CN107688877A - A kind of load distribution method based on new self-organizing grid dynamic base - Google Patents

Abstract

The present invention discloses a kind of load distribution method based on new self-organizing grid dynamic base, including：(1) initial grid dynamic base is established；(2) dynamic library searching positioning is carried out according to the instruction of full factory's total load and current complete each unit operation situation of factory, obtains sharing of load reference record；(3) assigned references are recorded with progress deviation to set to obtain new allocation result and send instruction；(4) each unit coal consumption and full factory's total consumption of coal are calculated according to final allocation result, and calculates corresponding coal consumption bias factor, record renewal grid dynamic base.On the premise of the present invention overcomes the shortcomings of the laod sharing modes such as existing intelligent optimization algorithm, take into full account the different operation characteristic of each unit, on the basis of guaranteeing safety, rational full factory laod sharing mode is provided, meet power plant safety, high-quality, economical operation needs, reach the effect of energy-saving and emission-reduction.

Description

A load distribution method based on a new self-organizing grid dynamic library

technical field

The invention relates to the field of energy-saving scheduling, and more specifically, relates to a load distribution method based on a novel self-organizing grid dynamic library.

Background technique

The core idea of the energy-saving dispatching system is to get rid of the traditional dispatching method, based on the energy consumption characteristics of each unit in the power plant, and with the help of various artificial intelligence optimization methods, to achieve the goal of operating the units in the whole plant with the lowest total energy consumption. Compared with the traditional dispatching method, the energy-saving dispatching system has changed the dispatching method of point-to-point direct control units. The dispatching center issues the load command of the whole plant, and the system automatically and rationally optimizes the load distribution of the units of the whole plant to reduce the overall coal consumption, thereby effectively Reduce the cost of power generation, and help to improve the stability and safety of the unit, and simplify the control objects of power dispatching.

Based on the above ideas, researchers have proposed a large number of intelligent optimization algorithms, including neural networks, genetic algorithms, and particle swarm optimization algorithms.

First of all, since the load scheduling of each unit in a power plant is a complex industrial problem with high dimensions, multiple constraints, and high requirements for dynamic response capabilities, although the existing intelligent algorithms have strong computing power for large-scale parameter optimization problems, they are too Relying on the established inherent model ignores factors such as unit characteristic changes, and some algorithms are prone to fall into local optimum, so the optimization results will not conform to the actual situation on site. Secondly, under the multi-constraint and multi-objective conditions, the calculation process of the optimization algorithm is complex, not only easy to fall into local optimum, but also often requires a lot of calculation time, which cannot meet the high dynamic response requirements.

Contents of the invention

In order to overcome at least one defect (deficiency) of the above-mentioned prior art, the present invention provides a load distribution method based on a novel self-organizing grid dynamic library. This method fully considers the different operating characteristics of each unit, and provides a reasonable load distribution method for the entire plant on the basis of ensuring safety, meeting the needs of safe, high-quality, and economical operation of the plant, and achieving the effect of energy saving and emission reduction.

In order to solve the problems of the technologies described above, the technical solution of the present invention is as follows:

A load distribution method based on a novel self-organizing grid dynamic library, comprising the following steps:

(1) Establish an initial grid dynamic library;

(2) Search and locate the dynamic library according to the total load command of the whole plant and the current operation status of each unit in the whole plant, and obtain the reference records of load distribution;

(3) Perform deviation setting on the allocation reference record to obtain a new allocation result and issue an instruction;

(4) Calculate the coal consumption of each unit and the total coal consumption of the whole plant according to the final distribution results, and calculate the corresponding coal consumption bias factor, record and update the grid dynamic library.

