Additive manufacturing heating power adaptive control method based on infrared thermal imaging

Abstract

The application relates to an additive manufacturing heating power self-adaptive control method based on infrared thermal imaging, which adopts an infrared thermal imaging camera to collect a global temperature field of a printing plane in real time, divides the temperature field into multiple heating sub-regions according to the spatial projection positions of curing lamp tubes, introduces an inter-region thermal coupling coefficient matrix to quantitatively separate the neighborhood thermal crosstalk suffered by each sub-region, extracts intrinsic temperature components reflecting only the heating contribution of each lamp tube, and constructs a real deviation signal after removing the coupling interference. The proportional-integral-derivative algorithm is used to independently calculate the power adjustment increment of each sub-region based on the intrinsic temperature deviation, the amplitude is corrected, and then the pulse width modulation driving signal is converted, so that the differential power self-adaptive adjustment of the lamp tubes at different spatial positions is realized. The scheme decouples the thermal crosstalk, so that the control decisions of each channel are independent of each other, the coupling oscillation caused by the shared heat conduction path of adjacent channels is avoided, and the printing surface temperature uniformity and product forming consistency are improved.

Description

Technical Field

[0001] This invention relates to the field of additive manufacturing process control technology, and in particular to an adaptive control method for additive manufacturing heating power based on infrared thermal imaging. Background Technology

[0002] In additive manufacturing, curing lamp arrays are arranged spatially adjacently above the printing plane to heat and cure the deposited material layer. The temperature uniformity of each area of ​​the printing plane directly determines the consistency of the material curing degree, thus affecting the dimensional accuracy, mechanical strength, and molding quality of the product. However, due to factors such as installation tolerances, differences in aging levels, and thermal radiation boundary attenuation inherent in the lamp array itself, spatial temperature gradients inevitably occur on the printing plane. If differentiated power compensation cannot be applied to address temperature deviations in different spatial locations, localized overheating or under-curing will occur, severely reducing the consistency of products between batches.

[0003] Existing temperature control solutions for additive manufacturing typically employ point-based non-contact infrared temperature sensors, using temperature samples from several discrete points as feedback signals. A proportional-integral-derivative (PI-DI) algorithm is then used to uniformly adjust the overall power of the lamp. This approach only acquires local information from a limited number of temperature measurement points, failing to characterize the spatial distribution differences of the overall temperature field and hindering independent power compensation for different spatial regions. Furthermore, even if the sensor is upgraded to an infrared thermal imaging device to obtain the overall temperature and control each region independently, the lack of complete insulation between heating sub-regions allows heat to diffuse laterally along the material matrix to adjacent regions, creating thermal crosstalk. This means that the measured temperature in each sub-region is actually the superposition of its intrinsic temperature generated by lamp heating and the extrinsic temperature diffused from neighboring regions. Under these conditions, if the mixed temperature is simply subtracted from the target temperature as the control deviation signal, controllers in adjacent channels may misjudge the thermal crosstalk, leading to alternating overcompensation and forming coupled oscillating loops. Ultimately, this causes the temperature field to oscillate continuously across multiple control cycles, making it difficult to converge to a steady state.

[0004] Therefore, an optimized adaptive control scheme for additive manufacturing heating power based on infrared thermal imaging is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide an adaptive control method for heating power in additive manufacturing based on infrared thermal imaging, comprising: S1. The infrared thermal imaging camera collects the temperature field of the printed plane to obtain the original infrared thermal image frame, which contains the infrared radiation intensity value of each pixel. S2, based on the material surface emissivity parameters, performs temperature calibration conversion on the original infrared thermal image frame to obtain a calibrated temperature distribution map; S3. Based on the spatial projection position of the curing lamp, perform multi-regional temperature field deviation spatial mapping on the calibration temperature distribution map to obtain the spatial temperature deviation matrix. S4, perform feedback-based power adjustment increment estimation on the deviation values ​​of each sub-region in the spatial temperature deviation matrix to obtain the spatial power adjustment increment vector; S5, superimpose the spatial power adjustment increment vector with the reference heating power configuration, and perform safety threshold limiting correction to obtain the adaptive heating power instruction set; S6, based on the pulse width modulation characteristics of the curing lamp driver hardware, performs duty cycle conversion and signal encoding on the adaptive heating power instruction set to generate a power drive signal sequence, wherein the power drive signal sequence is used to drive the curing lamps at each spatial position to achieve differentiated adaptive power adjustment.

[0006] Compared with existing technologies, this application proposes an adaptive control method for heating power in additive manufacturing based on infrared thermal imaging. It uses an infrared thermal imaging camera to replace traditional point-based temperature sensors, acquiring the full-domain temperature field of the additive manufacturing printing plane in real time. After material emissivity calibration, complete spatial temperature distribution information is obtained. Based on this, the temperature field is divided into multiple independent heating sub-regions according to the spatial projection position of the curing lamp array. An inter-regional thermal coupling coefficient matrix is ​​introduced to quantify and explicitly remove the neighboring thermal crosstalk experienced by each sub-region, decomposing the mixed temperature signal into intrinsic temperature components that only reflect the heating contribution of each lamp. Based on this, a true temperature deviation signal after removing coupling interference is constructed. Subsequently, a proportional-integral-derivative control algorithm is used to independently calculate the power adjustment increment based on the intrinsic temperature deviation of each sub-region. After safety threshold limiting correction, this is converted into a pulse width modulation drive signal, realizing differentiated adaptive power adjustment for curing lamps at different spatial positions. This solution achieves spatially differentiated independent zone control through global temperature sensing. At the same time, the thermal crosstalk decoupling mechanism makes the decision signals of each channel controller independent, avoiding coupling oscillations caused by adjacent channels sharing the heat conduction path. This allows the global temperature field to converge to the target distribution at a faster speed and be stably maintained within a small fluctuation range, improving the temperature uniformity of the printed surface and the consistency of the product molding quality. Attached Figure Description

[0007] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0008] Figure 1This is a flowchart of an additive manufacturing heating power adaptive control method based on infrared thermal imaging, according to an embodiment of this application. Figure 2 This is a schematic diagram of data flow for an adaptive control method of heating power in additive manufacturing based on infrared thermal imaging, according to an embodiment of this application. Figure 3 This is a flowchart illustrating the process of performing temperature calibration conversion on the original infrared thermal image frame to obtain a calibrated temperature distribution map based on the material surface emissivity parameter in an additive manufacturing heating power adaptive control method based on infrared thermal imaging, according to an embodiment of this application. Figure 4 This is a flowchart illustrating the spatial temperature deviation matrix obtained by spatially mapping a calibration temperature distribution map to a multi-region temperature field deviation based on the spatial projection position of a curing lamp tube, according to an embodiment of this application, in an additive manufacturing heating power adaptive control method based on infrared thermal imaging. Figure 5 This is a flowchart illustrating a method for adaptive control of heating power in additive manufacturing based on infrared thermal imaging, according to an embodiment of this application, which involves differential quantization of the average temperature vector of a sub-region with the target temperature value of the corresponding sub-region in a preset target temperature matrix to obtain a spatial temperature deviation matrix. Figure 6 This is a flowchart illustrating a method for adaptive control of heating power in additive manufacturing based on infrared thermal imaging, according to an embodiment of this application, which involves estimating the power adjustment increment vector by feedback based on the deviation values ​​of each sub-region in the spatial temperature deviation matrix. Figure 7 This is a flowchart illustrating an adaptive heating power control method for additive manufacturing based on infrared thermal imaging, according to an embodiment of this application. The method involves superimposing a spatial power adjustment increment vector with a reference heating power configuration and performing a safety threshold limiting correction to obtain an adaptive heating power instruction set. Detailed Implementation

