JPH0445210A - Method for controlling distribution of charged material in blast furnace - Google Patents

Abstract

PURPOSE:To stably keep heat load of a stave and to secure the adequate distribution of gas flow by using neural network based on each measured value of sensors in a blast furnace, obtaining variation tendency of the center flow of the furnace gas and controlling a swinging chute based on this. CONSTITUTION:The data from various kinds of the sensors 9 are inputted in data input means 5 from a process computer. Based on this inputted data, tendency index of gas distribution is obtd. from the values of distribution of the fixed sondes and gas utilization ratio distribution in shaft part at each position to radius direction in the furnace with the first neural network 6-1. By using this index and time sequential data of the heat load of stave, brick temp. on the stave, etc., the variation tendency of center flow of the gas in the furnace per the time sequential unit and the variation tendency of furnace wall flow of the gas are obtd. with the second neural network 6-2. This output is given to the swinging chute control unit 8.

Description

Detailed Description of the Invention [Industrial Application Field 1] The present invention is directed to a bell-less blast furnace system for stably maintaining stave heat load within a control range during blast furnace operation and ensuring appropriate gas flow distribution. This invention relates to a method for controlling distribution of materials.

[Prior Art] In a bell-less blast furnace, the conventional method for managing the furnace body heat load, that is, the gas flow in the furnace, is for the blast furnace operator to qualitatively judge the information from various sensors installed in the blast furnace. The method used is to evaluate the internal gas flow and heat load of the furnace body and change the tilting pattern of the rotating chute, but the results of this evaluation vary depending on the ability and experience of the operator, and the operational There were problems in that it was difficult to standardize actions and the evaluation was not quantitative, making operational analysis difficult.

In order to solve such problems, Japanese Patent Application Laid-Open No. 58-87209
A method has been proposed in which the gas flow in the furnace is quantitatively evaluated and the tilting position of the rotating chute is changed, as disclosed in the above publication. The material charging method for this bellless blast furnace is to separate the gas flow in the gas flow distribution in the furnace into a center flow, intermediate flow, and furnace wall flow based on various sensor information, and then arrange these at the vertices of a triangle. ~Create a triangular diagram displaying the charging pattern in which the side between the intermediate flow and the side between the furnace wall flow and the intermediate flow are scaled with the index of the rotating chute tilting position related to the combination of the number of rotations N and the tilting angle θ, While displaying the current raw material distribution position on this triangular diagram,
Based on the diagram display, the furnace heat load, gas temperature,
The next charging pattern is determined in response to the furnace condition index represented by the gas utilization rate, and the rotating chute is adjusted to perform charging, thereby achieving an appropriate gas flow distribution in the bellless blast furnace.

[Problem to be solved by the invention]

In the method disclosed in JP-A-58-87209, gas flow is determined only from sensor information at the time of measurement, and it is difficult to grasp the adjustment effect on changes in gas flow in the furnace by adjusting the rotating chute that was previously performed by the operator. However, the final adjustment decision is still left to the operator.

Further, such distribution control is only possible by embodying the knowledge of the operator, and for this purpose there is now a method of constructing an expert system applying AI technology.

However, as mentioned above, the operator's knowledge is complex, wide-ranging, and qualitative, making it difficult to extract knowledge and making it difficult for the system to replace the operations traditionally performed by operators. There was a point.

Furthermore, a particular difficulty with conventional methods lies in determining gas flow from sensor information.

Operators judge trends by looking at trend graphs of daily average values over several days (sometimes tens of days), but there is a lot of variation, and there are also effects of other disturbances, so it is difficult to accurately determine gas flow. It was difficult to determine. Furthermore, a fixed sonde is distribution data in the radial direction of the furnace, and although it was possible for an operator to visually evaluate the distribution intuitively, it was difficult to quantify it using a unified logic.