Preferably, the establishment process of the initial grid dynamic library of step (1) is:

(1a) Obtain the measured data through the thermal test of the unit, calculate the relationship between boiler efficiency and coal consumption characteristics in different load segments, and analyze the heat consumption characteristics of the steam turbine and the single factor change characteristics at the same time, and obtain the coal consumption and plant power consumption rate of each unit; obtain DCS/SIS Historical data, analyzing and obtaining the coal consumption characteristics of each unit based on the data drive; further combining the operation experience with the above thermal test analysis results and historical data analysis results for mutual verification, and finally fitting the coal consumption characteristic relationship of the unit;

(1b) Divide the operating points of each unit according to the external conditions of the operation of the main auxiliary machines. The historical operating conditions of the machine are divided into intervals. Considering that there are three main auxiliary machines, after removing the abnormal values of ambient temperature, coal calorific value, and power consumption of the feed water pump, they are divided into C1, C2, and C3 intervals according to the historical maximum/minimum values. , permutation and combination, a total of C1*C2*C3 working conditions are obtained);

(1c) Calculate the coal consumption of each unit under different working conditions in turn to form an initial record group;

(1d) Combine the different working conditions of each unit, and calculate the total coal consumption b _cp of the whole plant under all combinations, and form the initial dynamic library according to the coal consumption from low to high and the total load from low to high;

(1e) Further extract the historical operating conditions from the DCS/SIS operating data, screen out the load distribution mode with the lowest coal consumption for a single machine and the whole plant under each operating condition, and add them to the initial dynamic library in chronological order to form the initial grid dynamic library .

Preferably, in step (2), the specific process of performing dynamic library search and positioning according to the total load order of the whole plant and the operating conditions of each unit of the current whole plant is:

(2a) Search the dynamic library according to the total command of the load, obtain all M records that meet the conditions, and construct the record matrix A:

Among them: P _ij is the jth record of the i-th unit, ζ _ij is the offset factor of the j-th record of the i-th unit;

Correspondingly construct all recorded time vector T=[t ₁ ,t ₂ ,…t _M ] and coal consumption matrix B:

Among them: b _ij is the coal consumption recorded in item j of unit i;

Reconstruct all the records of the total plant coal consumption vector B _cp =[b _1,cp ,b _2,cp ,...b _M , _cp ];

Where t _j is the time of the jth record, b _j , _cp is the total coal consumption of the whole plant according to the distribution method of the jth record;

(2b) According to all M records sorted by coal consumption from low to high, select the record with the lowest coal consumption to distribute the results. The reference record of load distribution is screened out from the record, which includes the coal consumption of all units in the whole plant, and b _ij represents the coal consumption recorded in item j of unit i in the coal consumption matrix B.).

Preferably, in step (3), the specific process of setting the deviation to the allocation reference record is:

(3a) Set the coal consumption bias factor of different units under each working condition of the initial dynamic library to be 0;

(3b) On the basis of allocating reference records, consider the change of unit dynamic characteristics, and at the same time, under the condition that the total load remains unchanged, carry out load bias for each unit; if the bias factor is 0, then randomly determine the bias factor In the case of positive or negative, adjust the load of each unit according to the reference record according to the offset factor; if the offset factor is positive, then increase the load of the unit, otherwise decrease, to ensure that the total load remains unchanged after the offset, and issue it as a result Unit load command; if the bias factor is known, adjust the load of each unit proportionally according to the direction of the current bias factor to ensure that the total load remains unchanged after adjustment, and issue a distribution command.

Preferably, in step (4), the coal consumption b _i,new of each unit and the total coal consumption b _cp , _new of the whole plant are calculated according to the final distribution results, and the coal consumption bias factor of each unit under the current working condition is calculated:

Then, according to the load distribution results, coal consumption of a single unit, the total coal consumption of the whole plant, the offset factor of coal consumption, and the current operating status of each unit, the dynamic library is updated for the next call.

Compared with the prior art, the beneficial effect of the technical solution of the present invention is: the present invention divides the operating point according to the boundary conditions such as ambient temperature, extracts the optimal load distribution mode from the historical data and supplements it into the dynamic library, and takes the nearest record as a load distribution reference record. Considering the change of the dynamic characteristics of the unit, the load of each unit is biased, and the result after the bias is used as the load distribution command. On the premise of overcoming the shortcomings of the existing intelligent optimization algorithm and other load distribution methods, the present invention fully considers the different operating characteristics of each unit, and provides a reasonable load distribution method for the whole plant on the basis of ensuring safety to meet the safety, quality, and The needs of economic operation, to achieve the effect of energy saving and emission reduction.

