A temperature control method and system for variable aperture 3D printing

Abstract

This invention provides a temperature control method and system for variable-aperture 3D printing, relating to the field of 3D printing technology. The method includes: acquiring nozzle state data of the 3D printer; constructing a digital twin model to simulate the internal thermal-fluid coupling state of the nozzle; calibrating the parameters in the digital twin model based on the nozzle state data; obtaining a look-ahead command containing information on future nozzle aperture changes; inputting the look-ahead command into the calibrated digital twin model, outputting the nozzle heat load change trend and nozzle temperature change trend; generating a heating power command based on the nozzle heat load change trend and nozzle temperature change trend; and controlling the temperature of the variable-aperture 3D printing according to the heating power command. This invention can reduce temperature fluctuations caused by rapid changes in operating conditions, improve the stability and response speed of temperature control, and better meet the actual needs of high-end manufacturing fields for stable, reliable, and high-precision temperature control in the printing process.

Description

Technical Field

[0001] This invention relates to the field of 3D printing technology, and in particular to a temperature control method and system for variable aperture 3D printing. Background Technology

[0002] Variable nozzle diameter 3D printing technology can flexibly adapt to the printing accuracy and filling efficiency of different areas by adjusting the nozzle diameter in real time. It has significant advantages in the field of rapid prototyping of complex components. Stable control of nozzle temperature is the core link to ensure melt flowability, extrusion uniformity and interlayer bonding quality, which directly determines the molding accuracy and mechanical properties of the final printed part.

[0003] Currently, traditional 3D printing temperature control mostly adopts PID-based feedback control schemes, with some schemes supplemented by simple feedforward compensation. The heating power is adjusted by collecting the deviation between the real-time nozzle temperature and the set value, while the printing process is completed by relying on fixed instructions generated by the slicing software. Some systems will perform basic safety monitoring on the heating power and actuator status.

[0004] However, existing variable nozzle diameter 3D printing temperature control technology lacks the ability to predict future thermal load disturbances caused by planned changes in nozzle diameter in real time and accurately. It also cannot perform advanced and precise compensation control of the heating process based on the prediction results. This leads to problems such as large temperature fluctuations and lag response when the system changes rapidly, making it difficult to meet the actual needs of high-end manufacturing fields for printing process stability, reliability and high-precision temperature control. Summary of the Invention

[0005] To address the technical challenges of existing variable nozzle diameter 3D printing temperature control technologies, which lack the ability to accurately predict future thermal load disturbances caused by planned changes in nozzle diameter in real time, and cannot perform advanced and precise compensation control of the heating process based on the prediction results, resulting in problems such as large temperature fluctuations and lag response when the system experiences rapid changes in operating conditions, it is difficult to meet the actual needs of high-end manufacturing fields for printing process stability, reliability, and high-precision temperature control.

[0006] The technical solution provided by this invention is as follows: A first aspect of the present invention provides a temperature control method for variable aperture 3D printing, comprising: S1: Collect nozzle status data of the 3D printer; S2: Construct a digital twin model to simulate the internal thermal-fluid coupling state of the nozzle; S3: Based on the nozzle status data, calibrate the parameters in the digital twin model; S4: Obtain forward-looking instructions containing information on future nozzle diameter changes; S5: Input the look-ahead command into the calibrated digital twin model and output the nozzle heat load change trend and nozzle temperature change trend; S6: Generate heating power command based on the nozzle heat load change trend and nozzle temperature change trend; S7: Controls the temperature of variable diameter 3D printing according to the heating power command.

[0007] A second aspect of the present invention provides a temperature control system for variable aperture 3D printing, comprising: processor; The memory stores computer-readable instructions, which, when executed by a processor, implement the temperature control method for variable-diameter 3D printing as described in the first aspect.

[0008] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the temperature control method for variable-diameter 3D printing as described in the first aspect.

[0009] The beneficial effects of the technical solution provided by this invention include: By constructing and calibrating a thermal-fluid coupled digital twin model, and combining it with forward-looking instructions containing information on future nozzle diameter changes for advanced prediction, the model can output the nozzle heat load change trend, enabling effective prediction of future heat load disturbances. Based on the heat load and temperature change trends output by the calibrated model, a heating power command is generated, which can perform advanced and precise compensation control of the heating process. This helps to reduce temperature fluctuations caused by rapid changes in operating conditions, improve the stability and response speed of temperature control, and better meet the actual needs of high-end manufacturing fields for stable, reliable, and high-precision temperature control in the printing process. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a temperature control method for variable aperture 3D printing provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a temperature control system for variable aperture 3D printing provided in an embodiment of the present invention. Detailed Implementation

[0011] Reference manual attached Figure 1 The diagram shows a flow chart of a temperature control method for variable aperture 3D printing provided by an embodiment of the present invention.

