A variable frequency converter torque correction method based on process PID and arithmetic unit linkage

Abstract

This invention discloses a frequency converter torque correction method based on the linkage of process PID and arithmetic unit, belonging to the field of motor drive control technology. The arithmetic unit calculates the reference torque based on the process setpoint and real-time process variables; the process PID controller calculates the deviation signal based on the deviation between the setpoint and the feedback value; the deviation signal is used as a dynamic excitation signal, and the reference torque is multiplicatively modulated by a nonlinear modulation function to generate the final torque setpoint, which is then output to the torque channel of the frequency converter. Furthermore, the modulation depth coefficient is adjusted in real time according to the system's dynamic stiffness coefficient; the integral enable threshold is adaptively adjusted according to the rate of change of the reference torque. Compared with existing technologies, this method effectively reduces torque overshoot, enhances the system's adaptability to variable stiffness conditions, and effectively suppresses integral saturation. It is suitable for industrial scenarios requiring precise torque correction, such as constant tension control in winding and unwinding, constant pressure control in extruders, wire drawing machines, and cranes.

Description

Technical Field

[0001] This invention relates to the field of motor drive control technology, and in particular to a frequency converter torque correction method based on process PID and arithmetic unit linkage. It is particularly suitable for processes that require constant tension control, constant pressure control or constant torque control, including but not limited to winding and unwinding equipment, extruders, wire drawing machines, cranes and fan and pump loads. Background Technology

[0002] When driving AC motors, frequency converters typically employ V / f control, vector control, or direct torque control modes. In process applications such as constant tension control and constant pressure control, it is necessary to dynamically adjust the motor's output torque based on real-time feedback signals (such as tension sensors and pressure transmitters) to maintain constant process parameters.

[0003] In existing technologies, conventional methods for achieving closed-loop torque control include two types: process PID direct torque control and arithmetic unit feedforward and process PID feedback superposition control. Process PID direct torque control directly uses the output of the process PID controller as the torque setpoint for the frequency converter. It responds quickly when the load disturbance is large, but PID parameter tuning is difficult; an excessively large proportional gain can easily cause oscillation, while an excessively small gain results in a slow response. At low or zero speeds, pure torque control can easily lead to motor runaway or overspeed. Arithmetic unit feedforward and process PID feedback superposition control calculates the feedforward torque based on process variables (such as roll diameter and linear velocity) using the arithmetic unit, and the process PID calculates the corrected torque based on sensor feedback. The two are linearly added together to obtain the final torque setpoint, i.e., T = T0. AU +T PID This scheme is widely used in engineering, but the PID output and the arithmetic unit output are directly added together, and their weights do not change with the operating conditions. Under operating conditions with low reference torque (such as the initial stage of winding), the PID correction amount is relatively too large, which can easily cause over-adjustment; under operating conditions with high reference torque, the PID correction amount is relatively too small, and the correction effect is insufficient. Moreover, the sensitivity of the system to torque changes (i.e., process stiffness) varies significantly under different operating conditions, but the PID parameters of the existing scheme are fixed and cannot be adaptively adjusted. In addition, the integral action of traditional PID reduces the dynamic response speed of the system while eliminating steady-state error, and it is difficult to balance the two.

[0004] Therefore, there is an urgent need for a frequency converter torque correction scheme that can achieve deep integration of feedforward and feedback, has adaptive operating condition capabilities, and simultaneously considers response speed and stability. To this end, a frequency converter torque correction method based on the linkage of process PID and arithmetic unit is proposed. Summary of the Invention