Description of drawings

Fig. 1 is a schematic flow chart of the present invention.

detailed description

The drawings are for illustrative purposes only, and should not be construed as limitations on this patent; in order to better illustrate this embodiment, some parts in the drawings will be omitted, enlarged or reduced, and do not represent the size of the actual product;

For those skilled in the art, it is understandable that some well-known structures and descriptions thereof may be omitted in the drawings. The technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

As shown in Fig. 1, a kind of load distribution method based on novel self-organizing grid dynamic library proposed by the present invention, its realization process is:

S1. Establish an initial grid dynamic library:

11) Obtain data based on the power generation process and construct data sources; the data sources include: DCS/SIS historical data, unit thermal test data, offline data such as coal quality, and experience data of unit operators;

11) Based on the analysis of data sources, the coal consumption characteristic relationship of the unit is obtained, specifically: the measured data is obtained through the thermal test of the unit, the relationship between the boiler efficiency and the coal consumption characteristic of different load segments is calculated, and the heat consumption characteristics of the steam turbine and the single factor change characteristics are analyzed at the same time, to obtain the characteristics of each unit Coal consumption for power generation and plant power consumption rate; obtain DCS/SIS historical data, and dig out to obtain the data-driven coal consumption characteristics of each unit. Further combine the experience data of unit operators with the above thermal test analysis results and historical data excavation results for mutual verification, and finally fit the coal consumption characteristic relationship of the unit;

Under the thermal test conditions, the measured data and DCS operation data under multiple test conditions are obtained, and the system is pre-fitted into a quadratic polynomial before the system runs.

The steps of data mining method for fitting coal consumption characteristics are as follows:

Step1: Obtain DCS historical data

Step2: Judgment and screening of stable working conditions

Step3: Process and analyze the sample points under different load segments. Perform validity analysis on sample points to remove invalid working conditions; determine the number of valid working condition points for each load segment, and remove the load segment if the number of working condition points does not meet the requirements;

Step4: Fit the above stable data into a quadratic curve

Step5: Modify the constraint conditions and correct the coal consumption curve according to the current calculated working condition point or multiple stable working condition points in the near time period.

Finally, according to the operation experience of the on-site operators, observe the deviation between the above-mentioned coal consumption characteristics and the on-site operation results, and perform curve correction. Through continuous correction, it is ensured that the coal consumption characteristic curve always conforms to the current equipment status.

13) According to external conditions such as ambient temperature, coal calorific value, and operation of main auxiliary machines such as feed water pumps, the operating point of each unit is divided. Taking unit i as an example, under a given working condition point, the unit load upper and lower limit range [P _i,min ,P _i,max ], load change rate [v _i,min ,v _i,max ], load margin Limit ΔP _i,limit , critical load constraint P _i,critical , and calculate group coal consumption b _i,condition through coal consumption characteristic relationship;

14) Calculate the coal consumption of each unit under different working conditions in turn to form an initial record group.

15) Combine the different working conditions of each unit, and calculate the total coal consumption b _cp of the whole plant under all combinations, and form the initial dynamic library according to the coal consumption from low to high and the total load from low to high.

16) Further extract the historical operating conditions from the DCS/SIS operating data, screen out the load distribution mode with the lowest coal consumption of the single machine and the whole plant under each operating condition, and add them to the initial dynamic library in chronological order.

S2. Search and locate the dynamic library according to the total load command of the whole plant and the current operating conditions of the units of the whole plant, and obtain the load distribution reference records;

21) Search the dynamic library according to the total command of the load, obtain all M records that meet the conditions, and construct the record matrix A:

Among them: P _ij is the jth record of the i-th unit, ζ _ij is the offset factor of the j-th record of the i-th unit;

Correspondingly construct all recorded time vector T=[t ₁ ,t ₂ ,…t _M ] and coal consumption matrix B:

Among them: b _ij is the coal consumption recorded in item j of unit i;

Reconstruct all the records of the total plant coal consumption vector B _cp =[b _1,cp ,b _2,cp ,...b _M , _cp ];

Where t _j is the time of the jth record, b _j , _cp is the total coal consumption of the whole plant according to the distribution method of the jth record;

22) According to all M records sorted from low to high coal consumption, select the record with the lowest coal consumption to distribute the results.

