {Q Q}_{m m 11} {Q Q}_{m m 22} . . . . . . . . . . . . {Q Q}_{m m ((n no - - 11))} \end{matrix})$

上式第n台机组为满足其电负荷最大值与最小值之差最大并且最大热负荷与最小热负荷不等的机组，即：The nth unit in the above formula is the unit that satisfies the largest difference between the maximum value and the minimum value of its electrical load and the maximum heat load is not equal to the minimum heat load, that is:

P_nmax-P_nmin＞P_imax-P_imin P _nmax -P _nmin >P _imax -P _imin

Q_nmax-Q_nmin≠0 _Qnmax - _Qnmin ≠0

P_nmax，P_nmin表示被选出来的第n台机组的最大电负荷与最小电负荷；P_imax，P_imin表示剩余机组的最大电负荷和最小电负荷；Q_nmax，Q_nmin表示被选出来的第n台机组的最大热负荷与最小热负荷；P _nmax , P _nmin represent the maximum electric load and minimum electric load of the selected nth unit; P _imax , P _imin represent the maximum electric load and minimum electric load of the remaining units; Q _nmax , Q _nmin represent the selected The maximum heat load and minimum heat load of the nth unit;

2、构建适应度函数：通过适应度计算，实现个体的优化选择，同时使优化结果中第n台机组也满足第二个约束条件；2. Construct the fitness function: through the fitness calculation, the optimal selection of the individual is realized, and at the same time, the nth unit in the optimization result also satisfies the second constraint condition;

不符合条件的个体为：Ineligible individuals are:

P_n＜P_min OR P_n＞P_max P _n P _max

Q_n＜Q_min OR Q_n＞Q_max Q _n < Q _min OR Q _n > Q _max

热电负荷分配优化是求目标函数的最小值，遗传算法的优化目标是找到具有最大适应度的个体，故定义适应度函数ObjV定义如下：The optimization of thermoelectric load distribution is to find the minimum value of the objective function, and the optimization goal of the genetic algorithm is to find the individual with the maximum fitness, so the definition of the fitness function ObjV is defined as follows:

1)、对于符合条件的个体：1) For eligible individuals:

P_min＜P_n＜P_max&Q_min＜Q_n＜Q_max P _min < P _n < P _max & Q _min < Q _n < Q _max

$ObjV ObjV = = \frac{11}{\overset{&OverBar; &OverBar;}{R R}} - - - - - - ((55))$

2)、对于不符合条件的个体：P_n＜P_min OR P_n＞P_max，Q_n＜Q_min OR Q_n＞Q_max.采用指数尺度变换式(1)目标函数2) For unqualified individuals: P _n P _max , Q _n <Q _min OR Q _n >Q _max . Use exponential scale transformation formula (1) Objective function

$ObjV ObjV = = \frac{11}{\overset{&OverBar; &OverBar;}{{R R}_{max max}} \times \times exp exp ((αβ αβ))} - - - - - - ((66))$

其中：当P_n＜P_min时， $β = β_{p} = \frac{P_{\min} - P_{n}}{P_{\min}}$ Where: when P _n <P _min , $β = β_{p} = \frac{P_{\min} - P_{no}}{P_{\min}}$

当P_n＞P_max时， $β = β_{p} = \frac{P_{n} - P_{\max}}{P_{\max}}$ When P _n >P _max , $β = β_{p} = \frac{P_{no} - P_{\max}}{P_{\max}}$

对于Q_n同理可得：For Q _n, it can be obtained in the same way:

当Q_n＜Q_min时， $β = β_{Q} = \frac{Q_{\min} - Q_{n}}{Q_{\min}}$ When _Qn < _Qmin , $β = β_{Q} = \frac{Q_{\min} - Q_{no}}{Q_{\min}}$

当Q_n＞Q_max时， $β = β_{Q} = \frac{Q_{n} - Q_{\max}}{Q_{\max}}$ When Q _n > Q _max , $β = β_{Q} = \frac{Q_{no} - Q_{\max}}{Q_{\max}}$

对于β_P、β_Q同时存在时，β＝max(β_P，β_Q)When β _P and β _Q exist at the same time, β=max(β _P , β _Q )

为机组在运行过程中的最大热耗，通过电厂生产和试验数据得到； is the maximum heat consumption of the unit during operation, obtained from the production and test data of the power plant;

α为常系数，目标是使得当计算出的P_n，Q_n超过设定阈值100％时，其适应度值大于满足条件下适应度值的100倍，即β＝1，exp(α)＞100；在实验中α取5；α is a constant coefficient, the goal is to make when the calculated P _n and Q _n exceed the set threshold 100%, its fitness value is greater than 100 times the fitness value under the condition, that is, β=1, exp(α)>100; α is 5 in the experiment;

这样在选择过程中，适应度小的个体将有很大的概率被淘汰掉，同时不满足第二个约束条件的个体也将有很大的概率被淘汰掉，从而实现个体的优化选择，理想状况下得到最大适应度即总热耗值最小的个体；In this way, in the selection process, individuals with low fitness will have a high probability of being eliminated, and at the same time, individuals who do not meet the second constraint will also have a high probability of being eliminated, so as to realize the optimal selection of individuals, ideally The maximum fitness, that is, the total heat consumption value the smallest individual

3、完成上步骤后，再进行基于传统的遗传算法的选择、交叉、变异过程；当遗传代数达到终止条件N代时，遗传过程终止，输出满足所有机组的总热耗值最小时的各台机组的电负荷和抽汽量值最优解、各台机组的热耗值以及相应的所有机组的最小总热耗。3. After the above steps are completed, the selection, crossover, and mutation processes based on the traditional genetic algorithm are carried out; when the genetic algebra reaches the termination condition N generations, the genetic process is terminated, and the output satisfies the total heat consumption value of all units The optimal solution of the electric load and steam extraction value of each unit at the minimum hour, the heat consumption value of each unit and the corresponding minimum total heat consumption of all units.