[0009] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0010] Existing additive manufacturing temperature control solutions mostly employ point-type non-contact temperature sensors for localized temperature sampling, failing to acquire information on the overall temperature spatial distribution of the printed surface and hindering the achievement of independent, differentiated compensation for lamp power in different spatial regions. Even with the introduction of area-array infrared thermal imaging equipment for independent zone control, the continuous heat conduction medium formed by the printed material layer between each heating sub-region and the thermal diffusion from neighboring lamps cause the measured temperature to mix intrinsic temperature with extrinsic crosstalk components. This leads to alternating overcompensation and oscillating divergence in each channel controller due to coupling misjudgments, making it difficult for the temperature field to converge to a steady state. Therefore, this application proposes an adaptive control method for additive manufacturing heating power based on infrared thermal imaging. This method first achieves uniform and precise control of the temperature field of the printed surface through the coordinated operation of global temperature sensing, thermal crosstalk decoupling, and multi-channel independent closed-loop control. Specifically, firstly, an infrared thermal imaging camera is used to collect the global temperature field of the printing plane. After calibration of the material surface emissivity parameters, a calibrated temperature distribution map is obtained. Then, based on the spatial projection position of the curing lamp, the temperature field is divided into multiple heating sub-regions. A pre-calibrated inter-regional thermal coupling coefficient matrix is ​​introduced to quantify and remove the thermal crosstalk from the neighboring regions of each sub-region, extracting the intrinsic temperature that only reflects the heating contribution of the lamp itself. Based on this, a true spatial temperature deviation matrix is ​​constructed to remove coupling interference. Next, based on the proportional-integral-derivative control algorithm, the power adjustment increment is independently calculated for the intrinsic temperature deviation of each sub-region. The increment is superimposed with the reference power and a safety threshold limiting correction is performed. Finally, the adaptive power command is converted by duty cycle and encoded to generate a driving signal sequence, which drives the curing lamps at each spatial position to achieve differentiated adaptive power adjustment. This allows the global temperature field to quickly converge to the target distribution and be stably maintained within a small fluctuation range, improving the consistency of the product molding quality.

[0011] Figure 1 This is a flowchart of an additive manufacturing heating power adaptive control method based on infrared thermal imaging, according to an embodiment of this application. Figure 2 This is a schematic diagram of the data flow of an additive manufacturing heating power adaptive control method based on infrared thermal imaging, according to an embodiment of this application. Figure 1 and Figure 2As shown, an additive manufacturing heating power adaptive control method based on infrared thermal imaging according to an embodiment of this application includes the following steps: S1, acquiring the full-domain temperature field of the printing plane using an infrared thermal imaging camera to obtain an original infrared thermal image frame, wherein the original infrared thermal image frame contains the infrared radiation intensity value of each pixel; S2, performing temperature calibration conversion on the original infrared thermal image frame based on the material surface emissivity parameter to obtain a calibration temperature distribution map; S3, performing multi-region temperature field deviation spatial mapping on the calibration temperature distribution map based on the spatial projection position of the curing lamp to obtain a spatial temperature deviation matrix; S4, performing feedback-based power adjustment increment estimation on the deviation values ​​of each sub-region in the spatial temperature deviation matrix to obtain a spatial power adjustment increment vector; S5, superimposing the spatial power adjustment increment vector with the reference heating power configuration and performing safety threshold limiting correction to obtain an adaptive heating power instruction set; S6, performing duty cycle conversion and signal encoding on the adaptive heating power instruction set based on the pulse width modulation characteristics of the curing lamp driving hardware to generate a power driving signal sequence, wherein the power driving signal sequence is used to drive the curing lamps at each spatial position to achieve differentiated power adaptive adjustment.

[0012] Specifically, in step S1, an infrared thermal imaging camera is used to acquire the entire temperature field of the printed surface to obtain an original infrared thermal image frame, which contains the infrared radiation intensity value of each pixel. It should be noted that the temperature uniformity of the additive manufacturing printed surface directly affects the material curing quality. Traditional point-type non-contact infrared temperature sensors can only acquire temperature information at a limited number of discrete points, failing to obtain the overall temperature distribution across all spatial locations of the printed surface, making it difficult to provide a complete spatial temperature data foundation for subsequent multi-region differentiated power control. Therefore, the technical solution of this application first acquires the entire temperature field of the printed surface using an infrared thermal imaging camera to obtain an original infrared thermal image frame. Through the above processing, infrared radiation information of all spatial locations of the printed surface can be acquired at once using an array imaging method, providing complete two-dimensional temperature field data for subsequent zoned temperature extraction and deviation calculation based on the spatial projection position of the lamp tubes.

[0013] More specifically, in one particular example of this application, an infrared thermal imaging camera is mounted directly above the printing plane of the additive manufacturing equipment, with its optical axis perpendicular to the printing plane and its field of view covering the entire effective printing area. During each control cycle, the focal plane array detector of the infrared thermal imaging camera synchronously senses the infrared electromagnetic waves radiated from the material surface at various locations on the printing plane. Each pixel in the detector converts the received infrared radiation energy into a corresponding electrical signal, which, after analog-to-digital conversion, forms a two-dimensional data matrix containing the infrared radiation intensity values ​​of all pixels—the original infrared thermal image frame. The value of each pixel in this frame has a monotonic correspondence with the temperature of the material surface at its corresponding spatial location, and the spatial arrangement of the pixels forms a fixed geometric mapping with the physical coordinates of the printing plane. After acquisition, the original infrared thermal image frame is transmitted to the controller's data buffer via a data transmission interface for subsequent temperature calibration and conversion steps to read and process.

[0014] Specifically, in step S2, the original infrared thermal image frame is calibrated and converted to a calibrated temperature distribution map based on the material surface emissivity parameter. It should be noted that, given that the pixel values ​​in the original infrared thermal image frame acquired by the infrared thermal imaging camera are the radiation intensity signals of the detector response, rather than physical temperature values ​​that can be directly used for control feedback, and considering the differences in responsivity and random thermal noise interference between detector pixels, directly using the original radiation intensity data for subsequent deviation calculations without calibration and conversion will introduce measurement errors inconsistent with the actual temperature, causing the power control decision to deviate from the true temperature state. Therefore, the technical solution of this application further performs a temperature calibration and conversion on the original infrared thermal image frame based on the material surface emissivity parameter to obtain a calibrated temperature distribution map. Through the above processing, the original radiation intensity signal can be converted into calibration data characterizing the true physical temperature of each spatial location on the printed plane, providing an accurate temperature reference for subsequent multi-region temperature deviation spatial mapping.

[0015] Figure 3 This is a flowchart illustrating a method for adaptive control of heating power in additive manufacturing based on infrared thermal imaging, according to an embodiment of this application. The method involves temperature calibration transformation of an original infrared thermal image frame to obtain a calibrated temperature distribution map based on the material surface emissivity parameter. Figure 3 As shown, step S2 includes: S21, performing spatial domain noise reduction filtering on the original infrared thermal image frame to obtain a filtered thermal image frame; S22, performing focal plane array non-uniformity correction on the filtered thermal image frame based on the pre-calibrated and stored detector pixel gain coefficient matrix and bias coefficient matrix to obtain a corrected thermal image frame; S23, performing absolute temperature conversion calibration based on emissivity on the corrected thermal image frame based on the emissivity surface emissivity parameter to obtain a calibrated temperature distribution map.

[0016] In step S21, spatial domain noise reduction filtering is performed on the original infrared thermal image frame to obtain a filtered thermal image frame. It should be noted that due to the influence of the infrared detector's own thermal noise and environmental electromagnetic interference during operation, spatial random noise is inevitably superimposed on the original infrared thermal image frame. If directly used for subsequent temperature calibration and deviation calculation, the noise signal will be amplified and transmitted to the control decision-making stage. Therefore, the technical solution of this application further performs spatial domain noise reduction filtering on the original infrared thermal image frame to obtain a filtered thermal image frame. Through the above processing, pixel-level random noise interference can be suppressed, the spatial distribution trend characteristics of the temperature field can be preserved, and smooth input data can be provided for subsequent non-uniformity correction.

