A method for adaptive PWM power control of carbon nanotube heating films
By using an adaptive thermoelectric coupling state observation model and closed-loop regulation technology, the temperature and resistance changes of the carbon nanotube heating film are corrected in real time, and differential mode compensation components are generated. This solves the problems of uneven temperature distribution and insufficient total power tracking accuracy, and improves the stability and uniformity of the heating film.
Patent Information
- Application Number
- CN202611611578.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-19
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies struggle to detect the temperature distribution and equivalent impedance changes of carbon nanotube heating films in real time, resulting in insufficient total power tracking accuracy, weak active temperature uniformity adjustment capability, and a tendency for localized overheating and uneven heating, which negatively impacts the heating experience and reliability.
Real-time temperature distribution and equivalent phase resistance are obtained by an adaptive thermoelectric coupling state observation model. The thermal parameter polynomial coefficients are corrected online by using temperature deviation. A three-phase fundamental voltage modulation signal and differential mode compensation component are generated by combining the target total power and temperature uniformity. Closed-loop regulation and PWM drive are then performed to improve temperature uniformity and stability.
This improved the uniformity of temperature distribution and operational stability of the carbon nanotube heating film, thereby enhancing control accuracy and reliability under complex heat dissipation conditions.
Smart Images

Figure CN123124145A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of adaptive control, and in particular relates to a PWM power adaptive control method for a carbon nanotube heating film. Background Technology
[0002] In large-scale applications and complex working environments, carbon nanotube heating films exhibit significant thermoelectric coupling characteristics, and their equivalent resistance drifts nonlinearly with temperature changes. Traditional open-loop control or single closed-loop control typically adjusts based solely on set power, average temperature, or basic electrical parameters, making it difficult to reflect the internal temperature distribution and impedance changes of the heating film in a timely manner. This can easily cause the actual output power to deviate from the target requirement. Furthermore, due to inconsistent external heat dissipation conditions, local environmental disturbances, and microscopic differences in the material preparation process, the heating film is prone to localized overheating, uneven heating, or phase-to-phase heating imbalances during operation. This not only affects the heating experience and process quality but may also cause localized thermal stress concentration, accelerating material aging and reducing reliability.
[0003] Existing pulse-width modulation (PWM) power control methods primarily focus on inverter-side power conversion efficiency and basic electrical quantity control, paying insufficient attention to the spatial thermal field distribution at the load end and lacking an active adjustment mechanism for temperature uniformity. While zero-sequence voltage injection in three-phase modulation can improve DC bus voltage utilization, enhance modulation margin, or reduce switching losses, it typically does not alter the three-phase line voltage difference, making it difficult to directly achieve differentiated distribution of inter-phase heating power. Therefore, there is an urgent need for a PWM power adaptive control method tailored to the thermoelectric coupling characteristics of carbon nanotube heating films. This method should achieve total power tracking while incorporating real-time temperature distribution and equivalent phase resistance changes. Through modulation optimization and three-phase differential mode compensation, it should proactively improve the temperature uniformity of the heating film, enhancing the control stability and reliability of large-area heating films under complex heat dissipation conditions. Summary of the Invention
[0004] To address the problems in existing technologies such as difficulty in real-time sensing of temperature distribution and equivalent impedance changes in carbon nanotube heating films, insufficient total power tracking accuracy, and weak active temperature uniformity adjustment capability, in a first aspect, this invention provides a PWM power adaptive control method for carbon nanotube heating films, the method comprising: The real-time three-phase current and voltage of the three-phase inverter with a three-phase carbon nanotube heating film and the temperature values of multiple discrete temperature measurement points are obtained. The real-time temperature distribution matrix and equivalent phase resistance matrix are calculated using an adaptive thermoelectric coupling state observation model. The deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point is used as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model. Based on the target total power and target temperature uniformity, the total power deviation and temperature uniformity deviation are calculated by combining the real-time temperature distribution matrix and the equivalent phase resistance matrix. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of the physical region of each phase. The three-phase fundamental voltage modulation signal is algebraically superimposed with the corresponding phase compensation components in the zero-sequence modulation optimization component and the three-phase temperature equalization differential mode compensation component, and the DC bus voltage is limited to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
[0005] On the other hand, the present invention also provides a carbon nanotube heating film PWM power adaptive control system, the system comprising the following modules: The correction module is used to obtain the real-time three-phase current and voltage of the three-phase inverter of the three-phase carbon nanotube heating film and the temperature values of multiple discrete temperature measurement points. It uses an adaptive thermoelectric coupling state observation model to calculate the real-time temperature distribution matrix and the equivalent phase resistance matrix. The deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point is used as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model. The calculation module is used to calculate the total power deviation and temperature uniformity deviation based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B and C respectively, and are constructed based on the temperature deviation state of the physical region of each phase. The control module is used to algebraically superimpose the three-phase fundamental voltage modulation signal with the corresponding phase compensation components of the zero-sequence modulation optimization component and the three-phase temperature equalization differential mode compensation component, and perform DC bus voltage limiting constraint to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
[0006] This invention acquires multidimensional electrical parameters and discrete temperature measurement data to construct a thermoelectric coupling state observation model. It then uses temperature deviations to perform online recursive correction of the polynomial coefficients of the thermal parameters, obtaining the real-time temperature distribution and equivalent phase resistance of the heating film. Based on this, a three-phase fundamental voltage modulation signal is generated based on the total power deviation. Zero-sequence modulation optimization components are used to improve the modulation margin, and a three-phase temperature equalization differential mode compensation component is constructed based on the temperature uniformity deviation to provide differentiated compensation for the modulation reference signal of each phase. This method achieves closed-loop tracking of the target total power and improves the temperature distribution uniformity and operational stability of the carbon nanotube heating film. Attached Figure Description
[0007] Figure 1 A flowchart of a PWM power adaptive control method for carbon nanotube heating films; Figure 2 This is a schematic diagram of the complementary PWM drive signals for the upper and lower bridge arms of phase A. Detailed Implementation
[0008] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0009] In this application, a PWM power adaptive control method for a carbon nanotube heating film is disclosed, such as... Figure 1 As shown, the method includes: S1. Obtain the real-time three-phase current and voltage of the three-phase inverter with the three-phase carbon nanotube heating film, as well as the temperature values of multiple discrete temperature measurement points. Calculate the real-time temperature distribution matrix and equivalent phase resistance matrix using an adaptive thermoelectric coupling state observation model. Use the deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model.