本发明的有益效果是：The beneficial effects of the present invention are:

1、在整个优化过程中利用了耗差分析。充分考虑了冷凝器背压变化、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度等因素对热耗的影响，充分提高了优化结果的准确性。1. The consumption difference analysis is utilized in the whole optimization process. The influence of factors such as condenser back pressure change, main steam pressure, main steam temperature, reheat pressure, reheat steam temperature, and feed water temperature on heat consumption is fully considered, which fully improves the accuracy of the optimization results.

2、优化算法的改进。在优化过程中，解决了等式约束问题，同时，对适应度函数进行了修改，大大提高了系统的优化速度和准确性。2. Improvement of optimization algorithm. During the optimization process, the equality constraint problem is solved, and at the same time, the fitness function is modified, which greatly improves the optimization speed and accuracy of the system.

利用本方法在硬件上进行具体实现时，有如下有益效果：When the method is implemented on hardware, it has the following beneficial effects:

1、可实现提供设置机组热耗曲线和耗差曲线接口，能够得到经过耗差修正的机组实际热耗曲线。用户通过电厂实际数据，设置机组的实际热耗曲线。系统根据目前参数与设计参数的偏差，查询耗差曲线，得到各个参数对机组热耗的影响值，对实际热耗曲线进行修正，得到经过修正的机组实际热耗曲线。1. It can realize the interface of setting the heat consumption curve and the consumption difference curve of the unit, and can obtain the actual heat consumption curve of the unit after the correction of the consumption difference. The user sets the actual heat consumption curve of the unit through the actual data of the power plant. According to the deviation between the current parameters and the design parameters, the system queries the consumption difference curve, obtains the influence value of each parameter on the heat consumption of the unit, corrects the actual heat consumption curve, and obtains the corrected actual heat consumption curve of the unit.

2、管理人员可以对每台机组的热负荷和电负荷的范围进行限制。2. Managers can limit the range of heat load and electric load of each unit.

3、实现了在线实时优化和离线优化两种模式。本系统通过计算机和MIS系统的交互完成整个优化过程，计算机从MIS系统中读取当前需要优化机组的状态数据，优化之后再将优化结果写入MIS系统，实现了热电负荷分配的在线实时优化。用户也可以通过手动输入对机组进行离线优化，以便对历史数据进行分析比较。3. Realized two modes of online real-time optimization and offline optimization. This system completes the entire optimization process through the interaction between the computer and the MIS system. The computer reads the status data of the unit that needs to be optimized from the MIS system, and then writes the optimization results into the MIS system after optimization, realizing the online real-time optimization of thermal power load distribution. Users can also perform off-line optimization of the unit through manual input in order to analyze and compare historical data.

具体量化效果：综合考虑了冷凝器背压变化、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度等各参数影响的情况下，以某发电厂两台200MW和两台600MW机组供热期和非供热期某工况为例，供热期，机组热耗由优化前的7654.2609KJ/KWh降为优化后的7632.4441KJ/KWh，热耗降低21.817KJ/KWh；非供热期，机组热耗由优化前的8429.5307KJ/KWh降为优化后的8373.638KJ/KWh，热耗降低55.893KJ/KWh，具有巨大的节能潜力。Specific quantitative effect: Considering the influence of various parameters such as condenser back pressure change, main steam pressure, main steam temperature, reheat pressure, reheat steam temperature, and feed water temperature, a power plant with two 200MW and two Take a certain working condition of a 600MW unit in the heating period and non-heating period as an example. During the heating period, the heat consumption of the unit is reduced from 7654.2609KJ/KWh before optimization to 7632.4441KJ/KWh after optimization, and the heat consumption is reduced by 21.817KJ/KWh; During the heating period, the heat consumption of the unit is reduced from 8429.5307KJ/KWh before optimization to 8373.638KJ/KWh after optimization, and the heat consumption is reduced by 55.893KJ/KWh, which has great potential for energy saving.

附图说明 Description of drawings

图1是基于改进遗传算法的电厂热电负荷分配在线优化方法的逻辑框图(虚线框表示改进遗传算法)；Fig. 1 is a logic block diagram of an online optimization method for power plant heat and electricity load distribution based on the improved genetic algorithm (the dotted line box represents the improved genetic algorithm);

图2是利用本发明所述的基于改进遗传算法的电厂热电负荷分配优化系统结构示意图；Fig. 2 is a schematic structural diagram of a power plant thermoelectric load distribution optimization system based on an improved genetic algorithm according to the present invention;

图3是某电厂机组热耗曲线设置截图；Figure 3 is a screenshot of the heat consumption curve setting of a power plant unit;

图4是热电负荷分配优化平台界面截图。Figure 4 is a screenshot of the interface of the thermoelectric load distribution optimization platform.

具体实施方式 Detailed ways

具体实施方式一：如图1所示，本实施方式所述的基于改进遗传算法的抽汽供热机组热电负荷分配优化方法的具体过程为：Specific implementation mode 1: As shown in Figure 1, the specific process of the method for optimizing the thermal and electrical load distribution of steam extraction and heating units based on the improved genetic algorithm described in this implementation mode is as follows:

步骤一、设置机组实际热耗曲线：根据试验得到每台机组的实际热耗曲线；所述实际热耗曲线是指以功率P和抽汽量为Q为自变量(横坐标)，热耗值R为因变量(纵坐标)的一族曲线，即Step 1. Set the actual heat consumption curve of the unit: According to the test, the actual heat consumption curve of each unit is obtained; the actual heat consumption curve refers to the independent variable (abscissa) with power P and steam extraction Q R is a family of curves of the dependent variable (ordinate), namely

……...

第n台机组：R_n＝f(P_n，Q_n)；The nth unit: R _n = f(P _n , Q _n );

……...