S3. Perform deviation setting on the allocation reference record to obtain a new allocation result and issue an instruction;

31) Set the coal consumption bias factor of different units under each working condition of the initial dynamic library to be 0;

32) On the basis of allocating reference records, consider the change of unit dynamic characteristics, and at the same time ensure that the total load remains unchanged, carry out load bias for each unit; if the bias factor is 0, randomly determine the bias factor positive Negative situation, and adjust the load of each unit according to the reference record according to the offset factor; if the offset factor is positive, then increase the load of the unit, otherwise decrease, to ensure that the total load remains unchanged after the offset, and use this as a result to issue the unit Load command; if the bias factor is known, adjust the load of each unit proportionally according to the direction of the current bias factor to ensure that the total load remains unchanged after adjustment, and issue a distribution command

S4. Calculate the coal consumption b _i,new of each unit and the total coal consumption b _cp,new of the whole plant according to the final distribution results, and calculate the coal consumption bias factor of each unit under the current working condition based on the formula (1), and then according to the load distribution results, single unit Coal consumption, total coal consumption of the whole plant, coal consumption bias factor, and the current operating status of each unit, update the dynamic library, and prepare for the next call:

Taking four units of a certain power plant as an example, one month's network dispatching AGC instruction data is selected for real-time optimal distribution, and the typical working condition data covering the whole working condition range are selected and listed in Table 1 and Table 2.

Table 1: Optimal preload and coal consumption

Table 2: Load and coal consumption after optimization

Table 3: Comparison of coal consumption before and after optimization

Load command (MW) 1329.39 1550.52 1773.13 1975.40 2207.83 2409.46 Coal consumption before optimization (g/kWh) 344.48 334.62 330.62 329.10 328.07 326.52 Coal consumption after optimization (g/kWh) 343.08 333.95 329.10 327.15 326.60 326.07 Save coal consumption (g/kWh) 1.40 0.67 1.52 1.95 1.48 0.46 Coal saving rate/% 0.406 0.194 0.460 0.593 0.451 0.141

According to Table 1 and Table 2, the coal consumption before optimal distribution and the coal consumption rate after optimal distribution are compared in typical load segments, and combined with the coal consumption rate shown in Table 3, it can be seen that the coal consumption after optimal distribution is significantly better than that under the control of a single AGC in a power plant Coal consumption has achieved the optimization effect of saving fuel.

Apparently, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than limiting the implementation of the present invention. For those of ordinary skill in the art, other changes or changes in different forms can be made on the basis of the above description. It is not necessary and impossible to exhaustively list all the implementation manners here. All modifications, equivalent replacements and improvements made within the spirit and principles of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (5)

1. a kind of load distribution method based on new self-organizing grid dynamic base, it is characterised in that comprise the following steps：

(1) initial grid dynamic base is established；

(2) dynamic library searching positioning is carried out according to the instruction of full factory's total load and current complete each unit operation situation of factory, obtains load Assigned references record；

(3) assigned references are recorded with progress deviation to set to obtain new allocation result and send instruction；

(4) each unit coal consumption and full factory's total consumption of coal are calculated according to final allocation result, and calculates corresponding coal consumption bias factor, remembered Record renewal grid dynamic base.

2. a kind of load distribution method based on new self-organizing grid dynamic base according to claim 1, its feature exist In the process of establishing of the initial grid dynamic base of step (1) is：

(1a) obtains measured data by unit thermal test, calculates different load section boiler efficiency and coal consumption characteristic relation, together When analyze turbine heat rate characteristic, single factor test variation characteristic, obtain each unit generation coal consumption, station service power consumption rate；DCS/SIS is obtained to go through History data, analysis obtain each unit coal consumption characteristic based on data-driven；Tried further combined with operating experience and above-mentioned heating power Test analysis result and historical data analysis result is mutually authenticated, finally fit unit coal consumption characteristic relation；

(1b) is split according to the external condition of major pant item running situation to each unit operating point, and wherein major pant item is run Situation includes environment temperature, coal calorific value and feed pump；

(1c) calculates the unit coal consumption under each unit difference working condition successively, forms original records group；

Different operating modes after (1d) splits each unit are combined, and calculate all combination Xia Quan factories total consumption of coal b_cp, press According to coal consumption from low to high, the total load initial dynamic base of arrangement form from low to high；

(1e) extracts history run operating mode from DCS/SIS service datas, and it is minimum to filter out unit and full factory's coal consumption under each operating mode Laod sharing mode, be supplemented to initial dynamic base according to time order and function order, form initial grid dynamic base.