[0010] The microcontroller's analog-to-digital converter synchronously acquires the three-phase electrical parameters output from the Hall current and voltage sensors, and obtains temperature values at multiple discrete measurement points through a temperature acquisition module matched with thermocouples, NTC thermistors, or PT100 patch sensors. The carbon nanotube heating film adopts a three-phase partitioned structure.
[0011] An adaptive thermoelectric coupling state observation model is established by solving the partial differential equations of heat conduction, which include Joule heat source terms and natural convection heat dissipation terms, using the backward Euler discretization algorithm. The instantaneous heat generation power of each phase is calculated based on real-time three-phase current and voltage. Based on the three-phase partitioning structure and spatial current distribution mapping, the instantaneous heat generation power of each phase is allocated to discrete nodes within the corresponding physical region, forming a node Joule heat source matrix. This node Joule heat source matrix is then used as the heat source term input to the discretized heat conduction equations. A linear equations solving function is then called to solve the node temperature state equations, thereby outputting the real-time temperature distribution matrix.
[0012] Based on the thermo-resistivity fitting relationship between the resistivity and temperature of carbon nanotube materials, the local equivalent resistance of each discrete node is calculated by combining the real-time temperature distribution matrix, and the equivalent phase resistance matrix is obtained through network equivalent calculation.
[0013] Elements corresponding to the spatial coordinates of discrete temperature measurement points in the real-time temperature distribution matrix are extracted as predicted temperatures. The difference between the predicted and measured temperatures is calculated to obtain the temperature error vector. A recursive least squares algorithm with a forgetting factor is used as an adaptive update mechanism. The sensitivity Jacobian matrix of the predicted temperature relative to the coefficients of the thermal parameter polynomial is used as the regression matrix, and the temperature error vector is used as the error excitation signal. The controller calculates the gain matrix based on the sensitivity Jacobian matrix, the forgetting factor, and the historical covariance matrix, and calculates the parameter correction based on the gain matrix and the temperature error vector. The controller iteratively updates the coefficient vectors of the convective heat transfer coefficient polynomial and the thermal conductivity coefficient polynomial in the thermoelectric coupling state observation model to complete the online recursive correction.
[0014] In some implementations, the calculation of the real-time temperature distribution matrix and equivalent phase resistance matrix using an adaptive thermoelectric coupling state observation model includes: A thermoelectric coupling state equation is constructed, and the current temperature distribution matrix is iteratively calculated based on the temperature matrix of the previous control cycle, the real-time phase voltage and phase current, and the spatial current distribution mapping relationship. Based on the temperature resistance characteristics of carbon nanotube materials, the local equivalent resistance of each discrete node is calculated using a polynomial fitting model. Based on the electrical connection topology of discrete nodes within each phase physical region, network equivalent calculations are performed on the local equivalent resistance to obtain an equivalent phase resistance matrix containing three-phase transient resistance values.
[0015] In the calculation process of the adaptive thermoelectric coupling state observation model, the controller first discretizes the carbon nanotube heating film into a spatial grid, for example, dividing it into 100×100 discrete grid nodes, and establishes a temperature matrix according to the physical position of each grid node on the heating film. The discrete time step of the thermoelectric equation of state is preferably set in the range of 0.1ms to 5ms, for example, set to a 1ms control period.
[0016] The thermoelectric coupling equation of state includes a heat conduction term, a natural convection heat dissipation term, and a nodal Joule heat source term. The heat conduction term characterizes the heat transfer between adjacent discrete grid nodes, and its calculation parameters include the spatial connectivity between adjacent nodes and the material's equivalent thermal conductivity. The natural convection heat dissipation term characterizes the heat released from the heating film nodes to the ambient air, and its calculation parameters include the convective heat transfer coefficient and the ambient temperature. The nodal Joule heat source term characterizes the electric heating input, which is calculated based on the real-time phase voltage, phase current, and spatial current distribution mapping relationship. The convective heat transfer coefficient can be taken as 5 to 25 W / (m²K), for example, set to 15 W / (m²K). The material's equivalent thermal conductivity is set based on the calibration results of the carbon nanotube film material.