$\overset{&OverBar; &OverBar;}{R R} = = \frac{{P P}_{11} \times \times {R R}_{11} + + {P P}_{22} \times \times {R R}_{22} + + \cdot \cdot \cdot &Center Dot; \cdot \cdot {P P}_{n no} \times \times {R R}_{n no}}{{P P}_{11} + + {P P}_{22} + + \cdot \cdot \cdot \cdot \cdot \cdot + + {P P}_{n no}} - - - - - - ((11))$

设定约束条件：Set constraints:

即所有机组的总抽汽量Q_z和总电负荷P_z分别为常数。That is, the total steam extraction Q _z and the total electric load P _z of all units are constant respectively.

即每台机组的最大最小电负荷分别为：P_1min，P_1max；P_2min，P_2max；…；P_nmin，P_nmax；最大最小抽汽量分别为Q_1min，Q_1max；Q_2min，Q_2max；…；Q_nmin，Q_nmax；That is, the maximum and minimum electrical loads of each unit are: P _1min , P _1max ; P _2min , P _2max ; ...; P _nmin , P _nmax ; the maximum and minimum steam extraction volumes are Q _1min , Q _1max ; Q _2min , Q _2max ;...; _Qnmin , _Qnmax ;

1、初始种群设定1. Initial population setting

m为设定的个体数目，抽汽量Q_i和电负荷P_i均为满足第二个约束条件的随机数；上述初始种群采用满足第二个约束条件的约束编码的形式构造；m is the set number of individuals, the steam extraction Q _i and the electric load P _i are both random numbers satisfying the second constraint condition; the above initial population is constructed in the form of constraint code satisfying the second constraint condition;

对上述初始种群中前(n-1)个机组的电负荷和抽汽量进行满足第二个约束条件的编码，而最后一个机组通过下式计算：The electrical load and steam extraction capacity of the first (n-1) units in the above initial population are coded to meet the second constraint condition, and the last unit is calculated by the following formula:

$\{\begin{matrix} {P P}_{n no} = = {P P}_{z z} - - {P P}_{11} - - {P P}_{22} - - \cdot \cdot \cdot \cdot \cdot \cdot - - {P P}_{n no - - 11} \\ {Q Q}_{n no} = = {Q Q}_{z z} - - {Q Q}_{11} - - {Q Q}_{22} - - \cdot \cdot \cdot \cdot \cdot \cdot - - {Q Q}_{n no - - 11} \end{matrix} - - - - - - ((44))$

P_nmax-P_nmin＞P_imax-P_imin P _nmax -P _nmin >P _imax -P _imin

Q_nmax-Q_nmin≠0 _Qnmax - _Qnmin ≠0

不符合条件的个体为：Ineligible individuals are:

P_n＜P_min OR P_n＞P_max P _n P _max

Q_n＜Q_min OR Q_n＞Q_max Q _n < Q _min OR Q _n > Q _max

1)、对于符合条件的个体；P_min＜P_n＜P_max&Q_min＜Q_n＜Q_max 1), for qualified individuals; P _min <P _n <P _max &Q _min <Q _n <Q _max

2)、对于不符合条件的个体：P_n＜P_min OR P_n＞P_max，Q_n＜Q_min OR Q_n＞Q_max，采用指数尺度变换式(1)目标函数2) For unqualified individuals: P _n P _max , Q _n <Q _min OR Q _n >Q _max , use the exponential scaling formula (1) Objective function

对于Q_n同理可得：For Q _n, it can be obtained in the same way:

为机组在运行过程中的最大热耗，通过电厂生产和试验数据得到。 is the maximum heat consumption of the unit during operation, obtained from the production and test data of the power plant.

α为常系数，目标是使得当计算出的P_n，Q_n超过设定阈值100％时，其适应度值大于满足条件下适应度值的100倍，即β＝1，expα)＞100；在实验中α取5；α is a constant coefficient, and the goal is to make when the calculated P _n and Q _n exceed the set threshold 100%, its fitness value is greater than 100 times the fitness value under the condition, that is, β=1, expα)>100; In the experiment, α is set to 5;

针对本发明方法再进行进一步描述(步骤五)：Carry out further description (step 5) again for the inventive method:

遗传算法是以自然选择和遗传理论为基础，将生物进化过程中适者生存规则与群体内部染色体的随机信息交换机制相结合的高效全局寻优搜索算法。遗传算法摒弃了传统的搜索方式，模拟生物界的进化过程，采用人工进化的方式对目标空间进行随机优化搜索。它将问题中的可能解看做是群体中的一个个体，并将每个编码编成符号串的形式，模拟达尔文的遗传选择和自然淘汰的进化过程，对群体反复进行基于遗传的操作(遗传、交叉、变异)。根据预定目标的目标适应度函数对每个个体进行评价，依据适者生存、优胜劣汰的进化规则，不断得到最优的群体，同时以全局并行搜索方式来搜寻优化群体中的最优个体，以求得满足条件的最优解。Genetic Algorithm is an efficient global search algorithm based on natural selection and genetic theory, which combines the survival rule of the fittest in the process of biological evolution with the random information exchange mechanism of chromosomes within the population. The genetic algorithm abandons the traditional search method, simulates the evolution process of the biological world, and uses the artificial evolution method to carry out random optimization search on the target space. It regards the possible solutions in the problem as an individual in the group, and compiles each code into the form of a symbol string, simulating the evolution process of Darwin's genetic selection and natural elimination, and repeatedly performs genetic operations on the group (genetic , crossover, mutation). Evaluate each individual according to the target fitness function of the predetermined goal, and continuously obtain the optimal group according to the evolutionary rules of survival of the fittest and survival of the fittest, and at the same time search for the optimal individual in the optimized group with a global parallel search method, in order to The optimal solution that satisfies the conditions is obtained.