[0005] The main objective of this invention is to provide a frequency converter torque correction method based on the linkage of process PID and arithmetic unit, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, this invention provides a method for inverter torque correction based on the linkage of process PID and arithmetic unit, comprising the following steps: Includes the following steps: Step S1: Calculate the reference torque T using the arithmetic unit based on the process setpoint and real-time process variables. AU ; Step S2: Calculate the deviation signal e based on the deviation between the given value and the feedback value using the process PID controller; Step S3: Using the deviation signal e as a dynamic excitation signal, the reference torque T is modulated by a nonlinear modulation function. AU Modulation is performed to generate the final torque setpoint T. final ; Step S4: Set the final torque setpoint T final The torque channel output to the frequency converter drives the motor. The nonlinear modulation function is a multiplicative function, i.e., T final =T AU ×m, where the modulation coefficient m is a nonlinear function of the deviation signal e, and the range of m is limited to (1-α). max ,1+α max Within ) α max This is the preset maximum modulation depth.

[0007] Preferably, the nonlinear function is a hyperbolic tangent function, expressed as: m = 1 + α × tanh(β × e / Trated), where m is the modulation coefficient; α is the dynamic modulation depth coefficient, ranging from 0.05 to 0.25; β is the sensitivity coefficient, ranging from 1.0 to 3.0; e is the deviation signal; and T... rated is the rated torque of the motor; tanh is the hyperbolic tangent function, used to achieve smooth amplitude limiting.

[0008] Preferably, the nonlinear function can also be a piecewise linear saturated function, expressed as: E sat This is the saturation threshold.

[0009] Preferably, the nonlinear function can also be an sigmoid logistic function, expressed as: .

[0010] Preferably, the nonlinear function can also be a square root type function, expressed as: E ref This is a reference deviation value.

[0011] All of the above substitution functions can achieve the following characteristics: approximately linear output when the deviation signal is small, and smooth saturation to the limit value when the deviation signal is large.

[0012] Preferably, the method further includes a stiffness adaptive observation step: It also includes a stiffness adaptive observation step: The response characteristics of the real-time observation system to torque changes are used to calculate the dynamic stiffness coefficient K. stiffness =ΔPV / ΔT AU Where ΔPV is the change in the feedback value, and ΔT AU This represents the change in the reference torque. According to the dynamic stiffness coefficient K stiffness The modulation depth coefficient α is dynamically adjusted by a preset mapping relationship, wherein the larger the dynamic stiffness coefficient, the smaller the modulation depth coefficient α.

[0013] Preferably, the stiffness adaptive observation step further includes: The dynamic stiffness coefficients calculated over multiple consecutive sampling periods are subjected to a sliding window averaging filter to obtain smoothed stiffness coefficients, which are expressed as follows: K smooth K is the smoothed stiffness coefficient. inst (ki) is the dynamic stiffness coefficient, and N is the sliding window length; A first-order low-pass filter is applied to the modulation depth coefficients obtained by mapping the smoothed stiffness coefficients to suppress abrupt changes in the modulation depth coefficients, as shown below: α new (k) represents the modulation depth coefficient updated in the current period, α calc (k) represents the original modulation depth coefficient calculated for the current period, α old (k-1) represents the modulation depth coefficient updated in the previous cycle, and λ represents the filter coefficient.

[0014] Preferably, the method further includes a dynamic integral separation step: Calculate the rate of change of the reference torque with respect to time, dT AU / dt; The integral enable threshold ε is dynamically adjusted based on the rate of change of the reference torque. dynamic Among them, the wider the rate of change of the reference torque, the wider the integral enable threshold; the smaller the rate of change of the reference torque, the narrower the integral enable threshold, expressed as: ε dynamic =ε0+γ×|dT AU / dt|, ε0 is the basic integration threshold, and γ is the dynamic coefficient; When the absolute value of the deviation signal is less than the integral enable threshold, the integral term of the PID controller is updated; when the absolute value of the deviation signal is greater than or equal to the integral enable threshold, the integral term is frozen, and only the proportional and derivative terms are retained.

