CN-121995738-A - Self-learning fuzzy control method for hearth pressure of step heating furnace

CN121995738ACN 121995738 ACN121995738 ACN 121995738ACN-121995738-A

Abstract

The invention provides a self-learning fuzzy control method for furnace pressure of a stepping heating furnace, which comprises the steps of selecting two pressure taking points on two sides of two heating sections in the furnace, carrying out differential pressure detection with atmospheric pressure outside the furnace to obtain a furnace pressure actual value, setting a furnace pressure target value in a micro-positive pressure state, monitoring the furnace door action process in real time, identifying that a furnace door is in an opening stroke, a closing stroke or reaches a closed position according to a furnace door stroke detection signal, outputting a corresponding furnace door state mark, starting self-learning feedforward control during the furnace door enters the opening stroke, generating feedforward compensation quantity according to the furnace pressure history change trend, regulating the opening of a flue flashboard to inhibit pressure disturbance caused by furnace door opening from being transmitted to a flue control loop, continuously monitoring furnace pressure fluctuation until the furnace door is restored to a steady state after the furnace door reaches the closed position, exiting self-learning feedforward control and switching to conventional closed loop control, and keeping the furnace pressure in a target function interval by continuously regulating the opening of the flue flashboard.

Inventors

LIU ZHANQUAN
ZHAO GUOMING
LU JIANBIN
LIU DEHUI
SUN WEI
WANG ZHONGPENG
JIN LONG
LI MINGMIN
ZHANG WEICHUANG
YU RENFEI
CUI YAFENG

Assignees

鞍钢股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260205

Claims (7)