遗传算法的一般过程是：设置初始种群，计算适应度，选择，交叉，变异，产生新种群，重新计算适应度，依次循环迭代，直到迭代次数达到初始设定值，遗传结束，得到的最后一代种群为最优种群，种群里的个体为最优个体。在本例中，我们优化的目的是使所有机组的总热耗值最小，其中The general process of the genetic algorithm is: set the initial population, calculate the fitness, selection, crossover, mutation, generate a new population, recalculate the fitness, and iterate in turn until the number of iterations reaches the initial set value, the genetic end, and the last generation is obtained The population is the optimal population, and the individuals in the population are the optimal individuals. In this example, our goal of optimization is to make the total heat consumption value of all units minimum, of which

$\overset{&OverBar; &OverBar;}{R R} = = \frac{{P P}_{11} \times \times {R R}_{11} + + {P P}_{22} \times \times {R R}_{22} + + . . . . . . . . . . . . {P P}_{n no} \times \times {R R}_{n no}}{{P P}_{11} + + {P P}_{22} + + . . . . . . . . . . . . + + {P P}_{n no}} - - - - - - ((11))$

在界面中用户输入或者从系统中实时获取每个机组的抽汽量Q_i(用其表征热负荷)和电负荷P_i，可以通过查实际热耗曲线取得每个电负荷和抽汽量相应的热耗R_i(i∈[1，n])，通过R_i和上式便可以计算出所有机组的总热耗。优化方法如下：In the interface, the user inputs or obtains the extraction steam quantity Q _i (which represents the heat load) and the electric load P _i of each unit in real time from the system, and the corresponding electric load and the steam extraction quantity of each unit can be obtained by checking the actual heat consumption curve. The heat consumption R _i (i∈[1,n]), through R _i and the above formula can calculate the total heat consumption of all units. The optimization method is as follows:

1、初始种群设定1. Initial population setting

由于遗传算法不能直接处理问题空间的参数，因此必须通过编码把要求问题的可行解表示成遗传空间的染色体或者个体。常用的编码方法有二进制编码，格雷码编码，多级参数编码，有序串编码等。由于本优化问题是多维、高精度要求的连续函数优化问题，使用二进制等编码来表示个体会有一些不利之处，人们在一些经典优化算法的研究中所总结的一些宝贵经验也就无法加以利用，也不便于处理非平凡的约束条件。为了克服二进制编码方法的缺点，本文采用浮点数编码。对于种群设置的要求，每个个体必须是该优化问题的可行解，这样优化才有实际意义。举例说明：如果式(1)是一个纯数学的方程式，P_i、Q_i的任意取值均是式(1)的可行解，则编码过程只要生成一个2n*m的矩阵则能表示初始种群：Since the genetic algorithm cannot directly deal with the parameters of the problem space, the feasible solution of the required problem must be expressed as chromosomes or individuals in the genetic space through coding. Commonly used encoding methods include binary encoding, Gray code encoding, multi-level parameter encoding, and ordered string encoding. Since this optimization problem is a multi-dimensional, high-precision continuous function optimization problem, using binary codes to represent individuals will have some disadvantages, and some valuable experience that people have summarized in the research of some classic optimization algorithms cannot be used. , and is not easy to deal with non-trivial constraints. In order to overcome the shortcomings of the binary encoding method, this paper adopts floating-point encoding. For the requirements of the population setting, each individual must be a feasible solution to the optimization problem, so that the optimization has practical significance. For example: if equation (1) is a purely mathematical equation, and any value of P _i and Q _i is a feasible solution of equation (1), then the encoding process only needs to generate a 2n*m matrix to represent the initial population :

m为设定的个体数目，P_i、Q_i均为随机数。这样的编码称为为无约束编码。m is the set number of individuals, and P _i and Q _i are both random numbers. Such encodings are called unconstrained encodings.

但是在实际过程中每个机组的电负荷和抽汽量有上下界，即Q_max、Q_min和P_max、P_min，所有机组总体的电负荷和抽汽量为定值，即However, in the actual process, the electrical load and steam extraction of each unit have upper and lower bounds, namely Q _max , Q _min and P _max , P _min , and the overall electrical load and steam extraction of all units are fixed values, namely

Q_z＝Q₁+Q₂+…+Q_n＝const，P_z＝P₁+P₂+…+P_n＝constQ _z =Q ₁ +Q ₂ +...+Q _n =const, P _z =P ₁ +P ₂ +...+P _n =const

故电负荷和抽汽量为等式约束条件，而遗传算法由于其随机性，很难解决等式约束，或者即使采用有约束的编码，则初始种群则是在一个限定条件下产生的，这样就违背了遗传算法模拟生物进化的原则，个体的随机性和多样性受到了限制，算法的优化效果会大打折扣。所以我们的创新在于通过一系列的变换解决了等式约束的问题，并且能保证优化出来的最终个体均为最优解。Therefore, the electrical load and steam extraction are equality constraints, and the genetic algorithm is difficult to solve the equality constraints due to its randomness, or even if a constrained coding is used, the initial population is generated under a limited condition, so It violates the principle of genetic algorithm to simulate biological evolution, the randomness and diversity of individuals are limited, and the optimization effect of the algorithm will be greatly reduced. Therefore, our innovation lies in solving the problem of equality constraints through a series of transformations, and can ensure that the final individual optimized is the optimal solution.

已知约束条件1为：Q_z＝Q₁+Q₂+…+Q_n＝const，P_z＝P₁+P₂+…+P_n＝const (2)Known constraint condition 1 is: Q _z =Q ₁ +Q ₂ +...+Q _n =const, P _z =P ₁ +P ₂ +...+P _n =const (2)

即所有所有机组的总抽汽量Q_z和总电负荷P_z分别为常数。That is, the total steam extraction Q _z and the total electric load P _z of all units are constant respectively.