[0012] This invention provides a temperature control method for variable aperture 3D printing, which may include the following steps: S1: Collect the nozzle status data of the 3D printer.

[0013] It should be noted that the 3D printer is equipped with a data acquisition system that collects nozzle status data at a fixed control cycle (e.g., 100ms).

[0014] The nozzle status data specifically includes the current nozzle diameter, temperature at at least one location, extrusion flow rate, actuator status, and environmental status.

[0015] It should be noted that "at least one location" refers to at least one of the heating block core or nozzle throat locations.

[0016] The nozzle orifice diameter can be obtained by reading the value of the encoder or linear displacement sensor directly connected to the variable orifice mechanism, thereby obtaining the current flow channel outlet diameter.

[0017] Temperature can be obtained by using K-type thermocouples or fast-response RTD sensors placed in the core of the heating block, the throat of the nozzle, etc. Preferably, noise is suppressed by a combination of hardware filtering (such as RC circuit) and software filtering (such as moving average) to preserve the true dynamics.

[0018] The extrusion flow rate can be obtained directly from a micro flow meter or from the flow rate value calculated based on the pressure after the extrusion screw, or indirectly by reading the encoder feedback of the extrusion motor and combining it with the screw geometric parameters to calculate the equivalent extrusion speed or mass flow rate.

[0019] Specifically, the actuator status includes the current nozzle heater PWM duty cycle (or power) and the extrusion motor set speed.

[0020] The environmental conditions include ambient temperature, which is used to correct the heat loss model.

[0021] It should be noted that the nozzle status data is acquired through industrial real-time Ethernet (such as EtherCAT) or a microcontroller ADC-DMA channel with precise timing to ensure that the timestamp jitter of the data packets is less than 10% of the control cycle, thereby achieving hardware-level synchronization.

[0022] Furthermore, the steps of data diagnosis and filtering for the collected data specifically include determining whether the readings of each sensor are within the physically reasonable upper and lower limits such as 20–300°C. If they exceed the limits, the data is marked as invalid. The steps also include checking whether the sudden changes in adjacent period data exceed the maximum possible rate of change such as 10°C / second. If they do, they are considered abnormal jumps and the effective value or predicted value of the previous period is used instead. At the same time, the valid raw data that passes the check is low-pass filtered to remove high-frequency electrical noise while retaining the real dynamic frequency components caused by aperture changes, etc., so as to improve data reliability and model input quality.

[0023] S2: Construct a digital twin model to simulate the internal thermal-fluid coupling state of the nozzle.

[0024] It should be noted that a lightweight digital twin model is constructed to quickly and approximately simulate the key dynamic responses of the internal temperature field and melt flow during nozzle diameter changes. To meet real-time computation requirements, the digital twin model does not pursue high-fidelity full-physics simulation, but rather achieves rapid forward computation while ensuring prediction accuracy. The model can meet prediction accuracy under real-time computation conditions, where real-time computation means that the time for a single prediction is no more than 1 second; the model can complete predictions and output results that meet accuracy requirements within 100ms to 500ms, thus adapting to the real-time temperature control needs of variable-diameter 3D printing.

[0025] In one possible implementation, S2 specifically includes: S201: Based on the nozzle design data of the 3D printer, establish a parametric three-dimensional geometric model of the internal flow channel and heating area.

[0026] Specifically, based on the design data of the target variable-diameter nozzle, such as mechanical design drawings or 3D CAD models, a parametric 3D geometric model of the internal flow channel and heating area is established.

[0027] It should be noted that the parametric 3D geometric model is configured to parametrically represent the dimensions of key features.

[0028] The key feature dimensions specifically include the diameter and length of the variable diameter section and the distance from the heating zone to the nozzle throat. These key feature dimensions are used as the initial nozzle geometry parameters.

[0029] Furthermore, the parametric 3D geometric model is loaded with the material properties of the printing materials used, such as PLA and ABS.

[0030] The material properties specifically include thermophysical properties and rheological parameters.

[0031] The specific thermophysical parameters include: density, specific heat capacity, thermal conductivity, melting point, and latent heat of fusion.