[0017] Within each control cycle, the controller calculates the current heat generation power corresponding to phases A, B, and C based on the real-time phase voltage and phase current, and distributes the heat generation power of each phase to discrete nodes in the corresponding physical region through the spatial current distribution mapping relationship, thus obtaining a node Joule heat source matrix with the same dimension as the temperature matrix.
[0018] Specifically, the spatial current distribution mapping relationship is constructed by a three-phase partitioned mask matrix and node conductivity weights. Let the temperature grid be M×N, with phases A, B, and C corresponding to the mask matrix respectively. , and The elements in the mask matrix are used to indicate whether the corresponding grid node belongs to the corresponding physical heating region. The mask matrix is determined based on the three-phase electrode arrangement diagram of the carbon nanotube heating film, the interphase insulation boundary, the feed end position, the return end position, and the physical coordinates of each grid node. When the physical coordinates of a certain grid node fall within the effective heating region corresponding to phase A, phase B, or phase C, the element at that node position in the corresponding phase mask matrix is set to 1; otherwise, it is set to 0.
[0019] For each phase, the controller determines the effective conductive region and node current path of that phase based on the location of the feed end, the location of the return end, and the direction of the thin film trace. Then, it generates node conductivity weights based on the local conductivity values of each node and normalizes all node conductivity weights within the same phase so that the sum of all node weights within the same phase is 1. The controller then distributes the current heat generation power of that phase to each discrete node of the corresponding phase according to the normalized node conductivity weights, forming the node heat source distribution for that phase. The node heat source distributions of phases A, B, and C are then superimposed to obtain the node Joule heat source matrix.
[0020] When solving the real-time temperature distribution matrix, the controller expands the temperature matrix of the previous control cycle into a temperature state vector according to a preset node order, and expands the node Joule heat source matrix into a heat source vector in the same order. A heat capacity matrix is constructed based on the heat capacity parameters of each discrete node; a heat conduction matrix is constructed based on the spatial connection relationship between adjacent nodes and the equivalent thermal conductivity of the material; a convective heat transfer matrix is constructed based on the heat transfer relationship between exposed surface nodes or boundary nodes and the ambient air, as well as the convective heat transfer coefficient; and an environmental heat dissipation term is constructed based on the convective heat transfer matrix and the ambient temperature. Subsequently, the backward Euler method is used to discretize the heat conduction equation in time, forming a system of linear equations with the node temperature state vector of the current control cycle as the unknown. The coefficient matrix of the linear equation system is determined by the heat capacity matrix, heat conduction matrix, convective heat transfer matrix, and discrete time step, while the right-hand side is determined by the temperature state vector, heat source vector, and environmental heat dissipation term of the previous control cycle. Solving this system of linear equations yields the node temperature state vector of the current control cycle, which is then reconstructed into a real-time temperature distribution matrix according to the original spatial grid order.
[0021] The model updates the local equivalent resistance of each node based on the temperature resistance characteristics of the carbon nanotube material. Specifically, it reads the node temperature of the i-th discrete node in the real-time temperature distribution matrix. And calculate the node relative to the reference temperature. Temperature deviation The reference temperature A temperature of 25℃ can be used. The local equivalent resistance of this node is calculated using a third-order temperature resistance fitting polynomial, which represents the temperature deviation. Substituting the pre-calibrated temperature resistance fitting polynomial, the local equivalent resistance of the corresponding node is obtained. The zero-order coefficient of the temperature-resistance fitting polynomial represents the nodal equivalent resistance at the reference temperature, such as 10Ω; the first-order coefficient represents the basic temperature coefficient characteristic; and the second-order and third-order coefficients are used to correct nonlinear resistance drift in higher temperature ranges. By traversing each discrete node in the real-time temperature distribution matrix in the above manner, a local equivalent resistance matrix with the same dimension as the temperature matrix can be obtained.
[0022] During network equivalent calculation, the controller pre-stores the physical boundaries of the three-phase carbon nanotube heating film, the positions of the feed terminals and return terminals of each phase, the interphase insulation boundaries, and the connection relationships of adjacent grid nodes within each phase. For each phase, the controller extracts all nodes covered by the phase mask matrix and establishes the phase's resistance network based on the connection relationships between adjacent nodes. The local equivalent resistance of adjacent nodes is converted into the edge resistance or edge conductance of the corresponding connection edge. A unit test current or unit test voltage is applied between the feed terminal and return terminal of the phase to solve for the node potential distribution, and the equivalent resistance of the phase is obtained based on the ratio of port voltage to port current. For regular strip-shaped partitions, equivalent cumulative calculations can also be performed according to the series and parallel directions of nodes within the phase. This yields the equivalent phase resistances of phases A, B, and C, which are then combined to form an equivalent phase resistance matrix.
[0023] In some implementations, the method of using the deviation between the predicted and measured temperatures at discrete temperature measurement points as feedback to recursively correct the polynomial coefficients of the thermal parameters in the model includes: Calculate the temperature error based on the predicted temperature and the measured temperature at each discrete temperature measurement point; Based on the temperature error and the sensitivity of the predicted temperature to the polynomial coefficients, an adaptive parameter update method is used to perform online recursive correction of the polynomial coefficients.