约束条件2为：Q_i∈(Q_min，Q_max)，P∈(P_min，P_max) (3)Constraint condition 2 is: Q _i ∈ (Q _min , Q _max ), P ∈ (P _min , P _max ) (3)

为了要使编码满足机组运行的实际情况，即满足约束条件1，我们将编码修改为对上述初始种群中前(n-1)个机组的电负荷和抽汽量进行满足第二个约束条件的编码，而最后一个机组通过下式计算：In order to make the coding meet the actual situation of unit operation, that is, to meet the constraint condition 1, we modify the coding to satisfy the second constraint condition for the electrical load and steam extraction of the first (n-1) units in the initial population above. code, while the last unit is calculated by:

$\{\begin{matrix} {P P}_{n no} = = {P P}_{z z} - - {P P}_{11} - - {P P}_{22} - - \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot; - - {P P}_{n no - - 11} \\ {Q Q}_{n no} = = {Q Q}_{z z} - - {Q Q}_{11} - - {Q Q}_{22} - - \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot; - - {Q Q}_{n no - - 11} \end{matrix} - - - - - - ((44))$

得到的初始种群为：The resulting initial population is:

P_nmax-P_nmin＞P_imax-P_imin P _nmax -P _nmin >P _imax -P _imin

Q_nmax-Q_nmin≠0 _Qnmax - _Qnmin ≠0

P_nmax，P_nmin表示被选出来的第n台机组的最大电负荷与最小电负荷；P_imax，P_imin表示剩余机组的最大电负荷和最小电负荷；Q_nmax，Q_nmin表示被选出来的第n台机组的最大热负荷与最小热负荷。P _nmax , P _nmin represent the maximum electric load and minimum electric load of the selected nth unit; P _imax , P _imin represent the maximum electric load and minimum electric load of the remaining units; Q _nmax , Q _nmin represent the selected The maximum heat load and minimum heat load of the nth unit.

这样不仅初始种群的设置时完全随机的，第一个约束条件(式(2))也很好的满足了。但是对于式(3)，初始种群进行编码的时候不能很好的满足，也就是说，编码过程中会产生一些无意义的个体使得：In this way, not only the setting of the initial population is completely random, but also the first constraint condition (Formula (2)) is well satisfied. But for formula (3), the initial population cannot be well satisfied when encoding, that is, some meaningless individuals will be generated during the encoding process so that:

P_n＜P_min OR P_n＞P_max P _n P _max

Q_n＜Q_min OR Q_n＞Q_max Q _n < Q _min OR Q _n > Q _max

我们需要在后面的优化的过程中尽量将这些个体舍去，剩下来的就是符合条件的个体。We need to discard these individuals as much as possible in the subsequent optimization process, and the rest are eligible individuals.

2、适应度函数2. Fitness function

适应度函数是用来区分群体中个体好坏的标准，是进行自然选择的唯一依据。热电负荷分配优化是求函数的最小值，遗传算法的优化目标是找到具有最大适应度的个体，故定义适应度函数ObjV定义如下：The fitness function is the standard used to distinguish good and bad individuals in the group, and it is the only basis for natural selection. The optimization of thermoelectric load distribution is to seek the minimum value of the function, and the optimization goal of the genetic algorithm is to find the individual with the maximum fitness, so the definition of the fitness function ObjV is defined as follows:

1)、对于符合条件的个体：P_min＜P_n＜P_max&Q_min＜Q_n＜Q_max 1) For eligible individuals: P _min <P _n <P _max &Q _min <Q _n <Q _max

对于Q_n同理可得：For Q _n, it can be obtained in the same way:

α为常系数，目标是使得当计算出的P_n，Q_n超过设定阈值100％时，其适应度值大于满足条件下适应度值的100倍，即β＝1，exp(α)＞100；在实验中α取5。α is a constant coefficient, the goal is to make when the calculated P _n and Q _n exceed the set threshold 100%, its fitness value is greater than 100 times the fitness value under the condition, that is, β=1, exp(α)>100; α is 5 in the experiment.

这样我们就能得出这样的结论，当通过种群计算出的最后一个机组的电负荷和抽汽量Pn，Qn不满足条件时，其适应度值会根据其超出设定阈值的程度进行放大，超出越多，放大越厉害(指数增长)。(由式(5)和式(6)可知，超出设定阈值的程度越放大，适应度取值就越小)这样在选择过程中，适应度小的个体将有很大的概率被淘汰掉，从而实现个体的优化选择，理想状况下得到最大适应度即总热耗值最小的个体。In this way, we can draw the conclusion that when the electrical load and steam extraction volume Pn and Qn of the last unit calculated through the population do not meet the conditions, its fitness value will be enlarged according to the degree beyond the set threshold, The more it is exceeded, the greater the amplification (exponential growth). (From formula (5) and formula (6), it can be seen that the greater the degree of exceeding the set threshold, the smaller the fitness value) In this way, in the selection process, individuals with low fitness will have a high probability of being eliminated , so as to realize the optimal selection of individuals, and under ideal conditions, the maximum fitness is obtained, that is, the total heat consumption value smallest individual.

3、选择3. Choose

选择操作从旧群体中以一定的概率选择优良个体组成新的种群，以繁殖得到下一代个体。个体被选中的概率跟适应度值有关，个体适应度越高，被选中的概率越大。本文采用轮盘赌法，即基于适应度比例的选择策略，个体被选中的概率为：The selection operation selects excellent individuals from the old population with a certain probability to form a new population to reproduce and obtain the next generation of individuals. The probability of an individual being selected is related to the fitness value, the higher the individual fitness, the greater the probability of being selected. This paper uses the roulette method, which is a selection strategy based on the fitness ratio. The probability of an individual being selected is:

${p p}_{i i} = = \frac{{F f}_{i i}}{{&Sum; &Sum;}_{j j}^{N N} {F f}_{j j}}$

Fi为该个体的适应度值，为所有个体适应度值之和。Fi is the fitness value of the individual, is the sum of all individual fitness values.