[0015] Preferably, the process PID controller employs an incremental PID algorithm: In each control cycle, the changes in the proportional, integral, and derivative terms are calculated, and these changes are accumulated to the PID output value of the previous cycle to obtain the PID output value of the current cycle; wherein the accumulation of the integral term is conditional upon the judgment result of the dynamic integral separation step, expressed as: Δu(k)=K p ×[e(k)-e(k-1)]+K i ×e(k)+K d ×[e(k)-2e(k-1)+e(k-2)] u PID (k)=u PID (k-1)+Δu(k) Where e(k) is the current cycle deviation, e(k-1) is the previous cycle deviation, e(k-2) is the deviation between the previous two cycles, and K... p K i K d These are the proportional, integral, and differential coefficients, respectively.

[0016] Preferably, the method further includes amplitude limiting and rate limiting steps for the final torque: Amplitude limiting: This limits the final torque setpoint to between a preset maximum torque limit and a preset minimum torque limit, denoted as: T final =max(T min ,min(T max ,T final )), where T max T is the maximum torque limit value. min This is the minimum torque limit value; Rate limiting: Limits the change in the final torque setpoint within each sampling period to no more than the product of the preset maximum rate of change and the sampling period, expressed as: |T final (k)-T final (k-1)∣≤ΔT max ×T s , where ΔT max T is the maximum allowable rate of torque change per sampling period. s The sampling period.

[0017] Preferably, the reference torque calculation performed by the arithmetic unit uses different calculation methods depending on the process type: For the constant tension control process during winding, the reference torque is equal to the tension setpoint multiplied by half the real-time roll diameter, expressed as: T AU = F set ×D / 2, where F set Where D is the tension setting value and D is the real-time roll diameter; For the constant tension control process during unwinding, the reference torque is equal to the negative tension setpoint multiplied by half the real-time roll diameter, expressed as: T AU = -F set ×D / 2; For constant pressure extrusion processes, the reference torque equals the pressure setpoint multiplied by a preset proportional coefficient, expressed as: T AU =K p ×P set , where P set For the pressure setpoint, K p This is the proportionality coefficient; For constant air volume fan processes, the reference torque equals the square of the air volume setpoint divided by the current speed, then multiplied by a preset constant, expressed as: T AU = K×Q set 2 / n, where Q set Here, n is the airflow setpoint, n is the rotational speed, and K is a constant.

[0018] Preferably, the method operates in the programmable logic control unit inside the frequency converter, and the sampling period is synchronized with the pulse width modulation carrier period of the frequency converter, with a value range of 0.002~0.01 seconds; the arithmetic unit and the process PID controller are executed in parallel with the same sampling period; the stiffness adaptive observation step is executed at a preset frequency, and the modulation depth coefficient is updated once every preset sampling period (preferably 10~50).

[0019] Beneficial effects Compared with the prior art, the present invention has the following beneficial effects: This scheme uses multiplicative modulation instead of additive superposition. The PID output is no longer directly superimposed but is proportionally modulated to the reference torque. The absolute value of the correction is large for large torques and small for small torques, conforming to physical laws. Compared to existing superposition methods, this effectively reduces overshoot.

[0020] This solution automatically adjusts the modulation depth by observing the system's response characteristics to torque in real time, eliminating the need for repeated manual tuning of PID parameters and significantly improving the system's adaptability to different operating conditions.

[0021] This scheme uses dynamic integral separation to automatically adjust the integral enable threshold based on the rate of change of the reference torque. When the reference torque changes rapidly, integral intervention is allowed to compensate for dynamic errors, while the threshold is tightened in steady state to prevent the accumulation of static error, effectively balancing dynamic response and steady-state accuracy.

[0022] This scheme uses the tanh function to achieve smooth limiting of the correction amount, avoiding the jitter and shock introduced by traditional hard limiting.

