An automatic constant temperature control system for continuous induction heating of tool steel long bars
By collecting and analyzing coil electrical parameters in real time, and combining the dual criteria of resistance change rate and power factor, accurate judgment of Curie point crossing and transient power compensation were achieved. This solved the problems of temperature inhomogeneity and false triggering during the heating process of tool steel long bars, and improved heating efficiency and temperature uniformity.
Patent Information
- Application Number
- CN202611199674.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-10
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-08-10
AI Technical Summary
The existing continuous induction heating control system for tool steel long bars cannot respond in time to sudden changes in Curie point heating efficiency, resulting in excessive axial temperature difference in the bar and is prone to false triggering and temperature inhomogeneity, which cannot meet the production requirements of precision tools.
By employing techniques such as synchronous sampling of coil electrical parameters, sliding window calculation of load characteristics, Curie point crossing event discrimination, and transient power pulse superposition, coil voltage and current signals are acquired in real time. Combined with dual criteria of resistance change rate and power factor, accurate judgment of Curie point crossing and transient power compensation are achieved, avoiding false triggering and temperature fluctuations.
It effectively reduces the axial temperature difference of the bar stock, improves the uniformity of heating temperature, adapts to production conditions under different specifications and feeding speeds, avoids miscompensation and current surges, and ensures the stability and continuity of the temperature control process.
Smart Images

Figure CN122733017B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of induction heating control technology, and more specifically, to an automatic constant temperature control system for continuous induction heating of tool steel long bars. Background Technology
[0002] Continuous induction heating of tool steel long bars is a core process in tool steel heat treatment production. Multiple medium-frequency induction coils heat the continuously fed bars to the target quenching or tempering temperature. Existing production lines generally use an outlet infrared temperature measurement combined with PID closed-loop regulation of each coil's power, supplemented by fixed-end power compensation and speed feedforward to achieve constant temperature control. Tool steel contains a large amount of alloying elements such as chromium, molybdenum, vanadium, and tungsten, and its thermal conductivity is only 60% to 70% of that of ordinary carbon steel. When heated to the Curie point of 760 to 780 degrees Celsius, the material's magnetic permeability drops sharply from several hundred to 1, and the heating efficiency drops instantaneously by 30% to 50%. Existing control methods are based on the outlet temperature. The feedback has a pure lag of 1 to 3 seconds, which cannot respond in time to the sudden change in heating efficiency during Curie point crossing. It is easy to form a temperature depression in the axial direction of the bar stock. The fixed power compensation method cannot adapt to the Curie point crossing process under different diameters and feed speeds, which is prone to undercompensation or overcompensation. Feeding jitter and electromagnetic interference can also easily lead to false triggering. Power step changes can easily trigger the power supply overcurrent protection. Ultimately, the axial temperature difference of the tool steel bar stock can reach more than 20 degrees Celsius. The hardness uniformity after heat treatment cannot meet the production requirements of precision tools. In order to solve this technical problem, we provide a continuous induction heating automatic constant temperature control system for long tool steel bars. Summary of the Invention
[0003] The purpose of this invention is to provide an automatic constant temperature control system for continuous induction heating of tool steel long bars, so as to solve the problems mentioned in the background art.
[0004] To achieve the above objectives, an automatic constant temperature control system for continuous induction heating of tool steel long bars is provided. The system includes a coil electrical parameter synchronous sampling unit that collects the terminal voltage signal and loop current signal of each induction heating coil through voltage transformers and current transformers, and combines the voltage and current zero-crossing detection circuit to perform phase synchronous triggering sampling of AC cycles to obtain discrete voltage and current signals. The load characteristic sliding window calculation unit calculates the true effective value and voltage-current phase difference of the discrete voltage and current signals within the sliding time window, and outputs the load equivalent resistance value and power factor value of each coil in real time. The Curie point crossing event discrimination unit performs a first-order backward differential calculation on the equivalent resistance value of the load to obtain the resistance change rate. When the resistance change rate reaches the preset decrease threshold of the resistance change rate and the power factor value reaches the preset decrease threshold of the power factor in sync, the Curie point crossing trigger signal is output through dual criteria and logic triggering interlock. At the same time, the built-in anti-shake delay circuit filters out false triggers caused by feeding jitter and load fluctuation. After receiving the Curie point crossing trigger signal, the transient power pulse superposition unit superimposes a transient heating power feedforward pulse with a fixed amplitude and fixed width onto the PID closed-loop output of the corresponding coil. After the transient heating power feedforward pulse output reaches the preset duration, the automatic recovery unit automatically removes the superposition of the transient heating power feedforward pulse, so that the corresponding coil returns to the original PID closed-loop temperature control state.