[0017] More specifically, in a concrete example of this application, a two-dimensional Gaussian filtering algorithm is used to perform spatial domain convolutional noise reduction on the original infrared thermal image frame. A two-dimensional Gaussian convolution kernel of size (2k+1)×(2k+1) is pre-constructed, where k is the kernel radius parameter. The weight values ​​at each position in the convolution kernel follow a two-dimensional Gaussian distribution with the kernel center as the mean, and the weights decrease as the distance from the center increases. During the filtering operation, the Gaussian convolution kernel is slid sequentially to each pixel position in the original infrared thermal image frame. Taking the current pixel as the center, the radiation intensity values ​​of each pixel in its neighborhood are multiplied one by one with the weight value of the corresponding position of the convolution kernel, and then summed. The weighted sum is used as the filtered output value at the center pixel position. After performing the above convolution operation pixel by pixel on the entire image, each pixel position obtains a radiation intensity value after neighborhood weighted smoothing. Spatially abrupt random noise is effectively suppressed, while the overall spatial distribution gradient information of the temperature field is preserved. Finally, the filtered thermal image frame is output for use in the next step.

[0018] In step S22, based on the pre-calibrated and stored detector pixel gain coefficient matrix and bias coefficient matrix, focal plane array non-uniformity correction is performed on the filtered thermal image frame to obtain a corrected thermal image frame. It should be noted that due to the inherent differences in responsivity among pixels in an infrared focal plane array detector during semiconductor manufacturing, even when facing a uniform radiation source, the amplitude of the electrical signal output by different pixels is inconsistent, forming fixed pattern noise. Without correction, this will lead to different reading deviations at different pixel positions for the same temperature. Therefore, the technical solution of this application further performs focal plane array non-uniformity correction on the filtered thermal image frame based on the pre-calibrated and stored detector pixel gain coefficient matrix and bias coefficient matrix to obtain a corrected thermal image frame. Through the above processing, the spatial fixed deviation introduced by the difference in responsivity of each pixel can be eliminated, ensuring that the surface of an object at the same temperature produces a consistent radiation intensity output value at all positions in the image.

[0019] More specifically, in a specific example of this application, a two-point linear correction method is used to perform pixel-by-pixel non-uniformity compensation on the filtered thermal image frame. During the device's factory calibration phase, the infrared thermal imaging camera is aimed at two uniform blackbody radiation sources with known temperatures to acquire images. Based on the response output values ​​of each pixel at the two reference temperature points, the gain correction coefficient and bias correction coefficient corresponding to each pixel are calculated through linear fitting, forming a gain coefficient matrix and a bias coefficient matrix of the same size as the image resolution, respectively, and stored in the controller's non-volatile storage area. In each control cycle of actual operation, the radiation intensity value of each pixel position in the filtered thermal image frame is read, and the corresponding gain coefficient and bias coefficient are retrieved from the storage area. Linear compensation calculation is then performed on each pixel according to the two-point correction formula:

[0020] Where C(x,y) is the corrected radiant intensity value at coordinate position (x,y) in the corrected thermal image frame, K(x,y) is the gain correction coefficient corresponding to the pixel position in the gain coefficient matrix, F(x,y) is the pixel radiant intensity value at the corresponding position in the filtered thermal image frame, and B(x,y) is the bias correction coefficient corresponding to the pixel position in the bias coefficient matrix. After the above pixel-by-pixel linear compensation is completed, the response characteristics of each pixel are normalized to a unified linear scale, and the corrected thermal image frame is output for use in the subsequent temperature conversion calibration step.

[0021] In step S23, based on the surface emissivity parameters of the printing material, the corrected thermal image frame is calibrated using an absolute temperature conversion based on emissivity to obtain a calibrated temperature distribution map. It should be noted that since the pixel values ​​in the corrected thermal image frame are still in the dimension of radiation intensity after linear compensation, rather than physical Celsius temperatures that can be directly used in temperature deviation calculations, and different printing materials have different surface emissivity, different materials radiate different infrared energy at the same temperature. Without introducing material emissivity parameters for correction, a true absolute temperature cannot be obtained. Therefore, the technical solution of this application further calibrates the corrected thermal image frame using an absolute temperature conversion based on emissivity parameters of the printing material to obtain a calibrated temperature distribution map. Through the above processing, the radiation intensity signal can be converted into a true Celsius temperature value corresponding to the physical state of the material, providing an accurate temperature reference for subsequent multi-region deviation calculations.

[0022] More specifically, in a particular example of this application, a nonlinear inverse solution model between radiation intensity and absolute temperature is established based on the Stefan-Boltzmann radiation law, and temperature conversion calculations are performed on each pixel in the corrected thermal image frame. Before the additive manufacturing equipment is put into use, the surface emissivity constant of the material used in the current printing process is pre-determined and written within the operating temperature range. During the temperature calibration stage of each control cycle, the radiation intensity value in the corrected thermal image frame is read pixel by pixel. Combined with the pre-set material surface emissivity parameters and the Stefan-Boltzmann constant, the radiation intensity is converted into absolute temperature according to the inverse radiation thermometry formula, and then converted into Celsius temperature.

[0023] Where T(x,y) is the actual Celsius temperature value at coordinate position (x,y) in the calibration temperature distribution map, in degrees Celsius; C(x,y) is the corrected radiant intensity value at the corresponding position in the corrected thermal image frame; ε is the surface emissivity constant of the current printing material, with a value between 0 and 1; and σ is the Stefan-Boltzmann constant, with a value of 5.67 × 10⁻⁶. -8 W / (m 2 ·K 4 ), where 273.15 is the fixed offset for converting Kelvin temperature to Celsius temperature. After performing the above nonlinear conversion on all pixel positions of the entire image, the value of each pixel represents the true physical temperature of its corresponding printed plane spatial position, ultimately forming a two-dimensional calibration temperature distribution map, which is used for subsequent multi-region temperature field deviation mapping steps based on the spatial projection position of the lamp tube.

[0024] Specifically, in step S3, based on the spatial projection position of the curing lamps, a multi-region temperature field deviation spatial mapping is performed on the calibration temperature distribution map to obtain a spatial temperature deviation matrix. It should be noted that, given that the calibration temperature distribution map is continuous temperature data covering the entire printing plane, and the curing lamp array in the additive manufacturing equipment is arranged spatially adjacently above the printing plane, each lamp only has independent power adjustment capability for its projected local area. If the overall temperature data is not spatially partitioned according to the lamp's physical layout and the deviation between each region and the target temperature is not calculated, an independent power control feedback signal cannot be generated for each lamp. Therefore, the technical solution of this application further performs a multi-region temperature field deviation spatial mapping on the calibration temperature distribution map based on the spatial projection position of the curing lamps to obtain a spatial temperature deviation matrix. Through the above processing, the overall temperature field information can be converted into a spatial deviation quantification result corresponding one-to-one with the physical control channels of each lamp, providing channel-by-channel control error input for subsequent independent power adjustment increment calculations in each sub-region.

[0025] Figure 4This is a flowchart illustrating the spatial temperature deviation matrix obtained by spatially mapping a calibration temperature distribution map to multi-region temperature field deviations based on the spatial projection position of a curing lamp, according to an embodiment of this application for an adaptive control method of additive manufacturing heating power based on infrared thermal imaging. Figure 4 As shown, step S3 includes: S31, based on the pixel boundary range of each heating sub-region determined by the spatial topological geometric parameters of the curing lamp tube, performing binary mask cropping and region-by-region partitioning extraction on the calibration temperature distribution map to obtain a set of sub-region temperature maps; S32, performing average dimensionality reduction statistics on all effective pixel temperature values ​​within the coverage area of ​​each sub-region in the set of sub-region temperature maps to obtain the sub-region average temperature vector; S33, performing differential quantization between the sub-region average temperature vector and the corresponding sub-region target temperature value in the preset target temperature matrix to obtain the spatial temperature deviation matrix.