[0023] This solution can be implemented directly on the programmable platform of existing high-end frequency converters without incurring additional hardware costs. Attached Figure Description

[0024] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a diagram showing the relationship between the signal flow and functional modules in the method of the present invention. Figure 3 The characteristic curve of the hyperbolic tangent nonlinear modulation function is shown. Figure 4 This is a diagram showing the mapping relationship between stiffness adaptive observation and modulation depth coefficients. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0026] Example 1: Application of constant tension control during winding This embodiment uses the film winding process as an example to describe the specific implementation of the present invention. Equipment configuration: asynchronous motor (rated power 22kW, rated torque 140Nm), Huichuan MD800 series frequency converter (supporting user-programmable arithmetic units and process PID), tension sensor (range 0~2000N), roll diameter detection is performed by ultrasonic sensor or by calculation of linear speed / speed ratio.

[0027] (I) System Initialization After the frequency converter is powered on, perform the following initialization operations: (1) Read the parameters on the motor nameplate and calculate the rated torque T. rated =9550×22 / 1470≈143Nm; (2) Set the basic parameters as shown in the table below:

[0028] (3) Clear the historical buffer of the stiffness observer (100 sampling points).

[0029] (ii) Control cycle execution process (executed once every 4 milliseconds) Step 1: Calculate the reference torque T using the arithmetic unit. AU Collect current process parameters: Tension setpoint F set =800N (given via PLC communication); The current roll diameter D = 0.35m (calculated in real time using the linear velocity / rotation speed ratio).

[0030] Calculate: T AU =F set × D / 2 =800×0.35 / 2=140Nm.

[0031] Step 2: Calculate the deviation signal using process PID control. The tension sensor feedback value was collected: PV = 785N; Calculate the deviation: e raw =F set -PV = 800-785 = 15N.

[0032] Converted to equivalent torque deviation: e = e raw × D / 2 = 15 × 0.175 = 2.625Nm.

[0033] Incremental PID algorithm is used: Δu(k)=K p ×[e(k)-e(k-1)]+K i ×e(k)+K d ×[e(k)-2e(k-1)+e(k-2)] u PID (k)=u PID (k-1)+Δu(k) Where e(k) is the current cycle deviation, which is 2.625, e(k-1) is the deviation of the previous cycle, e(k-2) is the deviation of the two previous cycles, and K... p K i K d These are the proportional, integral, and differential coefficients, respectively.

[0034] In this embodiment, the deviation of the previous cycle is 2.100; the u of the previous cycle PID (k-1) is 1.850; Then we have: Δu(k) = 0.8×(2.625-2.100)+0.15×2.625+0 = 0.42+0.394 = 0.814Nm.

[0035] u PID (k) = 1.850+0.814 = 2.664Nm.

[0036] Step 3: Stiffness adaptive observation (executed once every 20 sampling periods) In this embodiment, the roll diameter D of the previous cycle is 0.3375 m, and the roll diameter T of the previous cycle is... AU =F set × D / 2 = 800 × 0.3375 / 2 = 135 Nm. The tension sensor feedback value PV changes from 785 N to 777 N; Then we have: Reference torque variation: ΔT AU =140-135 = 5Nm; Feedback change: ΔPV = 785 - 777 = 8N; Calculate instantaneous stiffness: K inst =ΔPV / ΔT AU = 8 / 5 = 1.6 (dimensionless); After sliding window filtering (N=100): ; Get K smooth =1.55; In this embodiment, a stiffness-modulation depth mapping representation example is given as follows:

[0037] Look up the mapping table: K smooth =1.55 falls within the interval 1.0~1.8, corresponding to α. calc =0.12; Low-pass filter update: α new (k)=λ×α calc (k)+(1-λ)×α old (k-1), α new (k) represents the modulation depth coefficient updated in the current period, α calc (k) represents the original modulation depth coefficient calculated for the current period, α old (k-1) represents the modulation depth coefficient updated in the previous cycle, and λ represents the filter coefficient. In this embodiment, the modulation depth coefficient α updated in the previous cycle is... old (k-1) is 0.12, and we take λ = 0.1, then we have: α new =0.1×0.12+0.9×0.12 = 0.12.