4、交叉操作4. Cross operation

由于本文采用浮点数编码，故相应的交叉策略选取算术交叉，是由两个个体的线性组合而产生出两个新的个体。假设在两个个体X_A，X_B之间进行算术交叉，则由算术运算后产生的两个新个体为：Since this paper adopts floating-point number encoding, the corresponding crossover strategy is arithmetic crossover, which produces two new individuals by the linear combination of two individuals. Assuming that arithmetic crossover is performed between two individuals X _A and X _B , the two new individuals produced by the arithmetic operation are:

X′_A＝aX_B+(1-a)X_A X′ _A ＝aX _B +(1-a)X _A

X′_B＝aX_A+(1-a)X_B X' _B ＝aX _A +(1-a)X _B

其中a为一个参数，a可以是一个常数，也可以是由进化代数所决定的变量。本文采用设置a为一个常数0.8。Where a is a parameter, and a can be a constant or a variable determined by the evolution algebra. In this paper, a is set as a constant 0.8.

5、变异5. Variation

变异能够改善遗传算法的局部搜索能力并能维持种群的多样性。常用的变异策略有基本位变异，均匀变异，边界变异等。热电负荷分配问题是复杂的非线性问题，能有很好的效果接近最优解，但难以确定性得搜索到最优解，为了解决这个问题，本文采用了高斯近似变异，能够改善遗传算法对重点搜索区域的局部搜索能力，并有一定概率使算法跳出局部极小点。具体操作时用符合均值为原有参数值，方差为原有参数值平方的正态分布的一个随机数来替换原有的基因值，由正态分布的特性可知，高斯变异也是重点搜索原有个体附近的局部区域。具体公式如下：Mutation can improve the local search ability of genetic algorithm and maintain the diversity of the population. Commonly used mutation strategies include basic bit mutation, uniform mutation, and boundary mutation. The heat and electricity load allocation problem is a complex nonlinear problem, which can be close to the optimal solution with good results, but it is difficult to search for the optimal solution with certainty. In order to solve this problem, this paper adopts Gaussian approximate variation, which can improve the genetic algorithm for The local search ability of the key search area, and a certain probability to make the algorithm jump out of the local minimum point. In the specific operation, replace the original gene value with a random number that conforms to the normal distribution whose mean is the original parameter value and whose variance is the square of the original parameter value. From the characteristics of the normal distribution, Gaussian variation is also the focus of searching for the original A local area near an individual. The specific formula is as follows:

$f f ((x x)) = = \frac{11}{\sqrt{22 π π} {q q}^{22}} {e e}^{- - \frac{{((x x - - q q))}^{22}}{22 {q q}^{22}}}$

其中q为原有基因值。Where q is the original gene value.

6、终止条件6. Termination conditions

遗传代数达到50代时，遗传过程自动终止，输出最优解和相应的最小总热耗。When the genetic algebra reaches 50 generations, the genetic process is automatically terminated, and the optimal solution and the corresponding minimum total heat consumption are output.

利用本发明方法的实施例(参考图1～4)：Utilize the embodiment of the inventive method (with reference to Fig. 1～4):

以某电厂4台机组为例：Take 4 units of a power plant as an example:

1对每台机组热耗曲线进行设置。进行现场试验，得出每台机组不同抽汽量下的热耗曲线。根据每台机组冷凝器背压变化、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度等因素偏离设计值的大小，查询耗差曲线，对热耗曲线进行修正。得到最终的进过修正的热耗曲线，如图3所示。1. Set the heat consumption curve of each unit. Field tests were carried out to obtain the heat consumption curves of each unit under different steam extraction volumes. According to the variation of condenser back pressure of each unit, main steam pressure, main steam temperature, reheat pressure, reheat steam temperature, feed water temperature and other factors deviate from the design value, query the consumption difference curve and correct the heat consumption curve. The final corrected heat consumption curve is obtained, as shown in Figure 3.

2对优化条件进行设置。输入每台机组的最大/小功率，最大/小抽汽量，如图4所示。2 Set the optimization conditions. Input the maximum/minimum power and maximum/minimum steam extraction of each unit, as shown in Figure 4.

3选择“自动优化”，进行优化。优化结果如图4所示，可以看出经过该系统优化后，电厂热耗减少了45.1kJ/kWh。优化后的结果可以写入MIS系统，运行人员根据结果对汽轮机进行控制。3 Select "Automatically optimize" to optimize. The optimization results are shown in Figure 4. It can be seen that after the optimization of the system, the heat consumption of the power plant is reduced by 45.1kJ/kWh. The optimized results can be written into the MIS system, and the operator can control the steam turbine according to the results.

4管理计算机从MIS系统中读取当前机组状态数据(冷凝器背压变化、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度)，得到目前用户需求的电负荷Q、热负荷P，通过查询管理计算机中的各机组设计工况下热耗曲线C和耗差修正曲线得到当前各机组热耗值R及电厂总热耗值 4 The management computer reads the current unit status data (condenser back pressure change, main steam pressure, main steam temperature, reheat pressure, reheat steam temperature, feed water temperature) from the MIS system, and obtains the current electric load Q, Heat load P, the current heat consumption value R of each unit and the total heat consumption value of the power plant are obtained by querying the heat consumption curve C and the consumption difference correction curve of each unit in the management computer

5管理人员可以通过计算机对每台机组的优化条件(如电负荷、热负荷的范围)进行限制。5. Managers can limit the optimization conditions (such as the range of electric load and thermal load) of each unit through the computer.