[0032] Among them, rheological parameters specifically include model parameters that describe the change of melt viscosity with temperature and shear rate.

[0033] It should be noted that the initial model parameters can use a simplified modified power-law model: in, m Indicates melt viscosity, K 0 represents the consistency coefficient at the reference temperature, and exp represents an exponential function with the natural constant as the base. α Indicates the temperature sensitivity coefficient. T Indicates the actual temperature of the melt. Tref Indicates reference temperature. Indicates shear rate, n This represents the power-law exponent.

[0034] S202: Based on the parametric three-dimensional geometric model, construct a simplified mathematical model for simulating the internal thermal-fluid coupling state of the nozzle.

[0035] It should be noted that, in order to achieve rapid or real-time prediction, the established parametric three-dimensional geometric model needs to be simplified and reconstructed into a computationally efficient mathematical model. Using the lumped parameter method, the originally continuously distributed heat transfer and fluid flow process is discretized into a dynamic system described by a finite number of lumped parameters, thus forming a simplified mathematical model of the heat-fluid coupling state composed of ordinary differential equations and a system of algebraic equations.

[0036] The simplified mathematical model specifically includes: a simplified heat transfer model, a simplified flow model, and a heat-flow coupling relationship.

[0037] The simplified heat transfer model abstracts the continuous temperature field inside the nozzle into a network consisting of multiple, such as 3-5, thermal capacity nodes and thermal resistances. Thermal capacity nodes represent key areas such as the heating block, melting zone, and nozzle throat, while thermal resistance describes the heat transfer capacity between these areas. The nozzle heater power input, the heat carried away by melt convection, and the latent heat of phase change serve as the system's source and sink terms. Its core energy conservation relationship can be expressed as a system of ordinary differential equations: in, C i Indicates the first i The heat capacity of each node, T i Indicates the first i Temperature of each node, T j Indicates the first j Temperature of each node, R ij Indicates the first i The node and the first j Thermal resistance between nodes, P heat,i Indicates the first i Heating power of each node, Indicates mass flow rate, c p ∆ represents specific heat capacity. T Indicates temperature difference. L h Indicates latent heat of fusion. f This indicates the melting rate.

[0038] The simplified flow model simplifies the complex melt flow within the flow channel into a lumped parameter model based on one-dimensional pipe flow theory. The macroscopic relationship between extrusion flow rate, driving pressure drop, nozzle diameter, and melt apparent viscosity is established using this simplified flow model, which can be expressed in the form of a modified Hagen-Poiseuille equation. in, Q Indicates the extrusion volumetric flow rate. D Indicates the nozzle diameter, ∆ P Indicates driving pressure drop, L Indicates the length of the flow channel. m app ( ) indicates the apparent viscosity of the melt.

[0039] Among them, the heat-fluid coupling relationship refers to the coupling of the two simplified models mentioned above into a whole through key variables. The key variables specifically include: flow rate affecting heat transfer, temperature affecting flow, and the core role of aperture.

[0040] Among them, the influence of flow rate on heat transfer is as follows: mass flow rate (or volume flow rate) is the key input to the convection term and latent heat term in the heat transfer model, and directly determines the rate at which heat is carried away by the material.

[0041] Among them, temperature affects flow as follows: the melt temperature calculated by the heat transfer model is the core parameter for calculating viscosity in the flow model, and viscosity directly determines flow resistance.

[0042] Among them, the core role of the nozzle diameter is that it is a key geometric parameter in the flow model (directly affecting flow resistance and flow rate), and also a key factor affecting the residence time of the melt in the heating zone (thus affecting heat transfer).

[0043] It should be noted that the simplified mathematical model constructed in the above manner for simulating the internal thermal-fluid coupling state of the nozzle significantly reduces computational complexity while retaining the main dynamic characteristics of the physical process, thus laying the foundation for subsequent rapid or real-time prediction and control.

[0044] S203: Perform initial calibration and verification of the simplified mathematical model to obtain the digital twin model.

[0045] It should be noted that an initial calibration procedure should be performed at least upon first use or after replacing important components / materials. The initial calibration should be performed offline.

[0046] In one possible implementation, S203 specifically includes: S2031: Under typical operating conditions, run the 3D printer to a thermally stable state.

[0047] Thermal steady state refers to the state in which, under given nozzle diameter, extrusion speed and heating power settings, the temperature change rate of each monitoring point tends to zero after the system has been running for a sufficient time, and its value only fluctuates randomly within a small range around the set value. When the 3D printer is in thermal steady state, the temperature change of each monitoring point is ≤±1°C.