[0024] A set of temperature measurement points, consisting of K sensors (such as NTC thermistors or PT100 patch sensors), is arranged at specific physical coordinates on the surface of the carbon nanotube heating film. The preferred range for K is 4 to 16 points; an exemplary arrangement is 8 uniformly distributed points. Within each control cycle or a set temperature sampling window, such as a 100ms cycle, the measured temperature values at these 8 locations are extracted in real time and subtracted from the predicted temperature output by the corresponding coordinate points of the state observation model to obtain a local temperature error vector of dimension 8×1. This error vector represents the degree of deviation of the current thermal parameter polynomial coefficients from the fit under the current operating conditions.
[0025] When using the recursive least squares algorithm, the controller uses the thermal parameter polynomial coefficient vector as the parameter vector to be updated, and the sensitivity of the predicted temperature at each discrete temperature measurement point to the parameter vector to be updated as the regression vector. The sensitivity is obtained using a finite difference method. Specifically, keeping the mapping relationship between real-time phase voltage, phase current, the temperature matrix of the previous control cycle, ambient temperature, and spatial current distribution unchanged, a preset small perturbation is applied sequentially to each polynomial coefficient to be corrected based on the current parameter values. The thermoelectric coupling state observation model is then rerun to obtain the predicted temperature after the perturbation. The difference between the predicted temperature after the perturbation and the predicted temperature before the perturbation is divided by the preset small perturbation amount to obtain the sensitivity value corresponding to that polynomial coefficient. This process is repeated for all polynomial coefficients to be corrected to obtain the sensitivity matrix.
[0026] The sensitivity row vectors corresponding to the K discrete temperature measurement points are combined into a sensitivity Jacobian matrix, and then input together with the temperature error vector into the recursive least squares update process to obtain a new thermal parameter polynomial coefficient vector.
[0027] Because the coefficients of the zeroth-order terms to higher-order terms differ significantly in magnitude, the recursive least squares update process can set corresponding scale normalization factors or parameter constraint ranges for polynomial coefficients of different orders. For example, different scale factors can be set for the coefficients of the zeroth-order, first-order, quadratic, and cubic terms, allowing coefficients of each order to participate in covariance matrix updates and parameter increment calculations at the same numerical magnitude. The controller calculates the parameter correction amount based on the sensitivity Jacobian matrix, temperature error vector, and forgetting factor, and then superimposes the parameter correction amount onto the current thermal parameter polynomial coefficient vector to obtain the updated thermal parameter polynomial coefficient vector.
[0028] S2, based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix, calculates the total power deviation and temperature uniformity deviation. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of each phase's physical region.
[0029] The actual total power is calculated by summing the products of the squares of the effective values of the three-phase currents and the corresponding phase resistances in the equivalent phase resistance matrix, and the difference between the target total power and the actual total power is taken as the total power deviation.
[0030] Based on the real-time temperature distribution matrix, the regional average temperature of the physical regions of phases A, B, and C is calculated respectively, and the dispersion of the three-phase regional average temperature relative to the global average temperature is calculated to obtain a temperature uniformity deviation metric. The target temperature uniformity corresponds to an allowable temperature non-uniformity deviation threshold in the control calculation; a smaller temperature uniformity deviation metric indicates a more uniform temperature across the three phase regions. When the temperature uniformity deviation metric exceeds the allowable temperature non-uniformity deviation threshold, the controller initiates temperature equalization differential compensation; when the temperature uniformity deviation metric is not greater than the allowable temperature non-uniformity deviation threshold, temperature equalization differential compensation is reduced or stopped.
[0031] The total power deviation is input into an incremental proportional-integral controller to calculate the fundamental voltage amplitude; based on the fundamental voltage amplitude and a three-phase sinusoidal reference signal with a fixed frequency and a phase difference of 120°, a three-phase fundamental voltage modulation signal is generated.
[0032] At each control moment, the maximum and minimum values of the three-phase fundamental voltage modulation signal are extracted in real time, and the sum of the maximum and minimum values is taken as the negative of the average value, which is used as the zero-sequence modulation optimization component.
[0033] When generating the three-phase temperature equalization differential mode compensation components, the controller determines the compensation direction for each phase based on the deviation direction of the average temperature of each phase region from the global average temperature. Phase regions with temperatures lower than the global average temperature correspond to enhanced compensation directions, while phase regions with temperatures higher than the global average temperature correspond to weakened compensation directions. When the temperature uniformity deviation metric exceeds the allowable temperature non-uniformity deviation threshold, the difference is multiplied by a preset temperature equalization compensation gain coefficient to obtain the temperature equalization compensation amplitude. When the temperature uniformity deviation metric is not greater than the allowable temperature non-uniformity deviation threshold, the temperature equalization compensation amplitude is reduced to a preset lower limit or set to zero. Based on the temperature equalization compensation amplitude, the compensation direction of each phase, and the synchronous electrical angle, the controller generates three-phase initial compensation voltage components corresponding to phases A, B, and C, respectively. The controller then performs instantaneous zero-sum processing on these three-phase initial compensation voltage components, ensuring that the algebraic sum of the processed phase A, phase B, and phase C compensation voltage components is zero at the same moment, thus forming the three-phase temperature equalization differential mode compensation components.