1. A self-learning fuzzy control method for hearth pressure of a step heating furnace is characterized by comprising the following steps: S1, selecting two pressure sampling points on two sides of two heating sections in a furnace, and performing differential pressure detection with the atmospheric pressure outside the furnace to obtain the actual value of the furnace pressure; s2, setting a hearth pressure target value to be a micro-positive pressure state which is enough to prevent a large amount of ambient cold air from being sucked into the hearth and inhibit the leakage of high-temperature gas in the hearth; S3, monitoring the furnace door action process in real time, recognizing that the furnace door is in an opening stroke, a closing stroke or reaches a closed position according to a furnace door stroke detection signal, and outputting a corresponding furnace door state identifier according to the furnace door stroke detection signal so as to determine whether self-learning feedforward control is started or a conventional closed loop control is switched to; s4, during the period that the furnace door enters the opening stroke, self-learning feedforward control is started, feedforward compensation quantity is generated according to the historical change trend of the furnace pressure, and the opening of the flue flashboard is adjusted in real time according to the feedforward compensation quantity, so that the pressure disturbance caused by the opening of the furnace door is restrained from being transmitted to the flue control loop; And S5, continuously monitoring the pressure fluctuation of the furnace chamber until the furnace chamber recovers to a steady state after the furnace door reaches the closed position, exiting the self-learning feedforward control and switching to the conventional closed-loop control, and continuously adjusting the opening of the flue flashboard to keep the pressure of the furnace chamber in a target functional interval.
2. The self-learning fuzzy control method of furnace pressure of a step heating furnace according to claim 1, wherein step S1 comprises: S11, arranging pressure taking ports at symmetrical positions of two sides of two heating sections in the furnace respectively to form pressure detection points on the inner side of the furnace; S12, leading the two pressure taking ports to a high pressure side and a low pressure side of the same differential pressure transmitter through pressure guide pipes respectively, and equalizing the pressures of the two sides; s13, accessing the atmospheric pressure outside the furnace as a reference pressure to a reference end of a differential pressure transmitter, and obtaining differential pressure signals inside and outside a hearth in real time; s14, filtering and temperature drift compensation are carried out on the differential pressure signals, and a stable hearth pressure actual value is output.
3. The self-learning fuzzy control method of furnace pressure of a step heating furnace according to claim 1, wherein step S2 comprises: s21, acquiring the environmental temperature and the atmospheric pressure standard of a furnace area, and taking the standard as a reference boundary of external cold air backflow and high-temperature gas overflow; S22, determining a minimum inward overpressure threshold value required for preventing cold air from flowing backwards according to the air flow organization in the furnace and the sealing grade of the furnace door; s23, determining a maximum outward overpressure threshold value for inhibiting the escape of high-temperature gas according to the furnace temperature, the blank oxidation burning sensitivity and the heat resistance limit of peripheral equipment; S24, setting a pressure interval between the minimum inward overpressure threshold and the maximum outward overpressure threshold as a hearth pressure target functional interval, and taking the intermediate state of the interval as a micro-positive pressure control target.
4. The self-learning fuzzy control method of furnace pressure of a step heating furnace according to claim 1, wherein step S3 comprises: S31, continuously acquiring a furnace door travel detection signal through a travel detection element arranged on a furnace door driving mechanism or a door body; S32, filtering the stroke detection signal to remove mechanical vibration and electrical noise and obtain stable and effective stroke variable; S33, dynamically comparing the filtered stroke variable with a preset opening threshold interval, a preset closing threshold interval and a preset closing interval, judging that the furnace door is currently in an opening stroke, a preset closing stroke or reaches a closed position, and generating a corresponding furnace door state identifier; s34, outputting a first trigger signal for starting self-learning feedforward control when the state identification is changed from closed state to open state; s35, outputting a second trigger signal switched to conventional closed-loop control after the state mark is changed from a closed stroke to a closed state and maintained until the furnace pressure fluctuation is attenuated to a steady state; And S36, if the state identifier jumps between the opening stroke and the closing stroke, the former effective trigger signal is kept until the next effective interval is confirmed, and the error switching is prevented.
5. The self-learning fuzzy control method of furnace pressure of a step heating furnace according to claim 1, wherein step S4 comprises: S41, latching the current furnace pressure instantaneous value as a reference disturbance zero point at the moment when the furnace door state mark enters the opening stroke, and starting a self-learning operation period; S42, reading the average furnace pressure change recorded in the two previous furnace door opening processes, and respectively recording as And (3) with Forming a history disturbance sample pair; S43, will And (3) with Inputting an exponential weighting filter to obtain a weighted average disturbance estimation value; S44, calculating feedforward compensation quantity required by the current furnace door opening by taking the deviation of the weighted estimation value and the furnace pressure target function value as disturbance error, wherein the calculation formula is as follows: Wherein, the Representing the reduction of the opening of the furnace chamber pressure regulator; indicating the reduction in the last furnace bore pressure regulator opening; Representing the average value of the furnace pressure change when the furnace door is opened last time; Representing the average value of the furnace pressure change of the last time the furnace door is opened; Representing the set value of the filter; A hearth pressure set value representing a furnace pressure regulator; Indicating the gain set point; s45, calculating the feedforward compensation quantity required by the current furnace door opening The compensation quantity is kept constant in the duration of the opening stroke, and the integration is stopped to continue accumulating; s46, after the state of the furnace door exits from the opening stroke, maintaining the feedforward compensation amount required by the opening of the furnace door Constant and record the actual pressure deviation for updating gain in the next operation And filter coefficients And the successive approximation optimal compensation is realized.
6. The self-learning fuzzy control method of furnace pressure of a step heating furnace according to claim 1, wherein step S5 comprises: s51, immediately taking the current furnace pressure instantaneous value as a disturbance attenuation monitoring starting point after the furnace door state identification confirms that the closed position is reached; s52, continuously collecting a hearth pressure sequence, and calculating a sliding window variance or a sliding average slope of the hearth pressure sequence in real time to serve as a fluctuation energy index; S53, when the fluctuation energy index is reduced to be below a preset steady state threshold value and is continuously maintained, judging that the airflow in the furnace is restored to be steady state, and generating a steady state confirmation mark; s54, closing a self-learning feedforward control channel according to a steady-state confirmation mark, and storing a final feedforward compensation quantity; s55, synchronously starting a conventional closed-loop control channel, continuously outputting the flue flashboard opening correction amount by using a PID or model prediction algorithm by taking a furnace pressure target function interval as a set value and real-time pressure as a feedback value; S56, continuously monitoring feedback deviation and an actuator saturation state in a closed-loop operation stage, dynamically setting closed-loop gain and integration time, and ensuring that the hearth pressure is always maintained in a target functional interval; And S57, if the furnace door is detected to enter the opening stroke again, immediately interrupting the conventional closed-loop control, returning to the step S41 to restart the self-learning feedforward control, and realizing the dual-mode seamless switching.
7. The self-learning fuzzy control method of the furnace pressure of the step-type heating furnace according to claim 1, wherein the adjustment of the opening degree of the flue shutter is realized by a butterfly-shaped flue valve arranged on a horizontal flue between the air heat exchanger and the vertical chimney, and the furnace pressure is in a proportional relation with the opening degree of a flue baffle.