[0026] In step S31, based on the pixel boundary range of each heating sub-region determined by the spatial topological geometric parameters of the curing lamp, the calibration temperature distribution map is subjected to binary mask cropping and region-by-region extraction to obtain a set of sub-region temperature maps. It should be noted that since the calibration temperature distribution map is a continuous two-dimensional temperature matrix covering the entire printing plane, and each curing lamp only has independent heating capability for the local area directly below its projection, the global temperature data needs to be divided into local temperature sub-images corresponding to each control channel according to the lamp's physical spatial layout. Based on this, the technical solution of this application further performs binary mask cropping and region-by-region extraction on the calibration temperature distribution map based on the pixel boundary range of each heating sub-region determined by the spatial topological geometric parameters of the curing lamp to obtain a set of sub-region temperature maps. Through the above processing, the global temperature field can be decomposed into multiple independent local temperature data blocks according to the lamp's spatial projection position, providing spatially isolated data input for subsequent region-by-region average temperature extraction.

[0027] More specifically, in a particular example of this application, the curing lamp array consists of N lamps arranged at fixed intervals above the printing plane. The spatial topological geometric parameters of each lamp, such as its center position, effective irradiation width, and length coverage, are determined after the equipment is assembled. Based on the spatial resolution of the infrared thermal imaging camera and the coordinate mapping relationship between the camera's field of view and the printing plane, the physical projection area of ​​each lamp on the printing plane is converted into the corresponding pixel coordinate boundary range in the calibration temperature distribution map. For the p-th heating sub-region, a binary mask matrix of the same size as the calibration temperature distribution map is constructed. The coordinate positions in this mask matrix that fall within the projection pixel boundary range of the p-th lamp are assigned a value of 1, and the remaining coordinate positions are assigned a value of 0. The calibration temperature distribution map is multiplied pixel-by-pixel by the binary mask matrix of the p-th sub-region. Pixels with non-zero values ​​in the product result are the effective temperature data within the projection coverage range of that lamp, while pixels with zero values ​​are masked out, thereby extracting the local temperature map corresponding to the p-th sub-region. Perform the mask clipping and extraction operations on all N heating sub-regions in sequence, and assemble the local temperature maps of all sub-regions in spatial index order. Finally, output the set of sub-region temperature maps for use in the subsequent mean temperature calculation step.

[0028] In step S32, the average dimensionality reduction statistics of all effective pixel temperature values ​​within the coverage area of ​​each sub-region in the sub-region temperature map set are performed to obtain the sub-region average temperature vector. It should be noted that since each sub-region temperature map contains temperature data of hundreds to thousands of pixels, directly using pixel-level data as control feedback input would result in excessively high control dimensionality, and residual noise from a single pixel could still affect decision accuracy. Therefore, it is necessary to reduce the high-dimensional temperature data within each region to a single scalar to characterize the overall thermal state of that region. Based on this, the technical solution of this application further performs average dimensionality reduction statistics on all effective pixel temperature values ​​within the coverage area of ​​each sub-region in the sub-region temperature map set to obtain the sub-region average temperature vector. Through the above processing, the spatial temperature distribution of the region corresponding to each lamp tube can be compressed into a representative average temperature, providing a dimensionally matched control variable for subsequent region-by-region difference comparison with the target temperature.

[0029] More specifically, in a concrete example of this application, for each sub-region temperature map in the set of sub-region temperature maps, the coordinates of all valid pixels within the mask area of ​​that sub-region are traversed, the total number of valid pixels is counted, and the Celsius temperature values ​​corresponding to all valid pixel positions are arithmetically summed. Then, the sum is divided by the total number of valid pixels to obtain the arithmetic mean temperature value of that sub-region. This operation can be expressed as:

[0030] in, This represents the value of the p-th element in the sub-region average temperature vector, i.e., the average temperature value of the p-th heated sub-region, in degrees Celsius. Let p be the total number of valid pixels in the p-th sub-region. Let be the set of valid pixel coordinates covered by the p-th sub-region. Let be the Celsius temperature value of the p-th sub-region in the set of sub-region temperature maps at coordinate position (x, y). Perform the above mean calculation sequentially on all N sub-regions, and arrange the resulting N average temperature values ​​into an N-dimensional vector according to spatial index order. This vector is the sub-region average temperature vector, used for subsequent deviation calculations from the preset target temperature.

[0031] In step S33, the average temperature vector of the sub-region is differentially quantized with the corresponding target temperature value of the sub-region in the preset target temperature matrix to obtain the spatial temperature deviation matrix. It should be noted that, since the average temperature vector of the sub-region only represents the current actual temperature state of the area covered by each lamp, while the control decision requires the deviation signal between the actual temperature and the desired temperature of each area, only by obtaining a clear quantized deviation value can it be determined whether the power of each lamp should be increased or decreased, and by how much. Based on this, the technical solution of this application further differentially quantizes the average temperature vector of the sub-region with the corresponding target temperature value of the sub-region in the preset target temperature matrix to obtain the spatial temperature deviation matrix. Through the above processing, the temperature state of each sub-region can be converted into a control error signal with direction and amplitude information, providing a channel-by-channel deviation input for the subsequent proportional-integral-derivative control algorithm.

[0032] Figure 5 This is a flowchart illustrating the spatial temperature deviation matrix obtained by spatially mapping a calibration temperature distribution map to multi-region temperature field deviations based on the spatial projection position of a curing lamp, according to an embodiment of this application for an adaptive control method of additive manufacturing heating power based on infrared thermal imaging. Figure 5 As shown, step S33 includes: S331, based on the inter-regional thermal coupling coefficient matrix obtained in advance through steady-state thermal calibration experiments, quantifying the neighborhood thermal crosstalk contribution of each channel temperature value in the sub-region average temperature vector to obtain the thermal crosstalk contribution vector; S332, based on the neighborhood thermal diffusion component encoded by each channel in the thermal crosstalk contribution vector, decoupling and isolating the intrinsic temperature of each channel in the sub-region average temperature vector to obtain the decoupled intrinsic temperature vector; S333, calculating the deviation and reconstructing the spatial layout between the target temperature of each sub-region in the preset target temperature matrix and the corresponding intrinsic temperature in the decoupled intrinsic temperature vector to obtain the spatial temperature deviation matrix.

[0033] In step S331, based on the inter-regional thermal coupling coefficient matrix obtained in advance through steady-state thermal calibration experiments, the thermal crosstalk contribution vector is obtained by quantifying the neighborhood thermal crosstalk contribution of each channel temperature value in the sub-region average temperature vector. It should be noted that, given that the heating sub-regions in the additive manufacturing printing plane are not completely separated by insulating material, but rather form a continuous heat conduction medium through the printing material layer and the air layer, when a lamp heats its covered area with high power, the heat diffuses laterally along the material matrix to the adjacent sub-regions, forming inter-regional thermal crosstalk. This means that the measured average temperature of each sub-region is actually the superposition of the intrinsic temperature component generated by its own lamp heating and the non-intrinsic temperature component diffused by heat conduction from the neighboring lamps. If the simple difference between this mixed temperature and the target temperature is directly used as the control deviation signal, the controllers of adjacent channels will generate alternating overcompensation due to thermal crosstalk misjudgment, forming a coupled oscillation loop, causing the temperature field to oscillate continuously in multiple control cycles and making it difficult to converge to a steady state. Based on this, the technical solution of this application further quantifies the neighborhood thermal crosstalk contribution of each channel temperature value in the sub-region average temperature vector by using the inter-region thermal coupling coefficient matrix obtained in advance through steady-state thermal calibration experiments, thereby obtaining the thermal crosstalk contribution vector. Through the above processing, the neighborhood thermal interference component originally implicit in the mixed temperature signal can be explicitly stripped and quantized into an independent data object, so that the subsequent deviation calculation has a data basis to distinguish its own thermal contribution from the neighborhood thermal interference.

