CN109107744B

CN109107744B - Medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method

Info

Publication number: CN109107744B
Application number: CN201810617355.3A
Authority: CN
Inventors: 司风琪; 田书耘; 祝康平
Original assignee: Cpi Shentou Power Generation Co ltd; Southeast University
Current assignee: Cpi Shentou Power Generation Co ltd; Southeast University
Priority date: 2018-06-15
Filing date: 2018-06-15
Publication date: 2020-09-01
Anticipated expiration: 2038-06-15
Also published as: CN109107744A

Abstract

The invention discloses a medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method, which comprises the steps of periodically updating sample data; establishing a soft measurement model of the coal powder fineness by a least square-support vector machine method to obtain an online coal powder fineness estimation value; defining a subsystem boundary condition range; performing multivariate synchronous clustering on the primary air volume and the loading oil pressure in each typical output neighborhood of the coal mill by using a K-means clustering algorithm, and excavating a class with the minimum conversion unit consumption corresponding to the loading oil pressure and the primary air volume of the mill under different inlet working conditions; taking the central point as the operating reference value of the primary air quantity and the loading oil pressure of the medium-speed mill in each typical output neighborhood region; and finally fitting an operation parameter curve of the medium-speed mill. The method of the invention obtains the primary air volume and loading oil pressure reference value curve through online real-time clustering, reduces the unit consumption of coal grinding as much as possible on the premise of meeting three indexes, and realizes dynamic optimization-approaching adjustment.

Description

Medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method

Technical Field

The invention relates to a method for adjusting the air-coal ratio and the oil pressure of a coal mill, in particular to a method for dynamically optimizing the air-coal ratio and the oil pressure of a medium-speed coal mill.

Background

The medium-speed coal mill has the advantages of short start-stop time, low power consumption and the like, and is widely applied to coal-fired power plants in China. The common grinding part of the medium speed coal mill extrudes and grinds the raw coal in the grinding part under the action of spring force, hydraulic force or other external forces, and finally the raw coal is crushed into coal powder; the pulverized coal is thrown to the air ring chamber by the rotation of the grinding component, the hot air flow flowing through the air ring chamber brings the pulverized coal to a pulverized coal separator at the upper part of the medium-speed coal mill, and the coarse pulverized coal is separated and ground again. In the process, besides the coal feeding amount, various input amounts including air volume, external grinding force and even adjustment of a separator influence the power consumption of the coal mill. However, currently, the medium speed mill operation of most power plants is primarily operated with operating parameters or field experience provided by the mill manufacturer. And the operation parameter reference value that the producer provided mostly is the setting of foreign manufacturer, lacks the operation parameter that combines together with the on-the-spot operation concrete condition, leads to export buggy to cross the phenomenon emergence up to standard or not up to standard occasionally, and the wearing and tearing of intermediate speed grinding roller increase. The manual field experience is adopted, the theoretical basis is lacked, and the misoperation event occurs sometimes. The system is extremely dependent on the operation parameters given by foreign manufacturers at the end of the root, and is not combined with the field reality.

The reference value refers to each parameter value corresponding to the optimal working condition of the unit under the current operation boundary, and is generally determined through design values, variable working condition calculation, tests and the like, and the optimal target value under the given working condition of the unit can also be determined based on a similar theory and field thermodynamic test results, or the optimal operation parameters are determined by utilizing a Lagrange optimization algorithm. However, the actual operation process often deviates from the mechanism, and factors such as equipment performance degradation and coal quality deviation from the design working condition may affect the reliability and accuracy of the mechanism model.

Disclosure of Invention

Aiming at the defects of the prior art, the invention provides a medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method, which solves the problem that the operation parameters cannot be adjusted timely when the actual operation process and the set working condition have deviation.