[0034] In some implementations, the calculation of total power deviation and temperature uniformity deviation metrics based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix, includes: The real-time estimated total power is calculated based on the effective value of the real-time three-phase current and the equivalent resistance of each phase, and the total power deviation is obtained based on the difference between the target total power and the real-time estimated total power. The regional average temperature of each phase physical region is calculated based on the real-time temperature distribution matrix, and the temperature uniformity deviation metric is obtained based on the degree of dispersion of the regional average temperature of each phase relative to the global average temperature.
[0035] The total power calculation process relies on a high-frequency ADC sampling channel, with a sampling frequency preferably between 10kHz and 20kHz, such as 16kHz, to acquire the transient currents of phases A, B, and C. A full-cycle root mean square algorithm is used to continuously extract the effective values of the three-phase currents. These values are then combined with the resistance values of each phase in the current iterative equivalent phase resistance matrix. The real-time total power is calculated by multiplying the square of the effective current of each phase by its resistance and summing the results. The target total power is typically set to a linearly adjustable range of 0% to 100% of the rated total capacity of the heating film. For example, the system's full-load rated output is set to 2000W. The difference between these two values in real time generates the total power deviation in watts (W), which serves as the closed-loop control input source for the power outer loop.
[0036] When extracting temperature uniformity characteristics, based on the factory physical wiring grid boundaries, the aforementioned 100×100 real-time temperature distribution matrix is divided into three subsets in the spatial domain. The arithmetic mean of all micro-nodes in the three independent heating regions (A-phase, B-phase, and C-phase) is calculated to obtain the real-time average temperature of each phase region. The sum of the three values divided by 3 yields the global average temperature of the three phases. The square root of the sum of the squares of the differences between the average temperature of each region and the global average temperature is then calculated, divided by 3, to obtain the temperature uniformity deviation metric. This value represents the degree of spatial imbalance in the three-phase thermal field of the carbon nanotube film.
[0037] In some implementations, the generation of the three-phase fundamental voltage modulation signal based on the total power deviation through a closed-loop regulator includes: The total power deviation is input into the proportional-integral controller to calculate the fundamental voltage amplitude. Based on the fundamental voltage amplitude and the synchronous electrical angle, a three-phase fundamental voltage modulation signal with a phase difference of 120° is generated.
[0038] The closed-loop regulator employs a digital proportional-integral controller with integral anti-saturation function. Since the dimension of the total power deviation at the input is W, and the dimension of the fundamental voltage amplitude at the output is V, the proportional gain coefficient can be set from 0.05 to 0.5 V / W, for example, 0.15 V / W in a system with a rated power of 2000W; the integral gain coefficient can be set from 0.5 to 5 V / (W·s), for example, 2.0 V / (W·s). During control operations, the controller clamps and limits the integral term. Specifically, when the output voltage amplitude reaches a preset upper or lower limit, it restricts the integral term from accumulating further along the saturation direction to suppress integral saturation caused by a step change in the target power. The upper limit of the fundamental voltage amplitude is preset based on the DC bus voltage and the linear modulation range; the upper limit of the fundamental phase voltage amplitude is set to not exceed 1 / 3 of the DC bus voltage. The fundamental voltage amplitude output by the closed-loop regulator is used to characterize the voltage modulation amplitude required to meet the target total heating power.
[0039] In the three-phase spatial modulation signal construction stage, the phase angle of the power grid is identified through a fully digital software phase-locked loop (SPLL), or the synchronization electrical angle is generated by an internal timer in off-grid pure AC heating mode. The AC heating output frequency can be fixed at 50Hz or configured with frequency conversion within the 30-100Hz frequency band. In each control cycle, the microcontroller multiplies the generated fundamental voltage amplitude with the cosine, minus 120° cosine, and plus 120° cosine values of the synchronization electrical angle, respectively, thereby constructing a three-phase fundamental voltage modulation wave control signal sequence with a 120° absolute electrical phase angle and completely symmetrical spatial amplitude.
[0040] In some implementations, the zero-sequence modulation optimized component is generated based on the principle of inverting the average value of the three-phase fundamental extreme values, including: The three-phase fundamental voltage modulation signal is compared in real time during each control cycle, and the maximum and minimum envelope values at the current moment are extracted. The maximum envelope value is the maximum value in the three-phase fundamental voltage modulation signal, and the minimum envelope value is the minimum value in the three-phase fundamental voltage modulation signal. The zero-sequence modulation optimized component is generated by summing the maximum and minimum envelope values and taking the negative of the average.
[0041] Taking the commonly used 10kHz inverter PWM switching frequency as an example, in each discrete control cycle with an execution length of 100μs, the instantaneous per-unit amplitude of the three-phase fundamental voltage modulation signal of the current cycle is synchronously identified and compared. Within one operation step, the maximum value among the three is found and assigned to the maximum value variable, while the minimum value is found and assigned to the minimum value variable.
[0042] After acquiring the real-time envelope extrema, the controller sums the maximum and minimum values, divides by 2, and then takes the inverse of the result to generate the zero-sequence modulation optimization component. In terms of the electrical physical control mechanism, this common-mode zero-sequence component injection logic is equivalent to the implicitly generated zero-sequence reference voltage in standard space vector pulse width modulation (SVPWM). It can transform the original pure sine wave envelope into a saddle-shaped wave in the digital domain. This approach, without altering the relative difference of the three-phase line voltages and the current topology of the heating phase, helps improve the linear modulation utilization rate of the three-phase inverter's DC bus voltage and reduces the risk of clipping when the modulation wave approaches the limiting boundary, thus adapting to the driving requirements of high-power heating applications using carbon nanotube films.