[0005] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention solves the problem of traditional infrared temperature measurement feedback at the outlet, which suffers from control lag and cannot respond promptly to sudden changes in Curie point heating efficiency, by real-time acquisition of coil terminal voltage and loop current and online calculation of load equivalent resistance and power factor. It enables immediate triggering of control actions when Curie point crossing occurs, effectively reducing the axial temperature difference of the bar stock and improving the heating temperature uniformity of tool steel long bars. By combining resistance change rate and power factor change as dual criteria with anti-jitter delay and post-trigger hardware interlocking, it addresses the issue of false triggering caused by feeding jitter and electromagnetic interference when using a single judgment condition, thus improving the reliability of Curie point crossing event judgment and avoiding false compensation. The resulting temperature anomalies are addressed by employing a technique that adaptively adjusts the pulse amplitude and compensation time according to the bar specifications and feeding speed, combined with power gradual-in / gradual-out control. This solves the problem that fixed compensation parameters cannot adapt to different production conditions. It can match the Curie point crossing process of bars of different specifications at different feeding speeds, avoiding insufficient or over-compensation. At the same time, it reduces the current impact of power step changes on the heating power supply. Through the PID integral tracking seamless switching technique, it solves the problem of power switching jump after transient compensation, achieving a smooth transition between the compensation process and normal PID closed-loop control, avoiding secondary temperature fluctuations during the switching process, and ensuring continuous and stable temperature control. Attached Figure Description
[0006] Figure 1 This is a system structure block diagram of the present invention. Detailed Implementation
[0007] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0008] This invention provides an automatic constant temperature control system for continuous induction heating of tool steel long bars. Please refer to [link to relevant documentation]. Figure 1 As shown, the coil electrical parameter synchronous sampling unit collects the terminal voltage signal and loop current signal of each induction heating coil through voltage transformers and current transformers, and combines the voltage and current zero-crossing detection circuit to perform phase synchronous triggering sampling of AC cycles to obtain discrete voltage and current signals. The load characteristic sliding window calculation unit calculates the true effective value and voltage-current phase difference of the discrete voltage and current signals within the sliding time window, and outputs the load equivalent resistance value and power factor value of each coil in real time. The Curie point crossing event discrimination unit performs a first-order backward differential calculation on the equivalent resistance value of the load to obtain the resistance change rate. When the resistance change rate reaches the preset decrease threshold of the resistance change rate and the power factor value reaches the preset decrease threshold of the power factor in sync, the Curie point crossing trigger signal is output through dual criteria and logic triggering interlock. At the same time, the built-in anti-shake delay circuit filters out false triggers caused by feeding jitter and load fluctuation. After receiving the Curie point crossing trigger signal, the transient power pulse superposition unit superimposes a transient heating power feedforward pulse with a fixed amplitude and fixed width onto the PID closed-loop output of the corresponding coil. After the transient heating power feedforward pulse output reaches the preset duration, the automatic recovery unit automatically removes the superposition of the transient heating power feedforward pulse, so that the corresponding coil returns to the original PID closed-loop temperature control state.
[0009] The terminal voltage signal output from the voltage transformer and the loop current signal output from the current transformer are first sent to the signal pre-conditioning circuit. The signal pre-conditioning circuit adjusts the terminal voltage signal to the input range of the high-speed analog-to-digital converter using a resistor divider, and converts the loop current signal into a voltage signal matching the range using a sampling resistor voltage drop converter. The signal pre-conditioning circuit has a built-in second-order Butterworth anti-aliasing low-pass filter. The cutoff frequency of the anti-aliasing low-pass filter is adjusted in real time according to the current induction heating operating frequency, and is set to 10 times the current operating frequency to ensure that the fundamental frequency and main harmonic components can pass normally while filtering out high-frequency sampling aliasing noise. After conditioning, the terminal voltage signal and loop current signal are split into two paths: one path is sent to the voltage and current zero-crossing detection circuit, and the other path is sent to the high-speed analog-to-digital converter. The analog input channel of the high-speed analog-to-digital converter incorporates two independent high-speed voltage comparators in its voltage and current zero-crossing detection circuit. The non-inverting input of each comparator receives the detection signal, while the inverting input connects to a zero-level reference voltage. When the signal changes from negative to positive and crosses zero, the comparator's output changes from low to high. The two comparators output positive zero-crossing pulses for the voltage and current respectively, serving as synchronization reference pulses. These pulses are then fed into the sampling trigger logic circuit. The circuit first performs Schmitt trigger shaping on the two synchronization reference pulses to eliminate noise glitches near the zero-crossing point. The shaped pulses are then fed into a digital frequency divider. The digital frequency divider calculates the division factor based on the real-time measured current induction heating operating frequency. Satisfy the formula , in the formula The global clock reference frequency for the sampling trigger logic circuit, in Hertz. This is the current induction heating operating frequency, measured in Hertz. For each AC cycle, a fixed integer number of sampling points are required. The frequency divider divides the global clock reference frequency according to the calculated division coefficient, generating an equally spaced sampling trigger sequence that is strictly synchronized with the AC cycle. This equally spaced sampling trigger sequence is simultaneously connected to the conversion start pins of both the voltage-channel high-speed analog-to-digital converter (ADC) and the current-channel high-speed ADC, ensuring that the ADCs of the two channels start at exactly the same time, with a time difference between the sampling times of the two channels. This will introduce a fixed phase error. The phase error satisfies the formula , in the formula The time difference between the start of sampling for the two channels, in seconds. Starting the conversion at the same time can... Controlled within 1 nanosecond, with a phase error of less than 0.0000126 radians at a 2000 Hz operating frequency, the sampling phase difference between two signals can be eliminated, avoiding fixed deviations in subsequent phase difference calculations. Taking a 2000 Hz operating frequency as an example, the global clock reference frequency is 25.6 MHz, and 32 points are sampled per AC cycle. Substituting into the formula, the frequency division coefficient is calculated to be 400. The digital frequency divider outputs a sampling trigger pulse every 400 global clock cycles, resulting in exactly 32 equally spaced trigger pulses per AC cycle. The time interval between adjacent trigger pulses is 15.625 microseconds. The high-speed analog-to-digital converter simultaneously performs analog-to-digital conversion on the terminal voltage signal and the loop current signal upon the arrival of each sampling trigger pulse. After conversion, it outputs quantized discrete voltage and current values. The unit of the discrete voltage value is volts, and the unit of the discrete current value is... The discrete voltage and current values obtained from each conversion are combined into discrete voltage and current signals according to the sampling order and sent to the subsequent load characteristic sliding window calculation unit. The fixed number of sampling points for an integer number of cycles can ensure that the probability of spectral leakage caused by non-integer cycle truncation is reduced when calculating the true RMS value. The true RMS value calculation result can be obtained without adding an additional window function. At the same time, the fixed sampling point interval can directly calculate the voltage and current phase difference through the difference of the sampling point sequence number corresponding to the zero crossing point, without the need for an additional high-precision timestamp measurement circuit. This adaptive synchronous sampling method can automatically adjust the frequency division coefficient when the induction heating operating frequency drifts slightly with the resonance state, always maintaining a fixed integer number of sampling points for each AC cycle, avoiding the sampling point misalignment error that occurs when the fixed sampling rate changes with frequency, and providing a discrete data basis for the subsequent calculation of the load equivalent resistance value and power factor value.