[0034] More specifically, in a specific example of this application, during the device calibration phase, each lamp is individually turned on and continuously heated at a constant power until the overall temperature field reaches a steady state. The steady-state temperature values ​​of all sub-regions are recorded at this point. By comparing the temperature rise of each non-heated region with the temperature rise of the heated region, the thermal conduction coupling ratio coefficient between any two sub-regions is calculated. All coupling coefficients are then used to construct a system with dimensions of [missing information]. The inter-region thermal coupling coefficient matrix is ​​stored in the controller's non-volatile storage area. During the thermal crosstalk quantization phase of each control cycle, the pre-stored inter-region thermal coupling coefficient matrix is ​​retrieved, and a cross-region weighted summation is performed on all channels in the sub-region average temperature vector. For the p-th sub-region to be analyzed, the average temperature of all other sub-regions (excluding itself) is multiplied by their corresponding thermal coupling coefficients, and then summed to quantify the equivalent temperature contribution introduced by neighboring thermal diffusion in the measured temperature of that sub-region. This operation is performed on all N channels one by one, and finally, the thermal crosstalk contribution values ​​of each channel are assembled in spatial order and output as a thermal crosstalk contribution vector. This operation can be expressed as:

[0035] in, This represents the equivalent temperature contribution of the p-th sub-region in the thermal crosstalk contribution vector to the thermal crosstalk from its neighboring region, expressed in °C. The element in the p-th row and j-th column of the inter-regional thermal coupling coefficient matrix represents the thermal conduction coupling ratio coefficient of the j-th sub-region to the p-th sub-region. It is dimensionless and is obtained in advance through steady-state thermal calibration experiments. Let be the measured average temperature value of the j-th sub-region in the sub-region average temperature vector, in °C, and N be the total number of heated sub-regions. It should be noted that when the temperature of a certain sub-region j is higher and the coupling coefficient between it and the target sub-region p is larger, the value of Q(p) is larger, which means that the sub-region is more strongly affected by the thermal interference from its neighbors, and the larger the non-intrinsic temperature components that need to be stripped off later.

[0036] Taking a lamp array containing 5 heating sub-regions as an example, assuming the average temperature vector of the sub-regions in the current control cycle is [180,195,175,190,182]℃, and the off-diagonal elements corresponding to the 3rd row of the inter-region thermal coupling coefficient matrix are α(3,1)=0.02, α(3,2)=0.08, α(3,4)=0.07, and α(3,5)=0.01 respectively, then the equivalent temperature contribution value of the neighboring thermal crosstalk to the 3rd sub-region is calculated to be 34.32℃. This value indicates that 34.32℃ of the measured temperature of 175℃ in the 3rd sub-region comes from the contribution of heat conduction and diffusion from the neighboring lamps, rather than from its own lamp heating. Subsequent deviation calculations need to remove this non-intrinsic component to obtain the true self-heating state deviation.

[0037] In step S332, based on the neighborhood heat diffusion component encoded in each channel of the thermal crosstalk contribution vector, the sub-region average temperature vector is decoupled and isolated channel by channel to obtain a decoupled intrinsic temperature vector. It should be noted that, given that the equivalent temperature contribution value of the neighborhood thermal crosstalk experienced by each sub-region has been explicitly quantized into an independent thermal crosstalk contribution vector in the previous step, and the measured temperature of each channel in the sub-region average temperature vector is still a mixture of the intrinsic temperature component generated by its own lamp heating and the non-intrinsic temperature component introduced by neighborhood heat diffusion, if this mixed temperature is directly used in the deviation calculation, the PID controller will make power adjustment decisions based on the distorted deviation signal, resulting in a coupled feedback path between adjacent channels through shared temperature signals. Therefore, the technical solution of this application further decouples and isolates the sub-region average temperature vector channel by channel by channel based on the neighborhood heat diffusion component encoded in each channel of the thermal crosstalk contribution vector to obtain a decoupled intrinsic temperature vector. Through the above processing, the subsequent deviation calculation and the input signal of the PID controller can truly reflect the heating state of each lamp, and cut off the path of coupling feedback formed by adjacent channels through shared temperature signals.

[0038] More specifically, in a specific example of this application, a channel-by-channel difference stripping operation is performed on the sub-region average temperature vector and the thermal crosstalk contribution vector. That is, the measured average temperature of each channel is subtracted from the corresponding channel's thermal crosstalk contribution value; the difference result is the independent contribution of the lamp's own heating power to the local temperature in that sub-region. For the p-th sub-region, the measured average temperature value of that channel in the sub-region average temperature vector is read, and simultaneously, the equivalent temperature contribution value of the corresponding neighborhood thermal diffusion in the thermal crosstalk contribution vector is read. Subtracting the latter from the former reveals the intrinsic temperature generated by the lamp's own heating in that sub-region. This difference operation is performed on all N channels one by one, and the results of all channel operations are assembled and output as a decoupled intrinsic temperature vector, represented as:

[0039] in, To decouple the intrinsic temperature value of the p-th sub-region in the intrinsic temperature vector, the unit is ℃, which only reflects the independent temperature contribution generated by the lamp heating in that region. Let be the measured average temperature value of the p-th sub-region in the sub-region average temperature vector. This represents the equivalent temperature contribution value of the neighborhood thermal crosstalk to the p-th sub-region in the thermal crosstalk contribution vector. The temperature data processed in this step eliminates the non-intrinsic temperature superposition introduced by heat conduction from neighboring lamps, ensuring that subsequent deviation calculations and the input signal to the PID controller accurately reflect the heating state of each lamp, rather than a superposition of multiple lamps' combined effects. This fundamentally cuts off the path for adjacent channels to form coupled feedback through shared temperature signals.

[0040] Using the scenario of the aforementioned five heating sub-regions, the measured average temperature of the third sub-region is 175℃. The equivalent temperature contribution value of the neighboring thermal crosstalk calculated in the previous step is 34.32℃. Therefore, the intrinsic temperature of this sub-region is calculated to be 140.68℃. This value indicates that the heating power of the lamp itself in the third sub-region only maintains the local temperature at 140.68℃, and the remaining 34.32℃ comes from the thermal diffusion contribution of the neighboring lamps. Subsequent deviation calculations will use 140.68℃ instead of 175℃ as the true temperature state of this channel for comparison with the target temperature, thereby avoiding the incorrect reduction of the lamp power due to the high mixed temperature.

[0041] In step S333, the deviation calculation and spatial layout reconstruction are performed on the target temperature of each sub-region in the preset target temperature matrix and the corresponding intrinsic temperature in the decoupled intrinsic temperature vector to obtain a spatial temperature deviation matrix. It should be noted that, given that the non-intrinsic components introduced by neighborhood thermal diffusion in the measured temperature of each sub-region have been removed in the previous step, resulting in a decoupled intrinsic temperature vector that only reflects the heating contribution of the lamp itself, this intrinsic temperature needs to be compared with the preset target temperature to generate a deviation signal that accurately corresponds to the power adjustment requirements of each lamp, rather than a fuzzy deviation mixed with uncontrollable neighborhood thermal interference. This allows each channel controller to independently and without mutual interference execute power adjustment decisions. Based on this, the technical solution of this application further performs deviation calculation and spatial layout reconstruction on the target temperature of each sub-region in the preset target temperature matrix and the corresponding intrinsic temperature in the decoupled intrinsic temperature vector to obtain a spatial temperature deviation matrix. Through the above processing, the deviation signal sent to the PID controller can accurately correspond to the required power adjustment range of each lamp, avoiding the problem of repeated overcompensation of adjacent channels caused by coupling misjudgment.