[0043] In some implementations, the three-phase temperature equilibrium differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of each phase's physical region, including: The target heating phase is determined based on the regional average temperature of each phase's physical region, and the compensation direction of each phase is determined based on the deviation direction of the regional average temperature of each phase from the global average temperature. Based on the temperature uniformity deviation metric, the temperature equalization compensation gain coefficient, and the synchronous electrical angle, three-phase compensation voltage components corresponding to phases A, B, and C are generated respectively, and the algebraic sum of the three-phase compensation voltage components is zero.
[0044] For heating compensation to address uneven heating of the heating film, the judgment is based on the average temperature of each phase's physical region. For example, assuming that phase A measures 35℃, phase B measures 40℃, and phase C measures 42℃ in the current acquisition cycle, the controller can identify phase A, which has the lowest temperature, as the compensation target and set the compensation direction for phase A to increase the equivalent modulation amplitude. For phases B and C, which have higher temperatures, the direction is set to decrease or weaken the equivalent modulation amplitude based on their deviation from the global average temperature. If the state machine identifies phase B or C as being in the lowest temperature state, then phase B or C is correspondingly identified as the compensation target.
[0045] The three-phase temperature equalization differential compensation component is constructed by subtracting the allowable temperature non-uniformity deviation threshold corresponding to the target temperature uniformity from the current temperature uniformity deviation measurement, taking the non-negative part of the difference, and then multiplying it by a preset temperature equalization compensation gain coefficient, preferably 1.0 to 2.0 V / ℃, for example, 1.5 V / ℃, to obtain the temperature equalization compensation amplitude. Three initial compensation coefficients are generated based on the deviations of the average temperatures of phases A, B, and C from the global average temperature. Specifically, the controller subtracts the average temperature of each phase region from the global average temperature to obtain the temperature compensation deviation for each phase; the temperature compensation deviation corresponding to low temperature is positive, and the temperature compensation deviation corresponding to high temperature is negative. The controller normalizes the sum of the absolute values of the temperature compensation deviations of each phase to obtain the initial compensation coefficients for phases A, B, and C; when the temperature compensation deviation is within the preset temperature dead zone range, the corresponding initial compensation coefficient is set to zero.
[0046] Subsequently, the initial compensation coefficients of each phase, the temperature equalization compensation amplitude, and the synchronous three-phase sinusoidal reference signals that are in the same frequency and phase as the fundamental waves of phases A, B, and C are multiplied to generate the initial compensation voltage components of phase A, phase B, and phase C. Finally, at each control moment, the instantaneous average value of the three initial compensation voltage components is calculated, and this instantaneous average value is subtracted from the initial compensation voltage components of each phase to obtain the compensation voltage components of phase A, phase B, and phase C. Through the above instantaneous zero-sum processing, the algebraic sum of the three-phase compensation voltage components at the same control moment is zero, belonging to the three-phase differential mode compensation quantity, which can be applied to the modulation reference signal of the corresponding phase respectively.
[0047] S3 performs algebraic superposition of the three-phase fundamental voltage modulation signal with the corresponding phase compensation components in the zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component, and performs DC bus voltage limiting constraint to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
[0048] The zero-sequence modulation optimization components are superimposed onto the three-phase fundamental voltage modulation signals, and the A-phase compensation voltage components, B-phase compensation voltage components, and C-phase compensation voltage components in the three-phase temperature equalization differential mode compensation components are superimposed onto the fundamental voltage modulation signals of the corresponding phases, thus completing the algebraic synthesis of the three-phase modulation reference signals.
[0049] The DC bus voltage value is acquired in real time through an isolation voltage sensor, and the superimposed three-phase voltage signal is input into the saturation limiting algorithm function. The upper threshold of the limiting algorithm function is set to 1 / 2 of the DC bus voltage value, and the lower threshold is set to -1 / 2 of the DC bus voltage value. When the amplitude of the superimposed signal exceeds the upper and lower thresholds, it is truncated to the boundary value to generate a three-phase modulated wave signal.
[0050] The three-phase modulated wave signal is normalized and mapped according to the DC bus voltage and the timer counting period to obtain the corresponding comparison register write value, which is then written into the comparison register of the microcontroller's internal advanced control timer. A center-aligned counting mode is configured in the advanced control timer, using the timer count value to form an equivalent high-frequency triangular carrier wave. The comparison register write value is compared with the timer count value; a high level is output when the comparison register write value is greater than the current timer count value, and a low level is output when the comparison register write value is not greater than the current timer count value. Combined with a preset dead-time generator, six complementary pulse width modulation drive signals with dead-time protection are generated, i.e., the PWM drive signals controlling the three-phase inverter. Taking phase A as an example, the complementary PWM drive timing with dead-time for the upper and lower bridge arms of the inverter is as follows: Figure 2 As shown.