[0010] After the discrete voltage and current values output from the preamplifier circuit are fed into the load characteristic sliding window calculation unit, DC bias compensation is first performed. The voltage discrete value is subtracted from the DC bias measured during power-on initialization, and the current discrete value is subtracted from the DC bias measured during power-on initialization. The DC bias values are obtained by taking the arithmetic mean of 1024 consecutive samples under zero-input conditions during power-on initialization, with units of volts and amperes, respectively. The compensated voltage and current discrete values are then subjected to range judgment. If the compensated voltage discrete value exceeds 5% or 5% of the full-scale voltage range, or if the compensated current discrete value exceeds 5% of the full-scale voltage range, the range is determined. When the value exceeds 5% or 5% of the full-scale current range, it is determined to be electromagnetic spike interference. The sampled value is not written to the ring buffer in this calculation; instead, the previously calculated true RMS voltage and current values are used as the output. When the compensated discrete voltage and current values are within the normal range, the earliest voltage sampled value stored at the head of the voltage ring buffer and the earliest current sampled value stored at the head of the current ring buffer are read, respectively. The voltage and current ring buffers are independent first-in-first-out circular storage structures. The depth of the voltage and current ring buffers is set to the total number of sampling points corresponding to an integer number of AC cycles. Total number of sampling points Satisfy the formula , in the formula This represents the number of AC cycles contained within the sliding time window, and its value is an integer greater than or equal to 2. The fixed number of sampling points per AC cycle, compared with the number used by the preceding sampling trigger logic circuit. To maintain consistency, using units of samples, update the voltage sum of squares recursive variables and current sum of squares recursive variables after reading the earliest sampled value. The update of the voltage sum of squares recursive variable satisfies the formula. , in the formula The sum of squared voltages is the recursive variable, with units of volts squared. The earliest voltage sample value, in volts, is stored at the head of the voltage ring buffer queue. The latest discrete voltage value after compensation is in volts. The current square sum and recursive variable update satisfy the formula. , in the formula The sum of squared currents is the recursive variable, with the unit being ampere squared. The earliest current sample value, in amperes, is stored at the head of the current ring buffer queue. The latest current discrete value after compensation, in amperes, is updated. After the sum of squares and recursive updates are completed, the latest voltage discrete value after compensation is written to the tail position of the voltage ring buffer, and the latest current discrete value after compensation is written to the tail position of the current ring buffer. The head and tail pointers of the voltage ring buffer are adjusted according to the total number of sampling points. The modulo operation increments, and the head and tail pointers of the current ring buffer are incremented according to the total number of sampling points. The modulus increments automatically, maintaining the number of sampling points within the sliding time window at a constant value. Then, based on the updated sum of squares and recursive variables, the true RMS values of voltage and current are calculated. The calculation of the true RMS value of voltage satisfies the formula... , in the formula The voltage is the true RMS value, in volts. The true RMS value of the current is calculated according to the formula. , in the formula This is the true RMS value of the current, in amperes. During the initialization phase, data is continuously collected. The sampled values fill the voltage and current circular buffers, and the initial voltage and current sum-of-squares recursive variables are calculated simultaneously. After filling, the normal recursive calculation state begins. Taking a sliding time window containing 32 sampling points per AC cycle and 3 AC cycles as an example, the total number of sampling points is... The system uses 96 samples. During the initialization phase, 96 samples are collected to calculate the initial sum of squares. Subsequently, each new sample only requires two subtraction operations, two addition operations, and two square root operations to update the true RMS value. It does not need to traverse all 96 samples within the window, reducing the single-step calculation workload to 1 / 96 of the traditional block calculation method. This makes it suitable for real-time operation in embedded microcontrollers. The calculated true RMS values of voltage and current are output in real time and fed into the subsequent circuit to calculate the equivalent resistance of the load. The recursive calculation method can update the true RMS value once after each sampling point. The output delay is one sampling control cycle, which reduces the output delay by one window length compared to the traditional window block calculation method. DC bias compensation can eliminate zero-point offset introduced by voltage transformers, current transformers, and signal pre-conditioning circuits. Outlier judgment can suppress the influence of sampling spikes caused by electromagnetic interference or feeding jitter on the true RMS calculation. The window length of integer cycles can avoid spectral leakage errors caused by non-integer cycle truncation. The true RMS calculation result can be obtained without applying an additional window function.