Description

Self-learning fuzzy control method for hearth pressure of step heating furnace Technical Field The invention relates to the technical field of detection and control of hearth pressure of a heating furnace, in particular to a self-learning fuzzy control method of hearth pressure of a stepping heating furnace. Background The detection and control of the furnace pressure of the heating furnace are key links for ensuring the efficient and stable operation of the heating furnace. The furnace pressure is usually detected by differential pressure, i.e. pressure tapping points are selected on both sides of the heating section in the furnace and compared with the atmospheric pressure outside the furnace, so as to obtain the actual value of the furnace pressure. The control range of the furnace pressure is generally-100 Pa to +100Pa, but the actual production is usually set in a micro-positive pressure state of +1Pa to +5Pa so as to ensure the stability of the combustion environment in the furnace. The adjustment of furnace pressure is realized mainly through adjusting the aperture of flue flashboard, and the flue flashboard is installed in the horizontal flue between air heat exchanger and chimney, adjusts flue extraction volume through changing its opening degree, and then influences furnace pressure. Reasonable control of the hearth pressure can effectively avoid flame extension or cold air suction, so that energy waste is reduced, heating quality is improved, and oxidation burning loss is reduced. At present, the control of the furnace pressure mainly depends on manual adjustment or a simple proportional control method, namely, the pressure adjustment is realized by adjusting the fixed proportional relation between the opening degree of a flue flashboard and the furnace pressure. However, the method has obvious limitations that firstly, the operation working condition of the heating furnace is complex and changeable, the furnace pressure is influenced by various factors such as fuel flow, combustion efficiency, flue resistance and the like, the fixed proportion control is difficult to adapt to the dynamically-changing environment, so that the adjustment precision is insufficient, secondly, the manual adjustment is dependent on the experience of operators, hysteresis and instability are easy to occur, the accurate micro positive pressure control is difficult to realize, and in addition, the traditional control method lacks self-adaptability, the control parameters cannot be dynamically adjusted according to the real-time change of the furnace pressure, the fluctuation of the furnace pressure is easy to be overlarge, and the heating quality and the equipment safety are influenced. Therefore, a hearth pressure control method capable of adapting to complex working conditions and improving control accuracy and stability is needed. Disclosure of Invention According to the technical problems, the self-learning fuzzy control method for the hearth pressure of the step-type heating furnace is provided. According to the invention, by adopting a dual-mode switching control technical means driven by a furnace door travel signal, the technical effects of self-learning feedforward compensation at the moment of furnace door opening, seamless transfer to conventional closed-loop control after furnace door sealing and air flow recovery are achieved, the furnace pressure is always kept in a micro-positive pressure functional interval which is enough to prevent a large amount of ambient cold air from flowing backward and inhibit high-temperature air from leaking, so that the impact of furnace door opening and closing actions on a gas field in the furnace is obviously weakened, furnace temperature drop and blank oxidation burning caused by cold air invasion are reduced, the energy waste and the thermal damage risk of surrounding equipment caused by flame outward spraying are reduced, the feedforward compensation quantity is continuously approximated to an optimal value along with the increase of operation times by means of a successive self-learning mechanism, the long-acting optimization of gradual reduction of the action amplitude of a flue flashboard and stable lifting of system regulation precision is realized, and finally, under the condition that the condition of not depending on specific numerical limitation, the furnace pressure control quality with stable, energy conservation and low loss can be ensured under various working conditions in a generalized and self-adaptive manner. The invention adopts the following technical means: a self-learning fuzzy control method for hearth pressure of a step heating furnace comprises the following steps: S1, selecting two pressure sampling points on two sides of two heating sections in a furnace, and performing differential pressure detection with the atmospheric pressure outside the furnace to obtain the actual value of the furnace pressure; s2, setting a hearth pressure target value to