[0048] Typical operating conditions refer to the 3D printer operating at a small diameter and low speed.

[0049] S2032: Obtain initial calibration data of the 3D printer under thermal steady-state conditions.

[0050] The initial calibration data specifically includes: nozzle heater power, multi-point temperature, and extrusion speed.

[0051] S2033: Adjust the key adjustable parameters in the mathematical simplification model based on the initial calibration data.

[0052] The key adjustable parameters include: equivalent thermal resistance and heat loss coefficient.

[0053] It should be noted that the initial data is imported into the mathematical simplification model, and the key adjustable parameters in the mathematical simplification model are adjusted so that the steady-state temperature distribution output by the mathematical simplification model is within the preset tolerance of the measured temperature.

[0054] Optionally, the preset tolerance is ±5°C.

[0055] S2034: Verify the adjusted simplified mathematical model to obtain the digital twin model.

[0056] It should be noted that small-amplitude dynamic perturbation tests were conducted on the digital twin model to verify whether the dynamic trend predicted by the digital twin model is consistent with the actual response trend.

[0057] Furthermore, by comparing the model output with the measured response of the physical nozzle under the same disturbance, the model's dynamic tracking capability and prediction accuracy under typical operating conditions such as variable diameter and variable flow rate are verified, ensuring that the model can truly reflect the dynamic characteristics of the nozzle's thermal-fluid coupling and meet the requirements for real-time prediction and feedforward control.

[0058] In this embodiment of the invention, through lightweight and parametric design and scientific calibration and verification, the computational complexity is significantly reduced while ensuring prediction accuracy. Prediction output can be completed within 100ms to 500ms, perfectly adapting to the real-time temperature control requirements of variable-diameter 3D printing. At the same time, the mathematical model is simplified through parametric geometric modeling and lumped parameter method, which not only retains the core physical characteristics of nozzle thermal-fluid coupling, but also avoids the tedious and time-consuming high-fidelity full-physics field simulation. After offline initial calibration and dynamic disturbance verification, it is ensured that the model can truly reflect the dynamic response of the nozzle under conditions such as variable diameter and variable flow rate. This provides accurate and reliable model support for subsequent online calibration, model prediction, feedforward control and adaptive iteration, effectively improving the temperature control accuracy, extrusion uniformity and printing quality of variable-diameter 3D printing, while reducing the risk of hardware failure and achieving stable and efficient operation of the system.

[0059] S3: Based on the nozzle status data, calibrate the parameters in the digital twin model.

[0060] In one possible implementation, S3 specifically includes: S301: Define the vector of parameters to be estimated in the digital twin model.

[0061] The parameter vector to be estimated specifically includes: the equivalent heat transfer coefficient from the nozzle heater to the melt, the total heat loss coefficient of the nozzle to the environment, and the apparent viscosity coefficient of the material in the current state.

[0062] S302: Construct the observation equation.

[0063] Specifically, in each control cycle, the preprocessed measured temperature is used as the system observation. The predicted temperature of the digital twin model under the previous state and the current input is expressed as a function of the parameters to be estimated, and used as the predicted observation.

[0064] S303: Based on the observation equation, the estimated value of the parameter vector to be estimated is calibrated by the recursive least squares method.

[0065] Among them, recursive least squares is an identification method that can optimize model parameters in real time, successively, and online during system operation. It does not need to store all historical data. Instead, it only makes an incremental update based on the parameter estimation results of the previous moment when a new set of sampled data is obtained, thereby continuously correcting the model parameters to make them closer to the actual physical characteristics. At the same time, the weight of new and old data can be flexibly adjusted through the forgetting factor, so that the algorithm can smoothly track the slow changes in parameters and respond quickly when the operating conditions change abruptly. It is particularly suitable for online calibration and adaptive parameter optimization of digital twin models.

[0066] Specifically, based on the prediction error at the current moment, the estimated value of the parameter vector to be estimated is updated using the recursive least squares method. By introducing a forgetting factor, the update process is made to focus more on recent observation data, thereby enabling the tracking of the slow time-varying characteristics of the parameters.

[0067] The calibration method for the estimated value is as follows: in, k Indicates the index of the control cycle. i hat ( k ) indicates the first k The estimated values ​​of the parameter vector to be estimated for each control cycle. i hat ( k -1) indicates the first k -1 estimated values ​​of the parameter vector to be estimated for the control cycle. K ( k ) indicates the first k The gain matrix for each control cycle, e ( k ) indicates the first k The prediction error for each control cycle, y ( k ) indicates the first k System observations for each control cycle, This means that given the parameter to be estimated i In the case of the first k Predicted observations for each control period.