[0011] After the true RMS values of voltage and current obtained from the pre-stage calculation are input, the positive zero-crossing pulses of the terminal voltage and loop current output by the voltage and current zero-crossing detection circuit are sent to the capture port of the sampling trigger logic circuit. Each time the rising edge of a zero-crossing pulse is captured, the sampling point number corresponding to the current moment is recorded. The sampling point number corresponding to the positive zero-crossing moment of the terminal voltage signal is denoted as... The sampling point number corresponding to the positive zero-crossing moment of the loop current signal is denoted as . After the positive zero-crossing event of the internal terminal voltage of the same AC cycle occurs, wait for the positive zero-crossing event of the loop current to occur, and record the difference in the number of sampling points between the two events. Difference in the number of sampling points Satisfy the formula , in the formula This refers to the number of sampling points between the zero-crossing of the terminal voltage and the zero-crossing of the current within the same cycle, expressed in units of points. The induction heating coil load is always an inductive load, and the loop current phase lags behind the terminal voltage phase. The value is always a positive integer, and the number of sampling points between two adjacent positive zero-crossing pulses of the terminal voltage is recorded as the total number of sampling points for the current cycle. Current total number of sampling points for the cycle The unit is the number of sampling points. The difference in the number of sampling points for the three most recent AC cycles is continuously recorded and stored in a zero-crossing difference queue of length 3. Each time a new cycle zero-crossing event occurs, the queue is updated by removing the earliest difference in the number of sampling points from the queue and adding the latest difference in the number of sampling points. The arithmetic mean of the differences in the number of sampling points in the queue is taken to obtain the average zero-crossing sampling point difference. Average zero-crossing sampling point difference The unit is points, and the voltage and current phase difference is calculated based on the average zero-crossing sampling point difference and the total number of sampling points in the current cycle. Voltage and current phase difference Satisfy the formula , in the formula The phase difference between voltage and current is expressed in radians. After calculating the phase difference, a range check is performed. The phase difference is considered valid when it is less than 0 or greater than 1. If the abnormal calculation result is determined to be caused by zero-crossing interference, the phase difference value will not be updated this time; the effective phase difference value obtained from the previous cycle will be used instead. When the phase difference is between 0 and... When the phase difference is within the specified interval, it is considered an effective phase difference. The real-time power factor is calculated based on this effective phase difference. The power factor is obtained by looking up a pre-stored cosine lookup table, which is generated during program initialization. This table stores the cosine values corresponding to all angles from 0 to π / 2 in 0.001 radians. During the lookup, the power factor PF is obtained by directly indexing the corresponding address based on the phase difference φ. The power factor PF is dimensionless. Subsequently, the equivalent impedance magnitude of the coil load is calculated, and this magnitude satisfies the formula... , in the formula This represents the equivalent impedance magnitude of the coil load, in ohms. This is the true RMS voltage value calculated by the preceding stage, in volts. The current is the true RMS value obtained from the previous stage calculation, in amperes. Finally, the equivalent resistance of the load is calculated, and the equivalent resistance of the load satisfies the formula. , in the formula The value is the equivalent resistance of the load, in ohms. Taking a 2000 Hz operating frequency with 32 sampling points per AC cycle as an example, the number of sampling points N_T between two adjacent zero-crossing points of the terminal voltage is 32. The differences in the number of sampling points for three consecutive cycles are 4, 4, and 3, respectively. The average zero-crossing sampling point difference is calculated. Approximately 3.6667, substituting into the formula yields the phase difference. With an angle of approximately 0.72 radians, the power factor (PF) obtained from the table is approximately 0.75. If the true effective value of the voltage is 500 volts and the true effective value of the current is 100 amperes, the equivalent impedance modulus of the coil load is calculated to be 5 ohms and the equivalent resistance of the load is 3.75 ohms. The calculated power factor and equivalent resistance values are sent to the Curie point crossing event discrimination unit in real time to provide input data for subsequent resistance change rate calculation and dual-criteria judgment. The zero-crossing time difference is calculated using the synchronous sampling point sequence number without the need for additional high-precision time measurement hardware. The zero-crossing point difference moving average can reduce the timing jitter error caused by noise affecting the zero-crossing comparator to one-third of that of a single measurement. The phase difference range verification can eliminate erroneous calculation results caused by external electromagnetic interference. The cosine lookup table method can avoid real-time floating-point trigonometric function calculation, reducing the power factor calculation time from microseconds to nanoseconds, meeting the time requirements for simultaneous real-time calculation of multiple coils.