3. a kind of load distribution method based on new self-organizing grid dynamic base according to claim 2, its feature exist According to the tool of each unit operation situation progress dynamic library searching positioning of the instruction of full factory's total load and current full factory in step (2) Body process is：

(2a) always instructs search dynamic base according to load, obtains all M bars record for the condition that meets, construction record matrix A：

Wherein：P_ijRecorded for i-th unit j-th strip, ζ_ijBias factor is recorded for i-th unit j-th strip；

Correspondingly construct the time arrow T=[t of all records₁,t₂,…t_M] and coal consumption matrix B：

Wherein：b_ijCoal consumption is recorded for i-th unit j-th strip；

Reconstruct and all record full factory's total consumption of coal vector B_cp=[b_1,cp,b_2,cp,…b_M,cp]；

Wherein t_jFor the time of j-th strip record, b_j,_cpThe full factory's total consumption of coal of the method for salary distribution is recorded for j-th strip；

(2b) is recorded according to all M bars and sorted from low to high according to coal consumption, selects the minimum record of coal consumption to carry out result distribution, coal When consuming identical, record is ranked up according to time order and function, selects nearest record as sharing of load reference record.

4. a kind of load distribution method based on new self-organizing grid dynamic base according to claim 3, its feature exist In the detailed process that in step (3) assigned references are recorded with progress deviation setting is：

(3a) sets the coal consumption bias factor of different units under each operating mode of initial dynamic base as 0；

(3b) considers unit dynamic characteristic situation of change on the basis of assigned references record, while is ensureing that total load is constant In the case of, load biasing is carried out to every unit；It is random to determine the positive and negative situation of bias factor if bias factor is 0, and to ginseng Examine and record each unit load and adjusted according to bias factor；If bias factor reduces just, to increase the load of the unit, Total load is constant after ensuring biasing, and unit load instruction is assigned in this, as result；If bias factor, it is known that if according to it is current partially Put factor direction and adjust each unit load in proportion, it is ensured that total load is constant after regulation, assigns distribution instruction.

5. a kind of load distribution method based on new self-organizing grid dynamic base according to claim 4, its feature exist According to each unit coal consumption b of final allocation result calculating in step (4)_i,newAnd full factory total consumption of coal b_cp,new, and calculate each unit Coal consumption bias factor under current working：

<mrow> <mi>Z</mi> <mo>=</mo> <mo>&amp;lsqb;</mo> <msub> <mi>&amp;zeta;</mi> <mn>1</mn> </msub> <mo>,</mo> <mo>...</mo> <mo>,</mo> <msub> <mi>&amp;zeta;</mi> <mi>i</mi> </msub> <mo>,</mo> <mo>...</mo> <mo>,</mo> <msub> <mi>&amp;zeta;</mi> <mi>N</mi> </msub> <mo>&amp;rsqb;</mo> <mo>=</mo> <mo>&amp;lsqb;</mo> <mfrac> <mrow> <msub> <mi>b</mi> <mrow> <mn>1</mn> <mo>,</mo> <mi>n</mi> <mi>e</mi> <mi>w</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>b</mi> <mrow> <mn>1</mn> <mi>j</mi> </mrow> </msub> </mrow> <msub> <mi>b</mi> <mrow> <mn>1</mn> <mi>j</mi> </mrow> </msub> </mfrac> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mfrac> <mrow> <msub> <mi>b</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>n</mi> <mi>e</mi> <mi>w</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>b</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> </mrow> <msub> <mi>b</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> </mfrac> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mfrac> <mrow> <msub> <mi>b</mi> <mrow> <mi>N</mi> <mo>,</mo> <mi>n</mi> <mi>e</mi> <mi>w</mi> </mrow> </msub> <mo>-</mo> <msub> <mi>b</mi> <mrow> <mi>N</mi> <mi>j</mi> </mrow> </msub> </mrow> <msub> <mi>b</mi> <mrow> <mi>N</mi> <mi>j</mi> </mrow> </msub> </mfrac> <mo>&amp;rsqb;</mo> <mo>;</mo> </mrow>

Shape is currently run further according to sharing of load result, single fighter coal consumption, full factory's total consumption of coal, coal consumption bias factor and each unit State, dynamic base is updated, called standby next time.