[0051] In this application, a carbon nanotube heating film PWM power adaptive control system is disclosed, the system comprising: The correction module is used to obtain the real-time three-phase current and voltage of the three-phase inverter of the three-phase carbon nanotube heating film and the temperature values of multiple discrete temperature measurement points. It uses an adaptive thermoelectric coupling state observation model to calculate the real-time temperature distribution matrix and the equivalent phase resistance matrix. The deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point is used as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model. The calculation module is used to calculate the total power deviation and temperature uniformity deviation based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B and C respectively, and are constructed based on the temperature deviation state of the physical region of each phase. The control module is used to algebraically superimpose the three-phase fundamental voltage modulation signal with the corresponding phase compensation components of the zero-sequence modulation optimization component and the three-phase temperature equalization differential mode compensation component, and perform DC bus voltage limiting constraint to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
[0052] In some implementations, the calculation of the real-time temperature distribution matrix and equivalent phase resistance matrix using an adaptive thermoelectric coupling state observation model includes: A thermoelectric coupling state equation is constructed, and the current temperature distribution matrix is iteratively calculated based on the temperature matrix of the previous control cycle, the real-time phase voltage and phase current, and the spatial current distribution mapping relationship. Based on the temperature resistance characteristics of carbon nanotube materials, the local equivalent resistance of each discrete node is calculated using a polynomial fitting model. Based on the electrical connection topology of discrete nodes within each phase physical region, network equivalent calculations are performed on the local equivalent resistance to obtain an equivalent phase resistance matrix containing three-phase transient resistance values.
[0053] In some implementations, the method of using the deviation between the predicted and measured temperatures at discrete temperature measurement points as feedback to recursively correct the polynomial coefficients of the thermal parameters in the model includes: Calculate the temperature error based on the predicted temperature and the measured temperature at each discrete temperature measurement point; Based on the temperature error and the sensitivity of the predicted temperature to the polynomial coefficients, an adaptive parameter update method is used to perform online recursive correction of the polynomial coefficients.
[0054] In some implementations, the calculation of total power deviation and temperature uniformity deviation metrics based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix, includes: The real-time estimated total power is calculated based on the effective value of the real-time three-phase current and the equivalent resistance of each phase, and the total power deviation is obtained based on the difference between the target total power and the real-time estimated total power. The regional average temperature of each phase physical region is calculated based on the real-time temperature distribution matrix, and the temperature uniformity deviation metric is obtained based on the degree of dispersion of the regional average temperature of each phase relative to the global average temperature.
[0055] In some implementations, the generation of the three-phase fundamental voltage modulation signal based on the total power deviation through a closed-loop regulator includes: The total power deviation is input into the proportional-integral controller to calculate the fundamental voltage amplitude. Based on the fundamental voltage amplitude and the synchronous electrical angle, a three-phase fundamental voltage modulation signal with a phase difference of 120° is generated.
[0056] In some implementations, the zero-sequence modulation optimized component is generated based on the principle of inverting the average value of the three-phase fundamental extreme values, including: The three-phase fundamental voltage modulation signal is compared in real time during each control cycle, and the maximum and minimum envelope values at the current moment are extracted. The maximum envelope value is the maximum value in the three-phase fundamental voltage modulation signal, and the minimum envelope value is the minimum value in the three-phase fundamental voltage modulation signal. The zero-sequence modulation optimized component is generated by summing the maximum and minimum envelope values and taking the negative of the average.
[0057] In some implementations, the three-phase temperature equilibrium differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of each phase's physical region, including: The target heating phase is determined based on the regional average temperature of each phase's physical region, and the compensation direction of each phase is determined based on the deviation direction of the regional average temperature of each phase from the global average temperature. Based on the temperature uniformity deviation metric, the temperature equalization compensation gain coefficient, and the synchronous electrical angle, three-phase compensation voltage components corresponding to phases A, B, and C are generated respectively, and the algebraic sum of the three-phase compensation voltage components is zero.
[0058] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A PWM power adaptive control method for a carbon nanotube heating film, characterized in that, Includes the following steps: The real-time three-phase current and voltage of the three-phase inverter with a three-phase carbon nanotube heating film and the temperature values of multiple discrete temperature measurement points are obtained. The real-time temperature distribution matrix and equivalent phase resistance matrix are calculated using an adaptive thermoelectric coupling state observation model. The deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point is used as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model. Based on the target total power and target temperature uniformity, the total power deviation and temperature uniformity deviation are calculated by combining the real-time temperature distribution matrix and the equivalent phase resistance matrix. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of the physical region of each phase. The three-phase fundamental voltage modulation signal is algebraically superimposed with the corresponding phase compensation components in the zero-sequence modulation optimization component and the three-phase temperature equalization differential mode compensation component, and the DC bus voltage is limited to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
2. The method according to claim 1, characterized in that, The calculation of the real-time temperature distribution matrix and equivalent phase resistance matrix using the adaptive thermoelectric coupling state observation model includes: A thermoelectric coupling state equation is constructed, and the current temperature distribution matrix is iteratively calculated based on the temperature matrix of the previous control cycle, the real-time phase voltage and phase current, and the spatial current distribution mapping relationship. Based on the temperature resistance characteristics of carbon nanotube materials, the local equivalent resistance of each discrete node is calculated using a polynomial fitting model. Based on the electrical connection topology of discrete nodes within each phase physical region, network equivalent calculations are performed on the local equivalent resistance to obtain an equivalent phase resistance matrix containing three-phase transient resistance values.