[0042] More specifically, in a concrete example of this application, after obtaining the decoupled intrinsic temperature vector after removing neighboring thermal interference, the element-by-element difference operation is performed between the target temperature of each sub-region in the preset target temperature matrix and the intrinsic temperature of the corresponding channel in the decoupled intrinsic temperature vector to calculate the true power control deviation of each spatial location under the condition of removing coupling interference. For the p-th sub-region, the expected target temperature value of the sub-region in the preset target temperature matrix is ​​read, and the intrinsic temperature value corresponding to the channel in the decoupled intrinsic temperature vector is read. The temperature deviation value of the sub-region is obtained by subtracting the intrinsic temperature from the target temperature. A positive value indicates that the intrinsic temperature is lower than the target and the power needs to be increased, and a negative value indicates that the intrinsic temperature is higher than the target and the power needs to be decreased. After completing the deviation calculation for all N channels one by one, the deviation value of each channel is reconstructed into a matrix form corresponding to the physical space lamp layout, and the spatial temperature deviation matrix is ​​output and passed to the subsequent PID incremental control stage, represented as:

[0043] in, This represents the temperature deviation value corresponding to the p-th sub-region in the spatial temperature deviation matrix, in °C. This represents the preset target temperature value corresponding to the p-th sub-region in the preset target temperature matrix, in °C. To decouple the intrinsic temperature value of the p-th sub-region in the intrinsic temperature vector. Since the deviation signal fed into the PID controller already accurately corresponds to the magnitude of the power adjustment required for each lamp itself, rather than the fuzzy deviation mixed with uncontrollable neighborhood thermal interference, each channel controller can independently and without mutual interference execute power adjustment decisions, avoiding the problem of repeated overcompensation of adjacent channels caused by coupling misjudgment.

[0044] Using the scenario of the aforementioned five heating sub-regions, assuming the target temperature of the third sub-region in the preset target temperature matrix is ​​170℃, and the intrinsic temperature of this sub-region calculated in the previous step is 141.28℃, then the calculated spatial temperature deviation value of this sub-region is 28.72℃. This positive deviation value accurately reflects the degree of insufficient heating power of the lamp itself in the third sub-region. The PID controller will then independently increase the power of the lamp based on this, and will not mistakenly determine that this region only needs slight adjustment or even needs to reduce power due to the mixed temperature of 175℃ caused by thermal crosstalk in the neighboring area, which is close to the target temperature of 170℃.

[0045] Specifically, in step S4, feedback-based power adjustment increment estimation is performed on the deviation values ​​of each sub-region in the spatial temperature deviation matrix to obtain the spatial power adjustment increment vector. It should be noted that the spatial temperature deviation matrix only provides the static deviation between the current temperature and the target temperature of each sub-region. In actual control, lamp power adjustment needs to consider not only the magnitude and direction of the current deviation but also the historical cumulative trend and rate of change of the deviation to achieve a comprehensive control effect of rapid response, elimination of static errors, and suppression of overshoot oscillations. Based on this, the technical solution of this application further performs feedback-based power adjustment increment estimation on the deviation values ​​of each sub-region in the spatial temperature deviation matrix to obtain the spatial power adjustment increment vector. Through the above processing, the temperature deviation signal of each sub-region can be converted into a power adjustment increment with dynamic adaptive characteristics, providing accurate incremental control output for subsequent superposition with the reference heating power to generate the final executable power command.

[0046] Figure 6 This is a flowchart illustrating a method for adaptive control of heating power in additive manufacturing based on infrared thermal imaging, according to an embodiment of this application. It describes the estimation of the power adjustment increment vector by performing feedback-based power adjustment increment estimation on the deviation values ​​of each sub-region in the spatial temperature deviation matrix. Figure 6 As shown, step S4 includes: S41, calculating the sum of the current deviation values ​​of each sub-region in the spatial temperature deviation matrix based on historical control cycles and the difference between adjacent cycles to obtain the integral and differential value sets; S42, based on preset proportional control coefficients, integral control coefficients, and differential control coefficients, performing proportional-integral-derivative weighted allocation on the current deviation values ​​and integral and differential value sets in the spatial temperature deviation matrix to obtain weighted control components; S43, performing channel-by-channel power adjustment increment linear synthesis on the proportional, integral, and differential components of each sub-region channel in the weighted control components to obtain the spatial power adjustment increment vector.

[0047] In step S41, the current deviation values ​​of each sub-region in the spatial temperature deviation matrix are accumulated and summed over historical control cycles, and the difference between adjacent cycles is calculated to obtain the integral and derivative value sets. It should be noted that since power adjustment is based solely on the instantaneous temperature deviation value of the current cycle, it is impossible to eliminate persistent static errors caused by factors such as lamp thermal inertia, nor can it predict the deviation change trend to suppress temperature overshoot in advance. Therefore, it is necessary to extract the historical accumulation and rate of change of the deviation signal over time. Based on this, the technical solution of this application further calculates the integral and derivative value sets by accumulating and summing the current deviation values ​​of each sub-region in the spatial temperature deviation matrix over historical control cycles and the difference between adjacent cycles. Through the above processing, time-series characteristic data reflecting the historical accumulation and rate of change of the deviation can be provided for subsequent PID weighting.

[0048] More specifically, in a concrete example of this application, within each control cycle, the deviation value of each sub-region in the spatial temperature deviation matrix at the current moment is read, and simultaneously, the sequence of deviation values ​​recorded for that sub-region in all previous control cycles is retrieved from the controller's historical data buffer. For the p-th sub-region, all historical deviation values ​​from the initial control moment to the current moment are arithmetically summed cycle by cycle to obtain the deviation integral term of that channel, which reflects the degree of continuous accumulation of deviation on the time axis. To prevent the integral term from excessively increasing due to long-term unidirectional deviation accumulation and causing control overshoot, when the integral term value exceeds a preset upper limit threshold, it is truncated and clamped to the upper limit value; when the integral term value is lower than a preset lower limit threshold, it is truncated and clamped to the lower limit value. Simultaneously, the deviation value of the current cycle is subtracted from the deviation value of the previous control cycle to obtain the deviation derivative term of that channel, which reflects the rate and direction of change of deviation between adjacent cycles. The above calculation can be expressed as:

[0049]

[0050] in, Let be the integral term of the deviation of the p-th sub-region at the current time t. Let be the differential term of the deviation of the p-th subregion at the current time t. Let be the temperature deviation value of the p-th sub-region in the spatial temperature deviation matrix during the τ-th historical control period. This is the current cycle deviation value. This represents the deviation value from the previous cycle. After calculating the integral and differential terms for each of the N sub-region channels, the integral and differential values ​​of each channel are combined and encapsulated in spatial index order, and the set of integral and differential values ​​is output for use in subsequent weighting and allocation steps.

[0051] In step S42, based on preset proportional control coefficients, integral control coefficients, and derivative control coefficients, the current deviation value and the integral and derivative value sets in the space temperature deviation matrix are subjected to proportional-integral-derivative weighted allocation to obtain weighted control components. It should be noted that since the deviation integral and derivative terms differ from the current deviation value in terms of numerical dimensions and physical meaning, they need to be uniformly calibrated as control components under the power adjustment dimension using their respective gain coefficients to form a coordinated integrated control output in subsequent synthesis. Based on this, the technical solution of this application further performs proportional-integral-derivative weighted allocation on the current deviation value and the integral and derivative value sets in the space temperature deviation matrix based on preset proportional control coefficients, integral control coefficients, and derivative control coefficients to obtain weighted control components. Through the above processing, the current value, historical cumulative amount, and rate of change of the deviation signal can be mapped to three independent control components with clear power adjustment significance.

[0052] More specifically, in a specific example of this application, pre-tuned proportional control coefficients, integral control coefficients, and derivative control coefficients are read from the control parameter configuration. These three coefficients determine the controller's immediate response strength to the current deviation magnitude, the strength of eliminating historical accumulated deviations, and the degree of proactive suppression of deviation change trends, respectively. For the p-th sub-region channel, the proportional component is obtained by multiplying the current deviation value of that channel in the spatial temperature deviation matrix by the proportional control coefficient; the integral component is obtained by multiplying the integral term of that channel in the integral and derivative value set by the integral control coefficient; and the derivative component is obtained by multiplying the derivative term of that channel by the derivative control coefficient. The above calculation can be expressed as:

[0053]

[0054]

[0055] in, This is the proportional control component for the p-th sub-region. For integral control components, These are differential control components, all in units of W. This is the proportional control coefficient. The integral control coefficient, The differential control coefficient, This is the current cycle deviation value. For the integral term of deviation, This is the deviation differential term. After performing the above three-way weighted operation on all N sub-region channels one by one, the proportional component, integral component and differential component of each channel are combined and encapsulated in spatial index order, and the weighted control component is output for use in the subsequent incremental synthesis step.

