Abstract This paper introduces the research and application of an automated system for a walking beam reheating furnace in a steel rolling mill. Through a series of strategies including software programming, dynamic compensation, encoder length measurement, and frequency converter optimization, existing problems were improved, resulting in a significant increase in the furnace's efficiency and automation level. Keywords Walking beam reheating furnace, program, dynamic compensation, encoder 1 Introduction In 2004, Jinan Iron and Steel Group's No. 1 small-scale rolling mill underwent a full continuous rolling mill renovation. To meet the requirements of an annual production of 800,000 to 1,000,000 tons of bar stock, it was equipped with a walking beam type air/gas dual regenerative, side-inlet/side-outlet reheating furnace with a heating capacity of 130 t/h (170 t/h for hot charging). Therefore, how to improve the tapping speed of the walking beam reheating furnace and the stability of the tapping system has always been a topic of great concern. 2. Problem Statement The entire electrical control system of the heating furnace area is implemented by a single S7-414 CPU. The original steel tapping cycle of the heating furnace was approximately 40 seconds, with an hourly output of about 100 billets. Based on a single billet weight of 1.7 tons, the hourly output was 170 tons. This is significantly lower than the heating capacity of the furnace during hot charging, a problem that became even more pronounced during the initial trial production when the single billet weight was only 0.8 tons. Furthermore, the walking beam's operation was highly unstable, with frequent malfunctions, further restricting the release of the heating furnace's heating capacity. The main reasons for this are analyzed as follows: 1. The logical interlocking relationships of the control system are redundant. For example, the furnace door must activate once during each tapping process; steel can only be fed after the walking beam returns to its lower position; the walking beam can only be activated when the positioning pusher returns to its rear position, etc. These unnecessary logical interlocking relationships consume a significant amount of time, greatly increasing the steel tapping cycle of the heating furnace. 2. The walking beam is driven by a hydraulic cylinder. During operation, because the walking beam's motion curve does not consider the inertia of the entire walking mechanism, the walking beam often lurches forward, causing uneven billet arrangement in the furnace. This necessitates manual adjustment of the step distance during tapping, affecting the furnace's tapping rhythm. 3. The original walking beam's motion logic interlock program has numerous logical dead zones. In practical applications, the program frequently enters an infinite loop, requiring the PLC to be powered on again to resume operation, thus affecting the furnace's tapping rhythm. 3. Control Strategy and Improvement Scheme 1. The main reason for the low tapping speed is the overly complex logic interlock of the tapping action and the instability of the furnace's electrical control system. While ensuring the safe operation of all furnace equipment, the logic interlock of the actions will be optimized to minimize the tapping cycle. This mainly involves the following aspects: ① Eliminating the original furnace-front positioning baffle can save approximately 2 seconds. Utilizing the braking of the furnace-front variable frequency roller conveyor ensures the billet stops accurately in front of the furnace. ② The furnace doors, which were previously opened and closed once for each billet, are now kept open continuously. Simultaneously, the variable frequency roller conveyor's braking mechanism prevents the billets from colliding with the walking beam. ③ The feeding command is issued earlier; instead of starting feeding when the walking beam returns to its lower rear position, feeding is now initiated when the walking beam reaches its upper front position. ④ The early start of the walking beam is changed from starting when the positioning pusher returns to its rear position to starting feeding when the positioning pusher reaches its front position. The improved electrical control system's logic interlocking relationship is shown in Figure 1. 2. Add a step distance compensation program. When the walking beam rises past the steel support point, it automatically memorizes the current translation encoder value and then calculates the value of the previous translation encoder based on the step distance. During the descent of the walking beam after reaching the upper front position, an adaptive digital PID algorithm is used to compensate and correct the previous position of the walking beam, ensuring that the forward step distance is the given step distance when the walking beam descends to the steel release point. ① Main program design for compensation (see Figure 2): In the main program design, the step distance compensation only works for a short period after the stepping beam reaches the upper front position and descends to the steel placement point. Furthermore, a dead zone (5mm) is designed to prevent stepping beam vibration. ② The flowchart of the digital PID algorithm for compensation is shown in Figure 3: This program is called in the PLC's cyclic timer interrupt OB32 (cycle is 200ms). The program remembers the position deviation within three scanning cycles and calculates it according to the formula: Un=Kp[E(n)×K1+E(N-1)×K2+E(n-2)×K3], where K1, K2, and K3 are pre-set coefficients, set to 0.8, 0.1, and 0.05 respectively based on the large inertia of the entire system. ③ Method for determining the adaptive coefficient: A compensation stop condition is set in the main program; when the step distance deviation is less than 5mm, step distance compensation stops. Simultaneously, the proportional coefficient Kp is corrected using the following formula: Kp+1 = (Xp - Xm) × 0.15 + Kp Where: Kp+1 is the digital PID gain for the next calculation, Kp is the digital PID gain for this calculation, Xp is the theoretically calculated previous encoder value, and Xm is the actual encoder value after compensation. 3. The original stepper beam program is optimized to ensure there are no logical dead zones. This is mainly achieved through the following aspects: ① A manual zero-return program is added, allowing the stepper beam program to manually return to the original point after entering a logical dead zone, while simultaneously re-initializing all program states. ② After the stepper beam decelerates to zero along the curve but has not yet reached the predetermined position, the program automatically applies a new speed to propel the stepper beam. ③ If, after applying speed, the predetermined position is still not reached within a certain time, the program judges whether the deviation between the actual position and the predetermined position is less than 10mm, and automatically proceeds to the next action to continue execution. 4. Addressing the issue of numerous detection elements and lack of fault tolerance in the original furnace front section, the control program was optimized to combine two laser object detectors into one. Fault tolerance was also implemented in the program to ensure operational reliability. This completely eliminated the previous issue of billet positioning failure within the furnace. 4. Conclusion and Usage Results After more than seven months of operation following this improvement to the furnace electrical control system, the equipment has performed well. The steel tapping cycle has been reduced from approximately 40 seconds to 29 seconds, demonstrating a significant speed improvement. Furthermore, zero failures have been achieved, meeting the needs of high-paced production tasks, and all indicators have met design requirements.