6利用改进遗传算法进行优化。输出优化结果，即每台机组的最优热电负荷分配方案和电厂热耗值、降低热耗值。6 Use the improved genetic algorithm to optimize. Output the optimization results, that is, the optimal heat and electricity load distribution plan for each unit, the heat consumption value of the power plant, and the heat consumption reduction value.

7优化结果输入MIS系统，运行人员根据优化结果对机组参数进行修改，实现对机组的控制。7 The optimization results are input into the MIS system, and the operating personnel modify the parameters of the unit according to the optimization results to realize the control of the unit.

如图2所示，利用本发明所述方法的软件写入在图2中的管理计算机中，供电厂管理人员进行节能监测并调整。管理计算机从MIS系统中读取当前机组的各项参数：电负荷、热负荷、冷凝器背压、主蒸汽压力、主蒸汽温度、再热压力、再热蒸汽温度、给水温度。通过查询机组设计热耗曲线和耗差修正曲线，得到当前各机组热耗值R₁，R₂，…，R_n和电厂总热耗值。管理人员可以通过管理计算机对每台机组的优化条件(最大最小电负荷P_1min，P_1max，P_2min，P_2max，…，P_nmin，P_nmax和最大最小抽汽量Q_1min，Q_1max，Q_2min，Q_2max，…，Q_nmin，Q_nmax)进行限制。As shown in Figure 2, the software using the method of the present invention is written in the management computer in Figure 2, and the management personnel of the power supply plant can monitor and adjust energy saving. The management computer reads various parameters of the current unit from the MIS system: electric load, heat load, condenser back pressure, main steam pressure, main steam temperature, reheat pressure, reheat steam temperature, and feed water temperature. By querying the unit design heat consumption curve and consumption difference correction curve, the current heat consumption values R ₁ , R ₂ ,..., R _n of each unit and the total heat consumption value of the power plant are obtained . Managers can optimize the conditions of each unit through the management computer (maximum and minimum electrical loads P _1min , P _1max , P _2min , P _2max , ..., P _nmin , P _nmax and maximum and minimum steam extraction Q _1min , Q _1max , Q _2min , Q _2max ,..., Q _nmin , Q _nmax ) to limit.

Claims

1. A method for optimizing the distribution of thermal and electrical loads of steam extraction heating units based on improved genetic algorithm, characterized in that: the specific process of the method is:

Step 1. Set the actual heat consumption curve of the unit: According to the test, the actual heat consumption curve of each unit is obtained; the actual heat consumption curve refers to the electric load P and steam extraction Q as independent variables, and the heat consumption value R' A family of curves for variables, namely

The first unit: R ₁ ′=f(P ₁ ,Q ₁ );

The second unit: R ₂ ′=f(P ₂ ,Q ₂ );

...

Unit n: R _n ′=f(P _n ,Q _n );

Step 2: Set the unit consumption difference correction curve, determine the total unit consumption difference correction coefficient θ _i , i is the unit number, i=1,2,...,n, n represents the number of units: based on the condenser back pressure, main steam pressure, When the six factors of main steam temperature, reheat pressure, reheat steam temperature and feed water temperature deviate from the design value, they will affect the heat consumption, and then find out the various influencing factors of each unit according to the consumption correction curve provided by the manufacturer or the power plant The heat consumption correction coefficients △ _1i , △ _2i , △ _3i ,..., △ _6i , △ _1i , △ _2i , △ _3i ,..., △ _6i are the condenser back pressure, main steam pressure and main steam temperature of each unit respectively , reheat pressure, reheat steam temperature, heat consumption correction coefficient of feed water temperature; let θ _i = △ _1i △ _2i △ _3i … △ _6i ;

Step 3. Obtain the design heat consumption curve of the unit: Correct the actual heat consumption of the unit according to the total coefficient θ _i of the consumption difference correction of each unit to obtain the design heat consumption curve of the unit. The design heat consumption curve of each unit is:

The first unit: R ₁ = θ ₁ f(P ₁ ,Q ₁ );

The second unit: R ₂ =θ ₂ ·f(P ₂ ,Q ₂ );

...

The nth unit: R _n = θ _n f(P _n , Q _n );

Step 4. Obtain the steam extraction Q _i and electrical load P _i of each unit: first measure the extraction steam Q _i and electrical load P _i of each unit, and then obtain the corresponding Heat consumption R _i , i∈[1,n],

The electrical loads of n units in the power plant are respectively P ₁ , P ₂ ,…,P _n , the steam extraction quantities are Q ₁ ,Q ₂ ,…,Q _n , and the heat consumption values are R ₁ ,R ₂ ,…,R _n , n is the number of units;

The purpose of optimization is to obtain the total heat consumption value of all units The allocation scheme of P ₁ , P ₂ ,…,P _n ,Q ₁ ,Q ₂ ,…,Q _n at the minimum, where the objective function is:

\overset{&OverBar;}{R} = \frac{P_{1} \times R_{1} + P_{2} \times R_{2} + . . . + P_{no} \times R_{no}}{P_{1} + P_{2} + . . . + P_{no}} - - - (1)

Set constraints:

The first constraint is: Q _z =Q ₁ +Q ₂ +...+Q _n =const, P _z =P ₁ +P ₂ +...+P _n =const (2)

That is, the total steam extraction Q _z and the total electric load P _z of all units are constant respectively;

The second constraint condition is: Q _i ∈(Q _imin ,Q _imax ),P _i ∈(P _imin ,P _imax ) (3)

That is, the minimum and maximum electrical loads of each unit are: P _1min , P _1max ; P _2min , P _2max ; ...; P _nmin , P _nmax ; the minimum and maximum steam extraction volumes are Q _1min , Q _1max ; Q _2min , Q _2max ;…; Q _nmin , Q _nmax ;

Step 5. Based on the improved genetic algorithm, calculate the electric load and steam extraction value of each unit when the total heat consumption value R of all units is the smallest: the specific process is as follows,

a. Initial population setting

The initial population can be represented by an m×2n matrix:

(\begin{matrix} {P P}_{1111} {P P}_{1212} . . . . . . . . . . . . {P P}_{11 n no} {Q Q}_{1111} {Q Q}_{1212} . . . . . . . . . . . . {Q Q}_{11 n no} \\ {P P}_{21 twenty one} {P P}_{22 twenty two} . . . . . . . . . . . . {P P}_{22 n no} {Q Q}_{21 twenty one} {Q Q}_{22 twenty two} . . . . . . . . . . . . {Q Q}_{22 n no} \\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \\ {P P}_{m m 11} {P P}_{m m 22} . . . . . . . . . . . . {P P}_{mn mn} {Q Q}_{m m 11} {Q Q}_{m m 22} . . . . . . . . . . . . {Q Q}_{mn mn} \end{matrix})

m is the set number of individuals, n is the number of units, the extraction steam Q _i and the electric load P _i are random numbers that satisfy the second constraint condition; the above initial population adopts the form of constraint code that satisfies the second constraint condition structure;

The electric load and steam extraction capacity of the first (n-1) units in the above initial population are constrained to be encoded to satisfy the second constraint condition, and the last unit is calculated by the following formula:

\{\begin{matrix} {P P}_{n no} = = {P P}_{z z} - - {P P}_{11} - - {P P}_{22} - - . . . . . . - - {P P}_{n no - - 11} \\ {Q Q}_{n no} = = {Q Q}_{z z} - - {Q Q}_{11} - - {Q Q}_{22} - - . . . . . . - - {Q Q}_{n no - - 11} \end{matrix} - - - - - - ((44))

In this way, we can obtain the initial population of constraint codes in which all units satisfy the first constraint condition and the first (n-1) units satisfy the second constraint condition:

(\begin{matrix} {P P}_{1111} {P P}_{1212} . . . . . . . . . . . . {P P}_{11 ((n no - - 11))} {Q Q}_{1111} {Q Q}_{1212} . . . . . . . . . . . . {Q Q}_{11 ((n no - - 11))} \\ {P P}_{21 twenty one} {P P}_{22 twenty two} . . . . . . . . . . . . {P P}_{22 ((n no - - 11))} {Q Q}_{21 twenty one} {Q Q}_{22 twenty two} . . . . . . . . . . . . {Q Q}_{22 ((n no - - 11))} \\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . \\ {P P}_{m m 11} {P P}_{m m 22} . . . . . . . . . . . . {P P}_{m m ((n no - - 11))} {Q Q}_{m m 11} {Q Q}_{m m 22} . . . . . . . . . . . . {Q Q}_{m m ((n no - - 11))} \end{matrix})

The nth unit in the above formula is the unit that satisfies the largest difference between the maximum value and the minimum value of its electrical load and the maximum steam extraction volume is not equal to the minimum steam extraction volume, that is:

P _nmax -P _nmin >P _imax -P _imin

_Qnmax - _Qnmin ≠0

P _nmax , P _nmin represent the maximum electric load and minimum electric load of the selected nth unit; P _imax , P _imin represent the maximum electric load and minimum electric load of any remaining unit i; Q _nmax , Q _nmin represent the The maximum steam extraction volume and the minimum steam extraction volume of the nth unit selected;

b. Construct the fitness function: realize the optimal selection of individuals through fitness calculation, and at the same time make the nth unit in the optimization result also meet the second constraint condition;

Ineligible individuals are:

P _n P _max

Q _n <Q _min OR Q _n >Q _max

The optimization of heat and electricity load distribution is to find the minimum value of the objective function, and the optimization goal of the genetic algorithm is to find the

Individuals with fitness, so the fitness function ObjV is defined as follows:

1) For eligible individuals: P _min <P _n <P _max &Q _min <Q _n <Q _max

ObjV ObjV = = \frac{11}{\overset{&OverBar; &OverBar;}{R R}} - - - - - - ((55))

2) For unqualified individuals: P _n P _max , Q _n <Q _min OR Q _n >Q _max , the objective function is represented by the exponential scaling formula (1)

ObjV ObjV = = \frac{11}{\overset{&OverBar; &OverBar;}{{R R}_{max max}} \times \times exp exp ((αβ αβ))} - - - - - - ((66))

Among them: when P _n <P _min ,

β = β_{p} = \frac{P_{\min} - P_{no}}{P_{\min}}

When P _n >P _max ,

β = β_{p} = \frac{P_{no} - P_{\max}}{P_{\max}}

For Q _n, it can be obtained in the same way:

When _Qn < _Qmin ,

β = β_{Q} = \frac{Q_{\min} - Q_{no}}{Q_{\min}}

When _Qn > _Qmax ,

β = β_{Q} = \frac{Q_{no} - Q_{\max}}{Q_{\max}}

When β _P and β _Q exist at the same time, β=max(β _P ,β _Q )

is the maximum heat consumption of the unit during operation, obtained from the production and test data of the power plant;

α is a constant coefficient, the goal is to make when the calculated P _n and Q _n exceed the set threshold 100%, its fitness value is greater than 100 times the fitness value under the condition, that is, β=1, exp(α)>100; α is 5 in the experiment;

In this way, in the selection process, individuals with low fitness will have a high probability of being eliminated, and at the same time, individuals that do not meet the second constraint will also have a high probability of being eliminated, so as to realize the optimal selection of individuals, ideally The maximum fitness, that is, the total heat consumption value the smallest individual

c. After the above steps are completed, the selection, crossover, and mutation processes based on the traditional genetic algorithm are carried out; when the genetic algebra reaches the termination condition N generations, the genetic process is terminated, and the output satisfies the total heat consumption value of all units The optimal solution of the electric load and steam extraction value of each unit at the minimum hour, the heat consumption value of each unit and the corresponding minimum total heat consumption of all units.