[0056] In step S43, the proportional, integral, and derivative components of each sub-region channel in the weighted control component are linearly synthesized channel-by-channel power adjustment increments to obtain a spatial power adjustment increment vector. It should be noted that since the proportional, integral, and derivative components of each sub-region channel in the weighted control component carry control information in different time dimensions, they must be synthesized into a single power adjustment increment value to serve as the direct input for subsequent power command superposition. Based on this, the technical solution of this application further performs channel-by-channel power adjustment increment linear synthesis of the proportional, integral, and derivative components of each sub-region channel in the weighted control component to obtain a spatial power adjustment increment vector. Through the above processing, the three control components can be uniformly synthesized into the power change required by each lamp in the current control cycle, providing a complete incremental control output for subsequent superposition with the reference power.

[0057] More specifically, in a concrete example of this application, for the p-th sub-region channel, the proportional component, integral component, and differential component of that channel are read from the weighted control components, and the three are algebraically summed. The sum is the incremental heating power adjustment value required by the curing lamp in that sub-region during the current control cycle. A positive value indicates that the power output needs to be increased, and a negative value indicates that the power output needs to be decreased. This operation can be expressed as:

[0058] in, This represents the heating power adjustment increment value corresponding to the p-th sub-region in the spatial power adjustment increment vector, in W. This is the proportional control component for this channel. For integral control components, This is the differential control component. After performing the above three linear summations on each of the N sub-region channels, the power adjustment increment values ​​of each channel are arranged in spatial index order into an N-dimensional vector, which is the spatial power adjustment increment vector, for subsequent superposition calculation with the reference heating power configuration.

[0059] Specifically, in step S5, the spatial power adjustment increment vector is superimposed on the reference heating power configuration, and a safety threshold limiting correction is applied to obtain an adaptive heating power command set. It should be noted that since the spatial power adjustment increment vector is only the power change of each sub-region relative to the steady-state operating point in the current control cycle, it needs to be superimposed on the reference heating power of each lamp to form a complete absolute power command. Furthermore, the physical hardware of the lamps has an upper power tolerance limit and a lower minimum operating limit. If the power command exceeds the safe range, it will cause the lamp to overheat and be damaged or fail to light up normally. Based on this, the technical solution of this application further superimposes the spatial power adjustment increment vector on the reference heating power configuration and applies a safety threshold limiting correction to obtain an adaptive heating power command set. Through the above processing, an executable power command that reflects both the dynamic temperature deviation compensation requirements and meets hardware safety constraints can be generated, providing an effective power setting input for subsequent drive signal conversion.

[0060] Figure 7 This is a flowchart illustrating an adaptive heating power control method for additive manufacturing based on infrared thermal imaging, according to an embodiment of this application. The method involves superimposing a spatial power adjustment increment vector onto a reference heating power configuration and performing a safety threshold limiting correction to obtain an adaptive heating power instruction set. Figure 7 As shown, step S5 includes: S51, superimposing the initial heating power command values ​​of each sub-region power increment in the spatial power adjustment increment vector with the corresponding sub-region reference power value in the reference heating power configuration to obtain the original heating power command matrix; S52, based on the maximum allowable power upper limit and minimum allowable power lower limit of a single zone of the curing lamp tube, performing safety domain amplitude truncation and limiting correction on each element in the original heating power command matrix to obtain the adaptive heating power command set.

[0061] In step S51, the power increment of each sub-region in the spatial power adjustment increment vector is superimposed with the corresponding sub-region reference power value in the reference heating power configuration to obtain the original heating power command matrix. It should be noted that since the values ​​in the spatial power adjustment increment vector are incremental changes relative to the steady-state operating point rather than absolute power values, they cannot be directly used as lamp driving commands. They need to be superimposed on the preset reference heating power of each sub-region to form a complete absolute power setting value. Based on this, the technical solution of this application further superimposes the power increment of each sub-region in the spatial power adjustment increment vector with the corresponding sub-region reference power value in the reference heating power configuration to obtain the original heating power command matrix. Through the above processing, the dynamic incremental compensation and static reference power can be combined into the absolute power target value of each lamp in the current cycle.

[0062] More specifically, in a specific example of this application, the reference heating power configuration is a steady-state operating power value of the lamps in each sub-region that is pre-set and stored during the equipment commissioning phase. This configuration reflects the basic power level required for each lamp to maintain its target temperature under conditions without temperature deviation disturbances. Within each control cycle, for the p-th sub-region, the reference power value corresponding to that sub-region is read from the reference heating power configuration, and simultaneously, the power adjustment increment value corresponding to that channel is read from the spatial power adjustment increment vector. The two are algebraically added together, and the sum is the original power command value of the lamps in that sub-region in the current cycle. This operation can be expressed as:

[0063] in, This represents the original power command value corresponding to the p-th sub-region in the original heating power command matrix, in W. The preset reference power value for the p-th sub-region in the reference heating power configuration, in W. This represents the power adjustment increment value corresponding to the p-th sub-region in the spatial power adjustment increment vector, in W. After performing the above superposition operation on all N sub-region channels one by one, the original power command values ​​of each channel are assembled according to the spatial layout order, and the original heating power command matrix is ​​output for use in the subsequent safety limit correction step.

[0064] In step S52, based on the maximum allowable power upper limit and minimum allowable power lower limit of a single zone of the curing lamp, the elements in the original heating power command matrix are truncated and limited by the safety domain amplitude to obtain an adaptive heating power command set. It should be noted that since the values ​​in the original heating power command matrix, after incremental superposition, may exceed the physical tolerance range of the curing lamp hardware, excessive power will lead to lamp overheating and damage or driver overload, while insufficient power will prevent the lamp from maintaining normal operation. Therefore, the technical solution of this application further truncates and limits the elements in the original heating power command matrix based on the maximum allowable power upper limit and minimum allowable power lower limit of a single zone of the curing lamp to obtain an adaptive heating power command set. Through the above processing, it can be ensured that the final power commands all fall within the safe operating range of the hardware, ensuring equipment operation safety while meeting control requirements.

[0065] More specifically, in a specific example of this application, the maximum allowable power upper limit and the minimum allowable power lower limit for a single zone are predetermined based on the hardware specifications of the curing lamp. The former is the highest electrical power value that the lamp and its driving circuit can safely withstand under continuous operation, and the latter is the minimum operating power value required for the lamp to maintain a stable lighting state. Within each control cycle, for the power command value of the p-th sub-region in the original heating power command matrix, a bilateral threshold truncation judgment is performed: if the value is greater than the maximum allowable power upper limit, it is forcibly truncated to the maximum allowable power upper limit value; if the value is less than the minimum allowable power lower limit, it is forcibly truncated to the minimum allowable power lower limit value; if the value is within the safe range between the upper and lower limits, the original value remains unchanged. This operation can be expressed as:

[0066] in, This represents the final adaptive power command value corresponding to the p-th sub-region in the adaptive heating power command set, in W. This represents the original power command value corresponding to the p-th sub-region in the original heating power command matrix. This is the maximum allowable power limit for a single zone of the curing lamp. To solidify the minimum allowable power limit for a single zone of the lamp, after performing the above-mentioned limiting correction on each of the N sub-region channels, the corrected power values ​​of each channel are assembled in spatial order, and an adaptive heating power command set is output for use in subsequent drive signal conversion steps.

[0067] Specifically, in step S6, based on the pulse width modulation characteristics of the curing lamp driver hardware, the adaptive heating power instruction set is subjected to duty cycle conversion and signal encoding to generate a power drive signal sequence. This power drive signal sequence is used to drive the curing lamps at each spatial location to achieve differentiated adaptive power adjustment. It should be noted that since the values ​​in the adaptive heating power instruction set are power setpoints in watts, which are logic quantities at the control level, they cannot be directly recognized and executed by the power switching circuit of the curing lamp driver hardware. They need to be converted into pulse width modulation signals that the driver hardware can receive to achieve physical-level power output control. Based on this, the technical solution of this application further utilizes the pulse width modulation characteristics of the curing lamp driver hardware to perform duty cycle conversion and signal encoding on the adaptive heating power instruction set to generate a power drive signal sequence. Through the above processing, the power setpoints at the control logic layer can be converted into physical drive signals that can be executed by the hardware, achieving differentiated adaptive power adjustment for the curing lamps at each spatial location.