[0012] The fixed control cycle is generated by a hardware timer. Each control cycle triggers a calculation of the resistance change rate. First, the latest calculated equivalent load resistance value for the current control cycle is read and denoted as... Simultaneously, it reads the load equivalent resistance value of the previous adjacent control cycle stored in the register, denoted as... k is the control cycle number, a positive integer starting from 0 and incrementing. The resistance difference between two adjacent control cycles is calculated. resistance difference Satisfy the formula , in the formula The resistance difference between adjacent control cycles, in ohms, is first determined by sign. If the resistance difference is greater than 0, it indicates the equivalent load resistance is in the rising phase, corresponding to the normal heating process of the bar stock. This does not meet the characteristic of a rapid decrease in resistance during Curie point crossing, so the resistance change rate is directly set to 0. If the resistance difference is less than 0, it indicates the equivalent load resistance is in the falling phase, entering the possible Curie point crossing range. The resistance change rate is then calculated, normalized based on the control cycle duration, denoted as . The unit is seconds. First, calculate the absolute resistance change rate per unit time, then divide it by the equivalent load resistance value of the current control cycle to obtain the relative resistance change rate. Satisfy the formula , in the formula This is the normalized rate of change of resistance per unit time, expressed in seconds. Normalization converts the absolute resistance change into a relative proportion to the current resistance value, eliminating differences in absolute resistance caused by different bar diameters and heating powers. The calculated relative rate of change of resistance is then filtered using a first-order hysteresis filter to obtain the output value of the resistance change rate. Satisfy the formula , in the formula The rate of change of resistance after filtering, in seconds. These are the first-order lag filter coefficients, with values ranging from 0 to 1. The current filtered resistance change rate, calculated in the previous control cycle, is saved to the register as the old value for the next cycle after the calculation is completed. Simultaneously, the current equivalent load resistance value is also saved. The overwrite value is saved in the register as the old resistance value for the next cycle. Taking a control cycle of 10 milliseconds as an example, the equivalent load resistance in the previous control cycle was 4 ohms, and the equivalent load resistance in the current control cycle is 3.94 ohms. The resistance difference is calculated. The resistance is -0.06 ohms, with a negative sign, and is used in the rate of change calculation. Substituting this into the formula, the relative resistance change rate is approximately -1.52 ohms per second, meaning the relative decrease in resistance per unit time is 152% per second. After a first-order hysteresis filter with a filtering coefficient of 0.3, the resistance change rate output value is obtained. The calculated filtered resistance change rate is then sent to the subsequent dual-criteria and logic triggering interlocking unit for comparison with the preset drop threshold of the resistance change rate. The sign judgment can directly filter out invalid falling edges caused by resistance fluctuations during normal heating. The normalized relative change rate calculation can adapt to heating scenarios of tool steel bars of different diameters and materials without needing to recalibrate the threshold for different production specifications. The first-order hysteresis filter can eliminate sampling noise and resistance change rate spikes caused by power supply fluctuations, avoiding false triggering.
[0013] The preset threshold for the rate of resistance change and the preset threshold for the power factor change were obtained during the commissioning phase through trial heating calibration with tool steel of the same specification. During calibration, three long bars of tool steel of the same batch and specification were continuously tested and heated. When a sudden change in the Curie point magnetic properties occurred during the heating process of each bar, the normalized rate of resistance change at that time was recorded. At the same time, the average power factor value within 500 milliseconds before the sudden change and the real-time power factor value at the time of the sudden change were recorded, and the power factor decrease was calculated. After the calibration of the three bars was completed, the arithmetic mean of the three recorded rates of resistance change was used to obtain the preset threshold for the rate of resistance change. The preset power factor decrease threshold is obtained by taking the arithmetic mean of the three recorded power factor decreases. The resistance change rate has a preset threshold decrease, measured in seconds. The power factor has a preset, dimensionless threshold decrease. Both thresholds, obtained from calibration, are permanently stored in a non-volatile parameter memory. During power-on, the device automatically reads these two thresholds from the non-volatile parameter memory and stores them in the corresponding comparator's threshold register. The dual-criteria and logic-triggered latch includes a digital comparator for the resistance change rate and a digital comparator for the power factor, along with hardware AND gates and RS flip-flops. The input of the digital comparator for the resistance change rate is connected to the filtered resistance change rate calculated in the previous stage. The digital comparator for the resistance change rate is set with a 10% hysteresis. retracement Satisfy the formula , in the formula The hysteresis of the resistance change rate comparator is expressed in seconds. When the filtered resistance change rate is less than or equal to a preset falling threshold, the digital comparator outputs a high level; when the filtered resistance change rate is greater than the preset falling threshold plus the hysteresis, the comparator outputs a high level. At this time, the digital comparator for the rate of change of resistance outputs a low level, and the input of the digital comparator for the power factor is connected to the power factor value calculated in real time. The digital comparator for the power factor first continuously calculates the average power factor value over the most recent 500 milliseconds. The power factor decrease is obtained by calculating the difference between the real-time power factor and the average power factor. Power factor decrease Satisfy the formula , in the formula The power factor decrease is dimensionless. For real-time power factor values, the power factor digital comparator is set with a 10% hysteresis. retracement Satisfy the formula , in the formula The power factor comparator hysteresis is dimensionless. When the power factor drop is greater than or equal to a preset power factor drop threshold, the power factor digital comparator outputs a high level; when the power factor drop is less than the preset power factor drop threshold minus the hysteresis, the power factor comparator outputs a high level. When the power factor digital comparator outputs a low level, the outputs of the resistance change rate digital comparator and the power factor digital comparator are connected to the two inputs of a hardware AND gate circuit. The output of the hardware AND gate circuit is connected to the set input of the RS flip-flop, which adopts a set-priority structure. The reset input is connected to the reset output of the subsequent automatic recovery unit of the control state. When both comparators output a high level simultaneously, the hardware AND gate circuit outputs a high level to set the RS flip-flop. The RS flip-flop outputs a Curie point crossing trigger signal, and simultaneously outputs an interlock level signal connected to the enable inputs of the two comparators, forcibly pulling down the outputs of the two comparators and locking the inputs until the reset input receives a reset signal to release the lockout. Taking the calibration of Cr12MoV tool steel bars with a diameter of 50 mm as an example, the resistance change rates recorded for the Curie point crossing of the three test bars are -1.4 sec, -1.5 sec, and -1.6 sec, respectively. The preset drop threshold of the resistance change rate is calculated. The recorded power factor decreases were 0.12, 0.11, and 0.10, with a rate of -1.5 per second. The preset power factor decrease threshold was then calculated. With a power factor of 0.11, during operation, when the rate of change of the filtered resistance is less than or equal to -1.5 seconds per second and the power factor decrease is greater than or equal to 0.11, both comparators simultaneously output a high level to trigger the RS flip-flop to set, outputting a Curie point crossing trigger signal. The Curie point crossing trigger signal is sent to the subsequent transient power pulse superposition unit to start the transient heating power feedforward pulse output. The pure hardware logic implementation of the judgment link response delay is less than 1 microsecond, requiring no software intervention in the judgment, and will not miss the transient process of Curie point crossing. The automatic locking of the comparator after triggering can avoid repeated triggering caused by load parameter fluctuations during Curie point crossing. The comparator hysteresis setting can prevent electrical noise near the threshold from causing output level jitter. Using the power factor decrease as the judgment basis instead of the absolute value of the power factor can adapt to the steady-state power factor differences under different coil turns and different bar diameters. There is no need to recalibrate the threshold for different operating conditions. The method of taking the average of three calibrations can reduce the threshold calibration error caused by the material difference of a single bar.