3. The method according to claim 1, characterized in that, The method of using the deviation between the predicted temperature and the measured temperature at discrete temperature measurement points as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model includes: Calculate the temperature error based on the predicted temperature and the measured temperature at each discrete temperature measurement point; Based on the temperature error and the sensitivity of the predicted temperature to the polynomial coefficients, an adaptive parameter update method is used to perform online recursive correction of the polynomial coefficients.
4. The method according to claim 2, characterized in that, The calculation of total power deviation and temperature uniformity deviation metrics based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix, includes: The real-time estimated total power is calculated based on the effective value of the real-time three-phase current and the equivalent resistance of each phase, and the total power deviation is obtained based on the difference between the target total power and the real-time estimated total power. The regional average temperature of each phase physical region is calculated based on the real-time temperature distribution matrix, and the temperature uniformity deviation metric is obtained based on the degree of dispersion of the regional average temperature of each phase relative to the global average temperature.
5. The method according to claim 1, characterized in that, The generation of a three-phase fundamental voltage modulation signal based on the total power deviation through a closed-loop regulator includes: The total power deviation is input into the proportional-integral controller to calculate the fundamental voltage amplitude. Based on the fundamental voltage amplitude and the synchronous electrical angle, a three-phase fundamental voltage modulation signal with a phase difference of 120° is generated.
6. The method according to claim 1, characterized in that, The zero-sequence modulation optimization component is generated based on the principle of inverting the average value of the three-phase fundamental extreme values, including: The three-phase fundamental voltage modulation signal is compared in real time during each control cycle, and the maximum and minimum envelope values at the current moment are extracted. The maximum envelope value is the maximum value in the three-phase fundamental voltage modulation signal, and the minimum envelope value is the minimum value in the three-phase fundamental voltage modulation signal. The zero-sequence modulation optimized component is generated by summing the maximum and minimum envelope values and taking the negative of the average.
7. The method according to claim 1, characterized in that, The three-phase temperature equilibrium differential mode compensation components correspond to phases A, B, and C, respectively, and are constructed based on the temperature deviation state of each phase's physical region, including: The target heating phase is determined based on the regional average temperature of each phase's physical region, and the compensation direction of each phase is determined based on the deviation direction of the regional average temperature of each phase from the global average temperature. Based on the temperature uniformity deviation metric, the temperature equalization compensation gain coefficient, and the synchronous electrical angle, three-phase compensation voltage components corresponding to phases A, B, and C are generated respectively, and the algebraic sum of the three-phase compensation voltage components is zero.
8. A PWM power adaptive control system for a carbon nanotube heating film, characterized in that, Includes the following modules: The correction module is used to obtain the real-time three-phase current and voltage of the three-phase inverter of the three-phase carbon nanotube heating film and the temperature values of multiple discrete temperature measurement points. It uses an adaptive thermoelectric coupling state observation model to calculate the real-time temperature distribution matrix and the equivalent phase resistance matrix. The deviation between the predicted temperature and the measured temperature value at the discrete temperature measurement point is used as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model. The calculation module is used to calculate the total power deviation and temperature uniformity deviation based on the target total power and target temperature uniformity, combined with the real-time temperature distribution matrix and equivalent phase resistance matrix. Based on the total power deviation, a three-phase fundamental voltage modulation signal is generated through a closed-loop regulator. The zero-sequence modulation optimization component and the three-phase temperature balance differential mode compensation component are calculated. The zero-sequence modulation optimization component is generated according to the principle of inverting the average value of the three-phase fundamental extreme values. The three-phase temperature balance differential mode compensation components correspond to phases A, B and C respectively, and are constructed based on the temperature deviation state of the physical region of each phase. The control module is used to algebraically superimpose the three-phase fundamental voltage modulation signal with the corresponding phase compensation components of the zero-sequence modulation optimization component and the three-phase temperature equalization differential mode compensation component, and perform DC bus voltage limiting constraint to generate a three-phase modulation wave signal. The three-phase modulation wave signal is then sent to the sinusoidal pulse width modulation unit to generate a PWM drive signal for controlling the three-phase inverter.
9. The system according to claim 8, characterized in that, The calculation of the real-time temperature distribution matrix and equivalent phase resistance matrix using the adaptive thermoelectric coupling state observation model includes: A thermoelectric coupling state equation is constructed, and the current temperature distribution matrix is iteratively calculated based on the temperature matrix of the previous control cycle, the real-time phase voltage and phase current, and the spatial current distribution mapping relationship. Based on the temperature resistance characteristics of carbon nanotube materials, the local equivalent resistance of each discrete node is calculated using a polynomial fitting model. Based on the electrical connection topology of discrete nodes within each phase physical region, network equivalent calculations are performed on the local equivalent resistance to obtain an equivalent phase resistance matrix containing three-phase transient resistance values.
10. The system according to claim 8, characterized in that, The method of using the deviation between the predicted temperature and the measured temperature at discrete temperature measurement points as feedback to perform online recursive correction of the polynomial coefficients of the thermal parameters in the model includes: Calculate the temperature error based on the predicted temperature and the measured temperature at each discrete temperature measurement point; Based on the temperature error and the sensitivity of the predicted temperature to the polynomial coefficients, an adaptive parameter update method is used to perform online recursive correction of the polynomial coefficients.