[0068] More specifically, in a concrete example of this application, firstly, based on the rated physical maximum power parameter of the curing lamp, a linear duty cycle mapping transformation from power to percentage is performed on the power wattage values ​​of each sub-region in the adaptive heating power instruction set. For the p-th sub-region, the power instruction value of that channel in the adaptive heating power instruction set is read, divided by the rated physical maximum power value of the lamp, and multiplied by 100% to obtain the target duty cycle percentage value of the pulse width modulation signal for that channel. This duty cycle has a linear correspondence with the actual output power of the lamp. After completing the above mapping for all N channels, the duty cycle values ​​of each channel are arranged in spatial order to form a duty cycle vector. This operation can be expressed as:

[0069] in, This represents the pulse width modulation duty cycle percentage value corresponding to the p-th channel in the duty cycle vector. This represents the power command value for the p-th sub-region in the adaptive heating power command set, in W. This refers to the rated physical maximum power of the curing lamp, expressed in watts (W).

[0070] Subsequently, based on the microcontroller timer auto-reload register value and the communication bus protocol frame format, the duty cycle vector is mapped to an integer value using register comparison and encoded serially. For each channel's duty cycle value, it is multiplied by the microcontroller timer's auto-reload register setting and rounded to the nearest integer to obtain the target write value for that channel's timer comparison register. This integer value directly determines the high-level duration counting period of the hardware timer's output pulse-width modulated square wave signal. This operation can be expressed as:

[0071] in, The target write value for the p-th channel timer compare register is a positive integer. This represents the duty cycle percentage value for that channel. This is the setting value for the microcontroller's timer auto-reload register. For rounding operations. After calculating the register values ​​for all channels, the target values ​​of the comparison registers for each channel are serially encoded according to the frame format of the communication bus protocol, and encapsulated into a binary data stream containing a frame header identifier, channel address, data field, and check field, generating a power drive signal sequence. This signal sequence is sent to the driver boards of each curing lamp via the physical communication interface. After parsing the data frame, the microcontroller of the driver board writes the corresponding comparison register value into a hardware timer, thereby physically outputting a pulse width modulated square wave signal with the corresponding duty cycle for each channel, controlling the on / off time ratio of the power switching devices, and ultimately driving the curing lamps in each spatial position to output heating at differentiated power levels.

[0072] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An adaptive control method for heating power in additive manufacturing based on infrared thermal imaging, characterized in that, include: S1. The infrared thermal imaging camera collects the temperature field of the printed plane to obtain the original infrared thermal image frame, which contains the infrared radiation intensity value of each pixel. S2, based on the material surface emissivity parameters, performs temperature calibration conversion on the original infrared thermal image frame to obtain a calibrated temperature distribution map; S3. Based on the spatial projection position of the curing lamp, perform multi-regional temperature field deviation spatial mapping on the calibration temperature distribution map to obtain the spatial temperature deviation matrix. S4, perform feedback-based power adjustment increment estimation on the deviation values ​​of each sub-region in the spatial temperature deviation matrix to obtain the spatial power adjustment increment vector; S5, superimpose the spatial power adjustment increment vector with the reference heating power configuration, and perform safety threshold limiting correction to obtain the adaptive heating power instruction set; S6, based on the pulse width modulation characteristics of the curing lamp driver hardware, performs duty cycle conversion and signal encoding on the adaptive heating power instruction set to generate a power drive signal sequence, wherein the power drive signal sequence is used to drive the curing lamps at each spatial position to achieve differentiated adaptive power adjustment.

2. The adaptive control method for additive manufacturing heating power based on infrared thermal imaging according to claim 1, characterized in that, Step S2 includes: Spatial domain noise reduction filtering is performed on the original infrared thermal image frame to obtain the filtered thermal image frame; Based on the pre-calibrated and stored detector pixel gain coefficient matrix and bias coefficient matrix, the filtered thermal image frame is subjected to focal plane array non-uniformity correction to obtain the corrected thermal image frame. Based on the surface emissivity parameters of the printing material, the corrected thermal image frame is calibrated by absolute temperature conversion based on emissivity to obtain a calibrated temperature distribution map.

3. The adaptive control method for heating power in additive manufacturing based on infrared thermal imaging according to claim 1, characterized in that, Step S3 includes: Based on the pixel boundary range of each heating sub-region determined by the spatial topological geometric parameters of the curing lamp tube, the calibration temperature distribution map is subjected to binary mask cropping and region-by-region partitioning extraction to obtain a set of sub-region temperature maps. The average temperature vector of the sub-region is obtained by averaging and dimensionality reduction of all valid pixel temperature values ​​within the coverage area of ​​each sub-region in the sub-region temperature map set. The spatial temperature deviation matrix is ​​obtained by differential quantization of the sub-region average temperature vector and the corresponding sub-region target temperature value in the preset target temperature matrix.

4. The adaptive control method for heating power in additive manufacturing based on infrared thermal imaging according to claim 1, characterized in that, Step S4 includes: The current deviation values ​​of each sub-region in the spatial temperature deviation matrix are accumulated and summed over historical control cycles, and the difference between adjacent cycles is calculated to obtain the set of integral and differential values. Based on the preset proportional control coefficient, integral control coefficient and derivative control coefficient, the current deviation value and the integral and derivative value sets in the spatial temperature deviation matrix are respectively subjected to proportional, integral and derivative weighting to obtain the weighted control components. The proportional, integral, and differential components of each sub-region channel in the weighted control component are linearly synthesized channel-by-channel power adjustment increments to obtain the spatial power adjustment increment vector.

5. The adaptive control method for heating power in additive manufacturing based on infrared thermal imaging according to claim 1, characterized in that, Step S5 includes: The initial heating power command value is superimposed on the power increment of each sub-region in the spatial power adjustment increment vector and the corresponding sub-region reference power value in the reference heating power configuration to obtain the original heating power command matrix. Based on the maximum allowable power upper limit and minimum allowable power lower limit of a single zone of a curing lamp, the element in the original heating power command matrix is ​​truncated and limited by the safety domain amplitude to obtain an adaptive heating power command set.

6. The adaptive control method for heating power in additive manufacturing based on infrared thermal imaging according to claim 1, characterized in that, Step S6 includes: Based on the rated physical maximum power parameter of the curing lamp, the power wattage value of each sub-region in the adaptive heating power instruction set is transformed into a duty cycle vector by a power-to-percentage linear mapping. Based on the microcontroller timer auto-reload register value and the communication bus protocol frame format, the duty cycle vector is mapped to integer register comparison values ​​and encoded serially to generate a power drive signal sequence, which is used to drive the curing lamps at each spatial position.

7. The adaptive control method for heating power in additive manufacturing based on infrared thermal imaging according to claim 3, characterized in that, The spatial temperature deviation matrix is ​​obtained by differential quantization of the sub-region average temperature vector and the corresponding sub-region target temperature value in the preset target temperature matrix, including: Based on the inter-regional thermal coupling coefficient matrix obtained in advance through steady-state thermal calibration experiments, the thermal crosstalk contribution vector of each channel temperature value in the sub-region average temperature vector is quantified by neighborhood thermal crosstalk contribution quantification. Based on the neighborhood thermal diffusion component encoded in each channel of the thermal crosstalk contribution vector, the sub-region average temperature vector is decoupled and isolated channel by channel to obtain the decoupled intrinsic temperature vector. The spatial temperature deviation matrix is ​​obtained by calculating the deviation between the target temperature of each sub-region in the preset target temperature matrix and the corresponding intrinsic temperature in the decoupled intrinsic temperature vector and reconstructing the spatial layout.

Images