[0014] The transient power pulse superposition unit includes a PID output register, a pulse amplitude register, an output adder, an output limiting circuit, and a gradual-in / gradual-out ramp generator. The PID output register is updated once per control cycle, storing the current power setpoint calculated by the infrared temperature measurement closed-loop PID temperature control unit. The unit is a percentage of rated power, ranging from 0 to 100%, corresponding to the induction heating power supply from 0 to rated output power. The pulse amplitude register stores the basic amplitude of the transient heating power feedforward pulse. The basic amplitude of the transient heating power feedforward pulse is stored in a non-volatile parameter memory along with the threshold parameter during calibration. During operation, the pulse amplitude is corrected according to the current bar diameter and real-time feed speed. The corrected transient heating power feedforward pulse amplitude is... Satisfy the formula , in the formula This is the corrected transient heating power feedforward pulse amplitude, expressed as a percentage of rated power. This refers to the current diameter of the heating rod, in millimeters. The diameter of the bar stock used for calibration is in millimeters. The feed rate is measured in millimeters per second in real time. The feed rate used for calibration is in millimeters per second. When no Curie point crossing trigger signal is received, the sloping in / out ramp generator outputs 0, and the output adder only outputs the current power setpoint from the PID output register. The coil operates in normal PID closed-loop temperature control mode. Upon receiving the Curie point crossing trigger signal, the sloping in / out ramp generator starts, operating at a fixed sloping time... The pulse output is linearly increased from 0 to the corrected transient heating power feedforward pulse amplitude, with a fixed asymptotic time. A fixed fade-out time of 100 milliseconds before the end of the transient heating power feedforward pulse duration. The pulse output is linearly reduced from the corrected transient heating power feedforward pulse amplitude to 0. The instantaneous pulse value output during the gradual in and out process is denoted as... The unit is a percentage of rated power. Each control cycle, the output adder adds the current power setpoint in the PID output register to the instantaneous pulse value output by the gradual-in / gradual-out ramp generator to obtain the total power setpoint. The total power setpoint satisfies the formula , in the formula The total power setpoint, expressed as a percentage of rated power, is calculated and fed into the output limiting circuit. This circuit restricts the total power setpoint to between 0% and 100%. When the total power setpoint is greater than 100%, the limited output is 100%; when it is less than 0, the limited output is 0%. When the total power setpoint is between 0% and 100%, it is output directly. The limited total power setpoint is then output to the power control terminal of the induction heating power supply. The IGBT conduction angle is adjusted to control the coil output power. Taking a calibrated bar diameter of 50 mm and a feed speed of 10 mm / s with a transient heating power feedforward pulse base amplitude of 45% as an example, when the actual heating bar diameter is 60 mm and the real-time feed speed is 12 mm / s, the corrected transient heating power feedforward pulse amplitude is calculated to be 64.8%. If the current power setpoint stored in the current PID output register... The value is 60%, and the total power setpoint is obtained by directly adding them together, which is 124.8%. After passing through the output limiting circuit, the output is 100% of the rated power. The gradual increase process increases linearly from 0 to 64.8% within 100 milliseconds. After the pulse lasts for 2 seconds, it decreases linearly to 0 within 100 milliseconds. The total power setpoint directly controls the output power of the induction heating power supply, realizing transient power compensation during Curie point crossing. The pulse amplitude is automatically corrected according to the bar diameter and feeding speed, which can adapt to the heating requirements of bars of different specifications. There is no need to reset the pulse parameters for different production specifications. The gradual in and gradual out ramp control can avoid power overcurrent impact caused by power step change. The output limiting circuit ensures that the power supply always works within the rated power range and will not overload. The pulse superposition process does not change the parameters and working state of the original PID closed loop. The feedforward power is only superimposed for a short time during Curie point crossing, which will not affect the normal temperature control process.