The technical scheme of the invention is as follows: a medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method comprises the following steps:

step 1, periodically updating collected sample data;

step 2, establishing a soft measurement model of the coal powder fineness by a least square-support vector machine method to obtain an online coal powder fineness estimation value;

step 3, limiting a plurality of boundary condition ranges of the subsystem;

step 4, dividing a plurality of typical output neighborhood regions in a medium-speed grinding output range;

step 5, performing K-means clustering with the class center number of 2 to K on the primary air volume and the loading oil pressure in each typical output neighborhood region, wherein K is more than or equal to 2;

step 6, calculating a Silhouette clustering effective evaluation function, and determining the number of the class centers;

step 7, searching an optimal Silhouuette contour map for the k-means cluster with the determined class center number, and searching a class with the minimum unit consumption of corresponding reduced powder at a class center point as a reference class;

step 8, taking the central point of the reference class as the operation reference value of the primary air quantity and the loading oil pressure of the medium-speed mill in each typical output neighborhood region;

and 9, fitting a medium-speed mill operation parameter curve with the primary air volume and the loading oil pressure operation reference value of the medium-speed mill in different typical output neighborhood regions.

Further, the radial basis kernel function of the least square-support vector machine in step 2 is:

wherein x_iAs kernel function center, if x and x_iIf very close, the value of the kernel function is 1, and if very different, the value of the kernel function is approximately 0. Since this function is similar to a gaussian distribution, it is also called a gaussian kernel function. σ is a width parameter of the function, and σ controls the radial range of action of the function.

Further, the boundary condition range in the step 3 includes an ambient temperature, a coal quality, a mill outlet temperature, a coal powder fineness, an air preheater outlet temperature and a mill hot primary air main pipe pressure.

Preferably, the environment temperature is-10-0 ℃, the coal quality is 17.19-17.89 MJ/kg, the mill outlet temperature is 70-80 ℃, and the fineness of the pulverized coal is R₉₀<27%, the outlet temperature of the air preheater is 290-320 ℃, and the pressure of the primary air main pipe subjected to heat grinding is 8-9 kPa.

Preferably, in the step 5, K is equal to or greater than (sample data number)^1/2And closest (sample data size)^1/2Is an integer of (1).

Preferably, the collection time interval is not more than 1min when the sample data is periodically updated in step 1, and the periodic update time interval is not more than 10 d.

Preferably, the number of the typical output neighborhood regions in the step 4 is not less than 6.

Preferably, the typical output neighborhood interval in the step 4 is 40 +/-0.2 t/h, 45 +/-0.2 t/h, 50 +/-0.2 t/h, 55 +/-0.2 t/h, 60 +/-0.2 t/h and 65 +/-0.2 t/h.

According to the technical scheme, the collected samples are updated periodically, a soft measurement model of the coal powder fineness is established by a least square-support vector machine method, an online coal powder fineness estimation value is obtained, a K-means clustering algorithm is further utilized, multivariate synchronous clustering is carried out on primary air volume and loading oil pressure under each typical output of a coal mill, grinding loading oil pressure and primary air volume reference values corresponding to different inlet working conditions are excavated, and the coal grinding unit consumption is reduced as far as possible on the premise that the three indexes are met. The reference value can be found through historical data and real-time data. The algorithm is simple and easy to implement, and can be applied to the field. In the actual operation of the unit, the reference value working condition library is dynamically adjusted according to the newly acquired index data so as to approach the optimal working condition of the current operation state of the mill, thereby providing operation guidance suggestions for field operators.

Drawings

FIG. 1 is a schematic flow chart of a medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method.

FIG. 2 is a Silhouette contour diagram under different centroid numbers K in the example.

FIG. 3 is a diagram of clustering results under the exemplary conditions of 55t/h in the example.

FIG. 4 is a schematic diagram illustrating the difference in powder consumption under the same output in the embodiment.

FIG. 5 is a diagram of the clustering effect under the exemplary condition of 55t/h in the example.

Fig. 6 is a primary air volume reference value distribution diagram in the example.