[0038] Step 4: Dynamic integration threshold determination Calculate the baseline torque change rate (sampling period 0.004 seconds): In this embodiment, the calculated reference torque for the previous cycle is 138 Nm; therefore, we have: dT AU / dt = (140-138) / 0.004 = 500Nm / s; Calculate the dynamic threshold: ε dynamic =ε0+γ×|dT AU / dt|, ε0 is the basic integration threshold, γ is the dynamic coefficient. In this embodiment, ε0 is 7.15 Nm; γ is 0.1, then: ε dynamic = 7.15 + 0.1 × 500 = 57.15 Nm; Currently, |e| = 2.664 Nm < 57.15 Nm, therefore the integral term is enabled, and the PID updates the integral normally.

[0039] It should be noted that the adaptive stiffness observation calculates the macroscopic stiffness, reflecting the system characteristics, and its data sampling period is once every 20 sampling periods; while the dynamic integral threshold judgment calculates the instantaneous rate of change and adjusts the integral threshold in real time, and its data sampling period is 0.004 seconds. Therefore, the reference torque of 138 Nm in the previous cycle in the dynamic integral threshold judgment step is not the same value as the reference torque of 135 Nm in the previous cycle in the adaptive stiffness observation step.

[0040] Step 5: Nonlinear Modulation Calculate the relative value of the deviation: e norm = u PID (k) / Rated torque T rated =2.664 / 143 =0.0186; Calculate the modulation coefficient: m = 1 + 0.12 × tanh(2.0 × 0.0186) = 1 + 0.12 × 0.0372 = 1.00446; Calculate the final torque: T final = 140 × 1.00446 = 140.62 Nm.

[0041] Step 6: Limiting and Output Amplitude Limit: T final =max(T min ,min(T max ,T final )), where T max T is the maximum torque limit value. min This is the minimum torque limit value; In this embodiment, the maximum torque limiting value T is taken. max 1.5T rated It is approximately 215 Nm, and the minimum torque limit is the reverse value, i.e., -1.5 T. rated Since it is approximately -215 Nm, then: T final = max(-215,min(215,140.62))=140.62Nm; Rate limit: Previous cycle T final = 139.80 Nm, change 0.82 Nm, less than the threshold ΔTmax ×T s =0.1×143 / 0.004×0.004 = 14.3Nm, output directly.

[0042] Write 140.62 Nm into the inverter torque setpoint register.

[0043] (III) Complete timing diagram Figure 1 A flowchart illustrating the overall process of the method of the present invention is shown. (Refer to...) Figure 1 After system initialization, it enters a periodic loop execution: the arithmetic unit calculates the reference torque, the process PID calculates the deviation, the stiffness adaptive observation (low frequency execution), the dynamic integral threshold judgment, nonlinear modulation, limiting and output, and then enters the next cycle.

[0044] Figure 2 A diagram showing the relationship between the signal flow and functional modules of the method of the present invention is provided. Figure 2 As shown, the output T of the arithmetic unit AU The output e of the process PID is sent to the nonlinear modulation module, and the stiffness adaptive observation module monitors T in real time. AU The dynamic integration threshold module outputs the modulation depth coefficient α to the modulation module based on the change in the feedback value, and the dynamic integration threshold module outputs the modulation depth coefficient α to the modulation module based on dT. AU / dt controls the enable state of the PID integral term.

[0045] Figure 3 The nonlinear modulation function m = 1 + α × tanh(β × e / T) is shown. rated The characteristic curve of ), where α=0.15, β=2.0, T rated =100. The horizontal axis represents the normalized deviation e / T. rated The vertical axis represents the modulation coefficient m. It can be seen that: the small deviation region exhibits approximately linear amplification, the large deviation region shows smooth saturation, and the region is continuously differentiable at zero crossings.