[0015] The automatic control state recovery unit includes a hardware timer, bumpless switching logic, and a PID integral tracking circuit. The duration of the transient heating power feedforward pulse is adaptively calculated based on the bar diameter and real-time feed speed. Satisfy the formula , in the formula The duration of the transient heating power feedforward pulse, in seconds. This is a dimensionless coefficient representing the length ratio of the Curie point interval, determined during calibration based on the material of the tool steel. This refers to the current diameter of the heating rod, in millimeters. The feed rate is measured in millimeters per second in real time. Multiplying by the bar diameter yields the axial length of the Curie point magnetic property transition zone. Dividing by the feed rate gives the time required for the bar to pass through the Curie point zone. This ensures the pulse duration completely covers the Curie point crossing process. Upon receiving the Curie point crossing trigger signal and simultaneously with the start of the transient heating power feedforward pulse output, a hardware timer loads the calculated count value corresponding to the transient heating power feedforward pulse duration. The count begins to decrease with a 1-millisecond time base. When the hardware timer counts to the remaining time equal to the fade-out time of 100 milliseconds, a pre-trigger signal is output to activate the PID integral tracking circuit. The value obtained by subtracting the current instantaneous pulse value from the current total power setpoint is written into the integral register of the PID controller, so that the integral term of the PID controller tracks the current actual power output value, eliminating the PID integral saturation phenomenon. When the hardware timer counts to 0, a reset signal is output. The reset signal is divided into three paths. The first path is connected to the reset terminal of the RS flip-flop, resetting the RS flip-flop to a low level, releasing the comparator interlock, and restoring the Curie point crossing event detection state. The second path is connected to the reset terminal of the fading-in and fading-out ramp generator, setting the ramp generator output to 0 and turning off the transient heating power feedforward pulse output. The third path connects to the release terminal of the PID integral tracking circuit. The PID controller resumes normal integral calculation. Since the integral term has already tracked the current power value, the power setpoint smoothly transitions to the PID closed-loop output value after the pulse is removed, without any power step jump. Taking a Curie point interval length proportional coefficient of 0.4, a bar diameter of 50 mm, and a real-time feeding speed of 10 mm / s as an example, substituting into the formula, the transient heating power feedforward pulse duration is calculated to be 2 seconds. The hardware timer loads 2000 1-millisecond count values and begins to count down. When the count reaches 1900 milliseconds, PID integral tracking is activated, and the current... The total power minus the instantaneous pulse value is written to the PID integral register. When the count reaches 2000 milliseconds, three reset signals are output to turn off the pulse and restore PID control. During the entire switching process, the power setpoint change rate does not exceed 0.1% of the rated power per millisecond, which will not cause sudden changes in power supply current or temperature fluctuations. The adaptive pulse duration can match the Curie point crossing time length under different diameters and feed speeds, avoiding overcompensation at high speeds or undercompensation at low speeds due to fixed compensation time. The PID integral tracking bumpless switching can eliminate power jumps when the pulse is removed, avoid secondary temperature fluctuations, and ensure continuous temperature control.
[0016] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. An automatic constant temperature control system for continuous induction heating of tool steel long bars, characterized in that, include: The coil electrical parameter synchronous sampling unit collects the terminal voltage signal and loop current signal of each induction heating coil through voltage transformers and current transformers, and combines the voltage and current zero-crossing detection circuit to perform phase synchronous triggering sampling of AC cycles, etc., to obtain discrete voltage and current signals. The load characteristic sliding window calculation unit calculates the true effective value and voltage-current phase difference of the discrete voltage and current signals within the sliding time window, and outputs the load equivalent resistance value and power factor value of each coil in real time. The Curie point crossing event discrimination unit performs a first-order backward differential calculation on the equivalent resistance value of the load to obtain the resistance change rate. When the resistance change rate reaches a preset decrease threshold and the power factor decrease simultaneously reaches a preset decrease threshold, the power factor decrease is the difference between the average power factor value over the most recent 500 milliseconds and the real-time power factor value. It outputs the Curie point crossing trigger signal through dual criteria and logic triggering interlocking, while the built-in anti-shake delay circuit filters out false triggers caused by feeding jitter and load fluctuations. After receiving the Curie point crossing trigger signal, the transient power pulse superposition unit superimposes a transient heating power feedforward pulse with fixed amplitude and fixed width onto the PID closed-loop output of the corresponding coil. The amplitude of the transient heating power feedforward pulse is corrected based on the current bar diameter and real-time feed speed, according to the calibrated basic amplitude of the transient heating power feedforward pulse. The corrected amplitude of the transient heating power feedforward pulse satisfies the formula... In the formula, Upuse is the corrected transient heating power feedforward pulse amplitude, in percentage of rated power; Upusebase is the transient heating power feedforward pulse base amplitude stored during calibration; d is the current heating bar diameter, in millimeters; d0 is the bar diameter used during calibration, in millimeters; v is the real-time feed rate, in millimeters per second; and v0 is the feed rate used during calibration, in millimeters per second. The automatic recovery unit automatically removes the superposition of transient heating power feedforward pulses after the output of the transient heating power feedforward pulse reaches the duration of the transient heating power feedforward pulse, restoring the corresponding coil to its original PID closed-loop temperature control state. The duration of the transient heating power feedforward pulse is adaptively calculated based on the current bar diameter and real-time feed speed, and satisfies the formula... In the formula, tcomp is the duration of the transient heating power feedforward pulse in seconds, kcomp is the Curie point interval length proportionality coefficient, which is dimensionless and determined according to the tool steel material during calibration, d is the current heating bar diameter in millimeters, and v is the real-time feed speed in millimeters per second.
2. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The coil electrical parameter synchronous sampling unit includes a signal pre-conditioning circuit, a high-speed analog-to-digital converter, and a sampling trigger logic circuit. The sampling trigger logic circuit is used to shape and divide the synchronous trigger pulse output by the voltage and current zero-crossing detection circuit to generate an equally spaced sampling trigger sequence synchronized with the AC cycle. This ensures that the analog-to-digital conversion of the voltage channel and the current channel starts at the same time to eliminate the sampling phase difference between the two signals. At the same time, it controls the number of sampling points in each AC cycle to be a fixed integer. The voltage and current zero-crossing detection circuit detects the positive zero-crossing time of the terminal voltage signal and the loop current signal respectively and outputs a synchronous reference pulse to the sampling trigger logic circuit. The trigger signal generated by the sampling trigger logic circuit controls the high-speed analog-to-digital converter to sample synchronously in equal phase in each AC cycle. The sampled signal is filtered by the signal pre-conditioning circuit to obtain discrete voltage and current signals.
3. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 2, characterized in that: The true RMS value calculation employs a sliding time window recursive root mean square algorithm. The discrete voltage and current values within the sliding time window are stored in independent voltage and current ring buffers. The depth of these buffers is set to a fixed number of sampling points corresponding to an integer number of AC cycles. After each new equal-phase synchronous sampling is completed, the latest obtained discrete voltage and current values are written to the tail of their respective buffers. Simultaneously, the earliest stored discrete voltage and current values at the head of the corresponding buffer are removed, maintaining a constant number of sampling points within the sliding time window. Using the length of an integer number of AC cycles as the sliding time window, the sum of the squares of the discrete voltage and current values within the sliding time window is accumulated, divided by the number of sampling points within the sliding time window, and then the square root is taken to obtain the true RMS voltage and true RMS current values.
4. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 3, characterized in that: The voltage-current phase difference is calculated by recording the time difference between the positive zero-crossing moment of the voltage signal and the positive zero-crossing moment of the loop current signal within the same cycle, and then converting it with the cycle duration of the current AC cycle. Finally, the cosine value of the voltage-current phase difference is taken to obtain the real-time power factor value.
5. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 4, characterized in that: The equivalent resistance value of the load is obtained by dividing the true effective value of the voltage by the true effective value of the current to obtain the equivalent impedance modulus of the coil load. Then, the equivalent impedance modulus of the coil load is multiplied by the power factor value to obtain the equivalent resistance value of the load.
6. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The first-order backward differential calculation uses a fixed control cycle as the interval, subtracts the load equivalent resistance value of the previous adjacent control cycle from the load equivalent resistance value in the current control cycle, divides by the control cycle duration, and normalizes to obtain the resistance change rate per unit time.
7. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The preset threshold for the rate of decrease in resistance and the preset threshold for the decrease in power factor are obtained by trial heating and calibration of tool steel of the same specification during the debugging phase. During the calibration process, the real-time rate of decrease in resistance and the decrease in power factor when the Curie point magnetic properties of the long bar of tool steel change abruptly occur are recorded and stored in a non-volatile parameter memory for runtime recall. The dual-criteria and logic-triggered latch includes a resistance change rate comparator, a power factor comparator, and a hardware AND gate. The resistance change rate comparator compares the resistance change rate with a preset threshold for the rate of decrease in resistance change. The power factor comparator compares the power factor decrease with a preset threshold for the decrease in power factor. When both the resistance change rate comparator and the power factor comparator output valid level signals, the hardware AND gate outputs a high-level trigger signal to the RS flip-flop, which latches the Curie point crossing trigger signal.
8. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The anti-shake delay circuit is integrated between the output of the hardware AND gate and the set terminal of the RS flip-flop. It adopts a re-triggerable monostable multivibrator. The anti-shake time window parameters, the preset drop threshold of the resistance change rate, and the preset drop threshold of the power factor are stored together in the non-volatile parameter memory during calibration. The high-level signal output by the hardware AND gate can be used as input to re-trigger the monostable multivibrator. When the duration of the high-level signal is less than the debounce time window, it is determined to be a false pulse, and the monostable multivibrator automatically resets and does not output a signal. When the duration of the high-level signal is greater than the anti-jitter time window, the monostable multivibrator outputs a stable high level to the subsequent rising edge detection circuit. The rising edge detection circuit extracts the level transition edge and generates a narrow set pulse of fixed width, which is then sent to the set terminal of the RS flip-flop. The RS trigger adopts a set-priority structure. The set terminal receives a narrow set pulse, and the reset terminal simultaneously receives the timing completion signal from the automatic recovery unit of the control state and the system power-on reset signal. The gate circuit output terminal outputs an interlock signal at the same time as outputting the Curie point crossing trigger signal to turn off the output enable of the resistance change rate comparator and the power factor comparator. This keeps the level of the Curie point crossing trigger signal stable throughout the entire transient heating power feedforward pulse compensation period, avoiding repeated triggering caused by load fluctuations during the crossing process of the Curie point crossing trigger signal.
9. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The transient power pulse superposition unit includes a PID output register, a pulse amplitude register, and an output adder. The PID closed-loop output value is stored in the PID output register, and the basic amplitude of the transient heating power feedforward pulse is stored in the pulse amplitude register. During operation, the basic amplitude of the transient heating power feedforward pulse is corrected according to the current bar diameter and real-time feeding speed to obtain the amplitude of the transient heating power feedforward pulse. After receiving a valid trigger signal, the output adder adds the values of the PID output register and the pulse amplitude register as the current coil power setpoint. At the same time, the output limiting circuit is set to limit the total power within the maximum allowable output range of the power supply.
10. The automatic constant temperature control system for continuous induction heating of tool steel long bars according to claim 1, characterized in that: The automatic control state recovery unit includes a hardware timer and a bumpless switching logic. The hardware timer starts counting while outputting the transient heating power feedforward pulse. When the counting time reaches the duration of the transient heating power feedforward pulse, the hardware timer outputs a reset signal to turn off the output of the pulse amplitude register. The bumpless switching logic smoothly switches the power setpoint back to the PID closed-loop output value, thus completing the automatic control state recovery.
Citation Information
Patent Citations
Hot-rolled high-strength steel full-continuous heat treatment production system and method
CN112143877A
Temperature control device
CN114365060A