[Means for Solving the Problems 1] The present invention provides a charge distribution control method for changing the settings of the rotating chute tilting pattern of a bell-less blast furnace based on the gas flow condition in the furnace, and the following steps (1), (2), and (2) This is a blast furnace charge distribution control method characterized by comprising:

■ Among various data such as stave heat load, stave brick temperature, fixed sonde temperature distribution, and shaft gas utilization rate distribution, we first calculate the values of the fixed sonde temperature distribution and shaft gas utilization rate distribution in the radial direction inside the furnace. The gas distribution trend index is obtained as an output using the first neural network input to the input barber unit, and ■ The obtained index and each piece of time series data such as stave heat load and stave brick temperature for the most recent days are input to the barber. Using the second neural network input to the unit, the increase/decrease tendency of the center gas flow in the furnace and the increase/decrease trend of the furnace wall flow are determined as outputs for each time series unit, and ■ The rotating chute is controlled based on this output. to do.

[Operation] The key point of the present invention is to utilize the intuitive pattern recognition ability of neural networks.

The radial distribution of temperature and gas in the furnace is a kind of pattern information. In addition, this control expresses time-series data of all kinds of information as a trend, and judges the trend as a pattern.

Judgments such as ``It has gone up'' or ``C has gone down'' are also pattern information judgments.

According to the present invention, among various blast furnace operation data, a pattern is determined by a neural network based on the tendency of increase/decrease in the center flow of the radial distribution of gas flow in the furnace or the tendency of increase/decrease in the furnace wall flow and its temporal change. Based on the pattern information and the current state value, the rotation angle pattern of the rotating chute is set. Since the tilting of the rotating chute is controlled based on this setting, the most appropriate action can be taken. Therefore, it becomes possible to operate the blast furnace with stable stave heat load and stable charging.

[Embodiment j] Hereinafter, an embodiment of the present invention will be described based on the drawings. FIG. 1 is an example of a system configuration for implementing the present invention.

Coke and ore, which are raw materials, are alternately supplied to the blast furnace 1 using a rotating chute 2. As shown in Fig. 2(a), the rotating chute 2 is a charging device with a high degree of freedom in which the rotating speed R and tilting angle α can be changed arbitrarily. The discharge speed from the upper bunker has been adjusted so that it can be loaded with Furthermore, the tilting angle of the swinging chute 2 can be changed for each swing. As shown on the vertical axis in Figure 2(b), the angle is defined as a certain number of notches (0) For example, α=55°: Notch 1 α=52°: Notch 2 α=45°: Notch 3.

By taking the 1st to 14th turns on the horizontal axis and setting the number of notches for each turning surface, the pattern shown in FIG. 2(b) can be obtained, and according to this, the turning chute control device 8 of FIG.
is activated, and the rotating chute 2 can be rotated and tilted as set.

The process computer 3 of the blast furnace has a function of centrally managing detection end information 9 at various locations in the furnace. therefore,
All measurement data necessary for the present invention and operation data calculated based on them are stored in this process computer 3 for a certain period of time, including past data.

The dedicated computer 4 that executes the present invention has the data 9
input means 5, and first and second neural networks 6 (6-1 for distribution data processing and 6-2 for time series data processing) for performing pattern judgment calculations using the input data. ). Furthermore, it has a means 7 for displaying the calculation results on a CRT, so that the results can be made known to the operator.

The above are specific configuration examples for implementing the present invention. Computers 3 and 4 may be the same device. The program of the invention runs once a day.

Next, more detailed examples will be described. Data input means 5
The data input from the process computer is as follows.

(a) Data showing the amount of heat removed from the stave heat load varnish tape, (b) Gas sampler Nt-Nto: This is the shaft gas utilization rate distribution measured by the gas sampler, and the subscript at the bottom right of N is the position in the radial direction. shows. 1 is the center side, 10 is the furnace wall side, and 2 to 9 are the intermediate points. The definition of gas utilization rate N(li! is (cl fixed sonde temperature TI-T
6: Radial gas temperature distribution measured with a fixed sonde,
T1 is the center side, T6 is the furnace wall side, and T2 to T5 are the intermediate portions.

(d) Shaft pressure loss: Difference between the measured values of the pressure gauges installed at the upper and lower stages of the shaft.