[0068] It should be noted that the obtained new parameter vector is updated to the corresponding parameters in the digital twin model. Simultaneously, physical constraints are applied to the updated parameters to prevent parameter estimation divergence due to sensor noise or transient data anomalies, thus ensuring that the model always conforms to physical laws. These physical constraints can include requirements such as ensuring the equivalent heat transfer coefficient is positive and the apparent viscosity coefficient is within a reasonable range of material properties.

[0069] Furthermore, the calibration process is subject to quality monitoring and adaptive processing. During system operation, the calibration quality and system status are monitored in real time, and the calibration strategy and control behavior are adaptively adjusted based on the monitoring results. Specifically, this includes calibration quality assessment and anomaly handling based on prediction errors, hardware and material anomaly warnings based on parameter mutations, and adaptive switching of calibration strategies based on operating condition identification.

[0070] Among them, the calibration quality assessment and anomaly handling based on prediction error are achieved by calculating the root mean square error of the prediction error in real time within the sliding time window, where the sliding time window is the past... NOne control cycle, N It is a positive integer greater than or equal to 3, optional. N =10, when the root mean square error value is in continuous M When the root mean square error value remains consistently below the excellent threshold within a control cycle, the digital twin model is considered to be in a highly calibrated state, and its prediction results can be adopted with high confidence by subsequent control modules; when the root mean square error value remains consistently below the excellent threshold, the model is considered to be in a highly calibrated state, and its prediction results can be adopted with high confidence by subsequent control modules. L If the error value remains above the abnormal threshold for an entire control cycle or if the error sequence exhibits significant non-convergent oscillations, the calibration is deemed abnormal. In this case, parameter updates are stopped and the model parameters are fixed to the values ​​of the most recent stable state. At the same time, the feedforward compensation weights are reduced or temporarily removed so that the system mainly relies on feedback control such as PID to maintain stability. Abnormal data and context information are recorded to generate a diagnostic log, where the abnormal threshold is much higher than the good threshold.

[0071] Among them, the hardware and material anomaly early warning based on parameter mutation is achieved by monitoring the unit time change rate of the parameter to be estimated. When a parameter undergoes a mutation that exceeds physical expectations, it is determined that there is a potential hardware failure or material anomaly, and an early warning is triggered and the printing task is suspended to prompt the operator to check. In a specific embodiment, when the equivalent heat transfer coefficient drops by more than 30% in the last 5 control cycles or 1 second, the system issues an early warning message and suspends the task.

[0072] Among them, the adaptive switching of the calibration strategy based on working condition identification is achieved through real-time analysis of the actual nozzle diameter and extrusion speed. When both are in continuous operation... P One control cycle (e.g.) P When the absolute values ​​of the changes within (=8) are less than the corresponding steady-state thresholds, the system is considered to be in a steady-state condition. In this case, the update frequency of the recursive least squares method can be reduced, or the forgetting factor can be set to a larger value to smooth the parameter estimation. When the change of any parameter exceeds the corresponding transient threshold, the system is considered to be in a transient condition. The high-frequency update of the recursive least squares method is then resumed, and the forgetting factor is set to a smaller value to improve the algorithm's sensitivity, enabling the model parameters to quickly track the dynamic changes of the system. At the beginning of each control cycle, the sensor data that has been verified to be valid at the current moment is injected into the digital twin model as an accurate initial state, ensuring that the internal state of the model is consistent with that of the physical printer. This ensures that the advance prediction starts from a zero-error initial point and avoids error accumulation.