FIG. 7 is a map of the reference value of the charging oil pressure in the embodiment.

Detailed Description

The present invention is further illustrated by the following examples, which are not to be construed as limiting the invention thereto.

Referring to fig. 1, for example, a 600MW subcritical unit in a power plant is provided, a pulverizing system is a cold primary air positive pressure direct blowing type, model of six coal mills is ZGM123, and coal is train bituminous coal, and its components are not changed much. The method for adjusting the medium-speed mill air-coal ratio and the oil pressure dynamic optimization trend comprises the following steps:

step 1, selecting historical operating data of a mill for one month in 1 month B in 2015 in an SIS (plant-level information monitoring system), wherein data labels are coal feeding amount (output), mill outlet temperature, environment temperature, average primary air temperature of an air preheater outlet, primary hot air main pipe pressure, primary mill air amount, oil pressure loading and milling power consumption conversion. The sampling interval is 1min, after data cleaning (removing stop points and obvious abnormal points), 37088 groups of samples are stored in the working condition database in total, and the database is updated at the interval of 10d in actual operation. Because the change condition of the coal quality can not be directly read from the SIS system of the power plant, and the current measuring means is difficult to realize the on-line analysis of the coal quality, the low-level calorific value of the coal as fired in the daily registry is selected to represent the coal quality characteristic.

Step 2, establishing a soft measurement model of the coal powder fineness by a least square-support vector machine method to obtain an online coal powder fineness estimation value, wherein a radial basis kernel function of the support vector machine model is as follows:

And all parameters which can represent the running state of the coal mill and can obtain measured values are used as the input of the coal powder fineness soft measurement model, and the coal powder fineness is used as the output. R₉₀＝f(B,Q,T_in,p_oilΔ p) wherein B is the amount of coal supplied, t.h^-1(ii) a Q is primary air quantity, t.h^-1；T_inInlet air temperature, deg.C; p is a radical of_oilLoading oil pressure for the mill, MPa; Δ p is inlet-outlet differential pressure, kPa; r₉₀Is the fineness of coal powder percent.

Step 3, defining subsystem boundaryThe environmental temperature is-10-0 ℃, the coal quality is 17.19-17.89 MJ/kg, the grinding outlet temperature is 70-80 ℃, and the fineness of the pulverized coal is R₉₀<27 percent, the outlet temperature of the air preheater is 290-320 ℃, and the pressure of the primary grinding and heating air main pipe is 8-9 kPa.

And 4, maintaining the output force between 35 t/h and 70t/h under the actual operation condition of the coal mill. The neighborhood region division is carried out respectively at six typical working conditions of 40 +/-0.2 t/h, 45 +/-0.2 t/h, 50 +/-0.2 t/h, 55 +/-0.2 t/h, 60 +/-0.2 t/h and 65 +/-0.2 t/h.

And 5, respectively carrying out K-means clustering with the class center number of 2 to K on the primary air volume and the loading oil pressure in each typical output neighborhood (setting i to traverse from 1 to 6 and corresponding to the neighborhood of six typical working conditions), wherein K is more than or equal to 2, and cf is (the number of sample data)^1/2；

Step 6 and FIG. 2 show a Silhouette profile graph when K-means clustering is carried out on the samples under a typical working condition of 55 +/-0.2 t/h and the value of K is 11. Under the working condition, the Silhouette values of all data points are closer to 1 when the data points are grouped into 4 types, and the number of unreasonable points is less than 5-11 types. Therefore, the type selection 4 is more reasonable in the typical working condition.

And 7, clustering k-means with the class center number of 4, finding an optimal Silhouuette outline map according to a clustering result shown in figure 3, finding a class with the minimum reduced coal pulverizing unit consumption at a class center point as a reference class, and locally enlarging a reduced coal pulverizing power consumption-coal feeding amount map near 55t/h in figure 4. It can be seen that the portions not separated by doping together may be caused by not accurately considering moisture, ash content, grindability and other limiting conditions of the coal, but the general trend is clear, and it can be said that the clustering basically achieves the ideal effect. Because each clustering center point is extracted from mass data under the current energy efficiency operation level of the unit, the optimal operation potential of the coal mill can be truly reflected, and therefore the reference value determined by the clustering algorithm has the advantage of being practically reachable.