[0046] (iv) Comparative Experiment On the same hardware platform, the method of this invention is compared with the existing superposition method (T). final = T AU +T PID A comparative test was conducted, with the working condition being a step change in the roll diameter from 0.2m to 0.5m.

[0047] The test results are shown in the table below:

[0048] Example 2: Adaptive effect under variable stiffness conditions This embodiment simulates a scenario where the stiffness of the process changes during operation. In the initial stage (roll diameter 0.2m), the system stiffness is low (K≈0.8); in the middle section (roll diameter 0.5m), the stiffness is medium (K≈1.5); and in the final section (roll diameter 0.8m), the stiffness is high (K≈2.5).

[0049] The results of the stiffness adaptive observation are as follows:

[0050] Figure 4 The diagram showing the mapping relationship between stiffness adaptive observation and modulation depth coefficient is presented.

[0051] Experiments show that when a fixed α=0.15 is used, the initial correction is insufficient (overshoot 15%), while the final correction is excessive (oscillation for 3 cycles). Using the adaptive scheme of this invention, the optimal correction strength is maintained at each stage, with no oscillation throughout, and the maximum overshoot is controlled within 8%.

[0052] Example 3: Verification of the effect of dynamic integral separation This embodiment simulates the winding acceleration process (accelerating from 0.5m / s to 2.0m / s).

[0053] During acceleration, the reference torque change rate dT AU / dt up to 800 Nm / s, dynamic integration threshold ε dynamic Automatically relaxed to 87 Nm, with the integral term enabled throughout, effectively compensating for dynamic tension deviations during acceleration.

[0054] After steady state, dT AU With / dt ≈0, the threshold shrinks to 7.15 Nm, and the integral action only intervenes under extremely small deviations, eliminating the risk of integral saturation.

[0055] Alternative implementation methods The above embodiments use the tanh function as the nonlinear modulation function. Those skilled in the art should understand that other functions with smooth saturation characteristics can also be used instead, for example: Alternative method 1: Piecewise linear modulation. When |e| ≤ E sat When |e| > E, the modulation coefficient changes linearly; sat When the modulation coefficient remains at its saturation value, it is expressed as: E sat This is the saturation threshold. This method requires less computation and is suitable for frequency converters with limited computing resources.

[0056] Alternative method 2: S-type logistic function, expressed as: The effect is equivalent to the tanh function.

[0057] Alternative method 3: Square root type function, represented as: E ref It uses the reference deviation value and has higher sensitivity in the small deviation region.

[0058] The method of this invention can be widely applied to the following industrial scenarios: Unwinding and winding equipment: constant tension control for films, paper, metal foils, and cables, with a roll diameter ratio of up to 10:1.

[0059] Extruder: Constant pressure control for plastic extrusion and rubber calendering.

[0060] Wire drawing machine: Torque distribution control in a multi-motor linkage wire drawing production line.

[0061] Cranes: Torque smoothing control to prevent load swaying.

[0062] Fans and pumps: Heavy-load starting scenarios requiring soft start and torque compensation.

[0063] This invention can be directly implemented on existing mainstream frequency converter platforms, including but not limited to: Siemens SINAMICS series, ABB ACS880 series, Huichuan MD800 series, Senlan Hope530 series, INVT Goodrive series and other frequency converter products that support user programmable arithmetic units.

[0064] 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 illustrative of the principles of 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 this invention is defined by the appended claims and their equivalents.