(e) Lo /Lc: Ore layer thickness from the sounding descent distance immediately before raw material charging (
Lo) and coke layer thickness (Lc) ratio.

(f) Density index: Σ ((difference distance) - (calculated charge required time)) (g) Profile meter furnace wall flat length (L) and inclination angle (θ): Measured with the profile meter The surface shape of the charged material is indexed as shown in Figure 3.

(h) Past action history: Actions taken over the past few days are quantified using the CPI index. Among these, the N value of the gas sampler and the fixed sonde temperature are shown in Figure 4 (a) and (b), for example. This is distribution information in the furnace radial direction. On the other hand, as shown in FIGS. 5(a) and 5(b), N1 to N10. Each data of T1 to TG is assigned to each unit, and a first neural network group (6-1
) has a three-layer structure, with one hidden layer in the middle. The number of units in the hidden layer is arbitrary. The internal function of the unit is that the input layer is "tthrough".
, the hidden layer and output layer are sigmoid functions. In addition to the layer structure, a bias unit that always has a value of 1 was set up and linked to each unit of the hidden layer and output layer. This also enables threshold learning of squashing functions.

Next, in addition to the gas flow tendency indexes No and To determined above, as shown in FIG.
Enter the daily average value of time series information such as CPI index for 30 days. For these, a second neural network group (6-) each has 30 input crab units and is provided with three units as output layers for instructing the center flow increase/decrease tendency, intermediate flow increase/decrease tendency, and furnace wall flow increase/decrease tendency. In this example, as shown in FIG. 6, a network structure similar to that of the first neural network group (6-11) was prepared.

First, this network provides several patterns of reference input and teacher output at that time based on the judgment results of a skilled operator.
Learn by pack propagation method. After learning, the network calculates the output in real time based on online information.6 Figure 6 shows an example of the second neural network that uses time-series data of stave heat loads. FIG. 7 shows a graph of trends resulting from online operation of this network. Along the way, the increasing tendency of the furnace wall flow becomes stronger, and as shown in Figure 7 (a), the furnace wall heat load increases, and the central flow tendency (Figure 7 (b)) and the furnace wall flow tendency (Figure 7 (
Since the change in (d)) was output, the operator took appropriate action based on the result (Fig. 7 (e)
)) We were able to restore proper gas distribution.

〔Effect of the invention〕

The present invention makes it possible to appropriately determine the gas flow distribution in a blast furnace, and enables more accurate and timely action than ever before. As a result, the stave thermal load and gas flow distribution are maintained stably, the number of snops, etc. is reduced, and unloading becomes stable.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of the overall configuration used to implement the method of the present invention, Fig. 2 is a histogram showing the pattern of the rotating chute, Fig. 3 is an explanatory diagram of the index obtained from the constant value of the furnace top profile meter-11, and Fig. 4 The figure is a graph showing an example of measured values from a gas sampler and a fixed temperature sonde, Figure 5 is a configuration diagram of a neural network for determining the distribution map shown above, and Figure 6 is a diagram of a neural network for estimating gas flow trends from time series data. The configuration diagram and FIG. 7 are graphs of actual operation examples. 1...Blast furnace 2...Swivel chute 3...-Process computer 4--Dedicated computer 5--Data input means 6--Neural network group 7--Data output display means 8...Swivel Chute control device 9... various sensors

Claims (1)

[Claims] 1. In a charge distribution control method that changes the setting of the rotating chute tilting pattern of a bell-less blast furnace based on the gas flow situation in the furnace, A gas distribution trend index is determined as an output using the first neural network in which the values at each position in the direction are input to each unit of the input layer, and the most recent number of time series data such as the determined index and stave heat load, stave brick temperature, etc. Using a second neural network that inputs daily values into each unit of the input layer, the increase/decrease trend of the center gas flow in the furnace and the increase/decrease trend of the furnace wall flow are determined as output for each time series unit. A blast furnace charge distribution control method characterized by controlling a rotating chute based on output.