Step 8 and FIG. 5 indicate that the reduced pulverizing power consumption corresponding to the 4 th class clustering center is smaller, so that the primary air volume 94.18t/h and the mill loading oil pressure 12.79MPa of the 4 th class clustering center are selected as the reference values of the target parameters of the working condition.

And 9, fitting the primary air volume of the medium-speed mill and the operation reference value of the loading oil pressure in all different typical output neighborhood regions to a medium-speed mill operation parameter curve, wherein the result is shown in fig. 6 and 7.

Claims

1. A medium-speed mill air-coal ratio and oil pressure dynamic optimization-approaching adjusting method is characterized by comprising the following steps:

step 1, periodically updating collected sample data;

step 2, establishing a soft measurement model of the coal powder fineness by a least square-support vector machine method to obtain an online coal powder fineness estimated value R₉₀＝f(B,Q,T_in,p_oilΔ p), B is the amount of coal supplied, Q is the primary air volume, T_inIs the inlet air temperature, p_oilOil pressure is applied to the mill, Δ p is the inlet-outlet differential pressure, R₉₀For the fineness of the coal powder, the radial basis kernel function of the least square-support vector machine is as follows:

wherein x_iIs the kernel function center, and σ is the width parameter of the function;

step 3, limiting a plurality of boundary condition ranges of the subsystem, wherein the boundary condition ranges comprise ambient temperature, coal quality, mill outlet temperature, coal dust fineness, air preheater outlet temperature and mill heat primary air main pipe pressure;

step 9, fitting an operation parameter curve of the medium-speed mill with the operation reference values of the primary air quantity and the loading oil pressure of the medium-speed mill in different typical output neighborhood regions;

and carrying out multivariate synchronous clustering on the primary air volume and the loading oil pressure under each typical output of the coal mill, and excavating corresponding mill loading oil pressure and primary air volume reference values under different inlet working conditions to reduce the unit consumption of powder preparation on the premise of meeting the output of the coal mill, the fineness of pulverized coal and the temperature of a mill outlet.

2. The method for adjusting the wind-coal ratio and the oil pressure dynamic trend toward an optimal state in the medium-speed mill according to claim 1, wherein the ambient temperature is-10 to 0 ℃, the coal quality is 17.19 to 17.89MJ/kg, the mill outlet temperature is 70 to 80 ℃, and the fineness of the pulverized coal is R₉₀Less than 27%, the outlet temperature of the air preheater is 290-320 ℃, and the pressure of the primary air main pipe subjected to heat grinding is 8-9 kPa.

3. The method according to claim 1, wherein K is equal to or greater than (number of sample data) in step 5^1/2And closest (sample data size)^1/2Is an integer of (1).

4. The method for adjusting the medium-speed mill draft coal-to-coal ratio and the oil pressure dynamic trend toward an optimal state according to claim 1, wherein the collection time interval is not more than 1min when the sample data is periodically updated in the step 1, and the periodic update time interval is not more than 10 d.

5. The medium-speed mill wind-coal ratio and oil pressure dynamic optimization adjusting method according to claim 1, wherein the number of typical output neighborhood regions in the step 4 is not less than 6.

6. The medium-speed grinding air-coal ratio and oil pressure dynamic optimization-tending adjusting method according to claim 1, characterized in that typical output neighborhood intervals in the step 4 are 40 ± 0.2t/h, 45 ± 0.2t/h, 50 ± 0.2t/h, 55 ± 0.2t/h, 60 ± 0.2t/h and 65 ± 0.2 t/h.