Claims

1. A method for torque correction of a frequency converter based on the linkage of process PID and arithmetic unit, characterized in that, include: The reference torque is calculated using the arithmetic unit based on the process setpoints and real-time process variables; The process PID controller is used to calculate the deviation signal based on the deviation between the given value and the feedback value; The deviation signal is used as a dynamic excitation signal, and the reference torque is modulated by a nonlinear modulation function to generate the final torque setpoint. The final torque setpoint is output to the torque channel of the frequency converter to drive the motor. Wherein, the nonlinear modulation function is a multiplicative function; in the nonlinear modulation function, the modulation coefficient is a nonlinear function of the deviation signal, and the range of the modulation coefficient is limited to a preset maximum modulation depth range; The nonlinear function is a hyperbolic tangent function, expressed as: m = 1 + α × tanh(β × e / Trated), where m is the modulation coefficient; α is the modulation depth coefficient, ranging from 0.05 to 0.25; β is the sensitivity coefficient, ranging from 1.0 to 3.0; e is the deviation signal; T rated is the rated torque of the motor; tanh is the hyperbolic tangent function, used to achieve smooth amplitude limiting; It also includes a stiffness adaptive observation step: The response characteristics of the real-time observation system to torque changes are used to calculate the dynamic stiffness coefficient K. stiffness =ΔPV / ΔT AU Where ΔPV is the change in the feedback value, and ΔT AU This is the change in the reference torque; According to the dynamic stiffness coefficient K stiffness The modulation depth coefficient α is dynamically adjusted through a preset mapping relationship f, wherein the larger the dynamic stiffness coefficient, the smaller the modulation depth coefficient α.

2. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, The stiffness adaptive observation step also includes: The dynamic stiffness coefficients calculated from multiple consecutive sampling periods are subjected to sliding window averaging filtering to obtain smoothed stiffness coefficients. A first-order low-pass filter is applied to the modulation depth coefficients obtained by mapping the smoothed stiffness coefficients to suppress abrupt changes in the modulation depth coefficients.

3. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, It also includes a dynamic integral separation step: Calculate the rate of change of the reference torque relative to time; The integral enable threshold is dynamically adjusted based on the rate of change of the reference torque: the larger the rate of change of the reference torque, the wider the integral enable threshold; the smaller the rate of change of the reference torque, the narrower the integral enable threshold. When the absolute value of the deviation signal is less than the integral enable threshold, the integral term of the PID controller is updated; when the absolute value of the deviation signal is greater than or equal to the integral enable threshold, the integral term is frozen, and only the proportional and derivative terms are retained.

4. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, The reference torque calculation performed by the arithmetic unit uses different calculation methods depending on the process type: For the constant tension control process during winding, the reference torque is equal to the tension setpoint multiplied by half the real-time roll diameter; For the unwinding constant tension control process, the reference torque is equal to the negative tension setpoint multiplied by half the real-time roll diameter. For constant pressure extrusion processes, the reference torque is equal to the pressure set value multiplied by a preset proportional coefficient; For constant air volume fan technology, the reference torque is equal to the square of the air volume setting value divided by the current speed and then multiplied by a preset constant.

5. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, The process PID controller uses an incremental PID algorithm: In each control cycle, the changes in the proportional, integral, and derivative terms are calculated, and these changes are accumulated to the PID output value of the previous cycle to obtain the PID output value of the current cycle.

6. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, It also includes a final torque limiting step: Amplitude limiting: Limits the final torque setpoint between the preset maximum torque limit value and the minimum torque limit value; Rate limit: Limits the change in the final torque setpoint within each sampling period to no more than the product of the preset maximum rate of change and the sampling period.

7. The inverter torque correction method based on process PID and arithmetic unit linkage according to claim 1, characterized in that, The nonlinear function can also be any one of the following: a piecewise linear saturated function, a sigmoid logistic function, or a square root function.

8. A method for torque correction of a frequency converter based on the linkage of process PID and arithmetic unit according to any one of claims 1-7, characterized in that, The method operates within the programmable logic control unit inside the frequency converter. The sampling period is synchronized with the pulse width modulation carrier period of the frequency converter, and the value range is 0.002~0.01 seconds. The arithmetic unit and the process PID controller are executed in parallel with the same sampling period. The stiffness adaptive observation step is executed at a preset frequency, and the modulation depth coefficient is updated once every preset sampling period.

Images