[0073] In this embodiment of the invention, through scientific parameter calibration and full-process quality monitoring, real-time online adaptive optimization of digital twin model parameters is achieved, ensuring that the model always closely matches the actual operating state of the nozzle. By clearly defining the parameter vector to be estimated and constructing reasonable observation equations, combined with the incremental update characteristics of the recursive least squares method, key parameters such as the equivalent heat transfer coefficient and heat loss coefficient can be quickly corrected without storing all historical data. Simultaneously, by leveraging the forgetting factor to flexibly adapt the weights of new and old data, it can both smoothly track slow parameter changes and quickly respond to sudden changes in operating conditions. By applying physical constraints, parameter estimation divergence is effectively avoided, ensuring... The fault model conforms to physical laws; through three major monitoring and processing methods—calibration quality assessment, hardware and material anomaly early warning, and adaptive switching of working conditions—it can not only judge the calibration effect in real time and handle calibration anomalies in a timely manner, but also provide early warning of hardware failures and material problems. It can also dynamically adjust the calibration strategy according to steady-state and transient working conditions, and with the state reset at the beginning of the control cycle, it avoids error accumulation. Ultimately, it continuously improves the prediction accuracy and reliability of the digital twin model, provides accurate parameter support for subsequent model prediction and feedforward control, ensures the stability and printing quality of the variable diameter 3D printing process, and reduces failure risk and maintenance costs.

[0074] S4: Obtain forward-looking instructions containing information on future nozzle diameter changes.

[0075] S5: Input the look-ahead command into the calibrated digital twin model and output the nozzle heat load change trend and nozzle temperature change trend.

[0076] In one possible implementation, S5 specifically includes: S501: Parse the look-ahead instructions to obtain the key feature sequence.

[0077] It should be noted that a look-ahead buffer can be set up to cache and parse future time-domain look-ahead instructions.

[0078] The future time domain is 1-4 seconds after the current moment, and the specific duration can be configured according to the printing speed, control cycle and computing power.

[0079] Among them, the forward-looking instruction is the planned printing instruction. The planned printing instruction refers to the complete sequence of control instructions generated by the slicing software or path planner to complete the printing of the predetermined 3D model. It includes the nozzle movement trajectory, extruder control signals and nozzle diameter change timing. Such as a G-code file, this instruction sequence is determined before printing begins, so that the control system can read the future information contained therein.

[0080] The key feature sequences include: planned caliber sequence, planned extrusion flow rate sequence, special event markers, and printhead movement speed.

[0081] Among them, the planned nozzle diameter sequence is a time series formed by the target nozzle diameter value for each future control cycle, with the target nozzle diameter value at the end of every 0.1 seconds.

[0082] Among them, the planned extrusion flow rate sequence is the equivalent mass flow rate time series obtained by converting the extrusion motor control command.

[0083] Among them, special event markers are used to identify command events such as retraction and travel that can cause sudden changes or interruptions in traffic.

[0084] The printhead speed can be used to indirectly assess the change in the residence time of the melt within the printhead.

[0085] S502: By using a calibrated digital twin model, dynamic forward simulation is performed on key feature sequences to obtain the predicted trajectory of key states.

[0086] The key state prediction trajectory specifically includes: temperature prediction curve, heat load prediction curve, and apparent viscosity prediction curve.

[0087] Among them, the temperature prediction curve is the prediction curve of the nozzle throat temperature.

[0088] The heat load prediction curve is used to determine the theoretical heating power required to maintain the predicted state.

[0089] The apparent viscosity prediction curve reflects the change in flow resistance.

[0090] It should be noted that the simulation step size in the forward simulation can be set to be equal to the control cycle to synchronize the simulation process with real time and complete a full prediction calculation within each control cycle. Alternatively, if computational resources permit, the simulation step size can be set to be less than the control cycle to improve simulation accuracy. At least thermal state prediction and flow regime and thermal load prediction should be performed during the forward simulation.

[0091] It should be noted that for thermal state prediction, at each simulation step, the model solves the simplified heat balance equation based on the current virtual temperature, the current step input (aperture, flow rate) and material properties, and calculates the temperature distribution for the next step. It should predict the change trajectory of the nozzle throat predicted temperature and the heating block core predicted temperature.

[0092] It should be noted that for flow regime and heat load prediction, the flow model calculates the pressure drop and actual extrusion capacity based on the orifice diameter and viscosity (determined by the predicted temperature). Additionally, the heat load is configured for real-time calculation; the heat load refers to the theoretical heating power required to maintain the current planned flow rate and predicted temperature, estimated by the following formula: in, Indicates the predicted heat load, Indicates the planned quality flow rate. c p Indicates specific heat capacity. This indicates the predicted nozzle temperature. T room Indicates room temperature. L h Indicates latent heat of fusion. f melt This indicates the currently predicted melting rate. P loss This indicates the predicted heat loss. T pred Indicates the predicted temperature. T amb Indicates ambient temperature.

[0093] S503: Perform uncertainty quantification and credibility assessment on the predicted trajectory of critical states to obtain a comprehensive credibility score.

[0094] In one possible implementation, S503 specifically includes: S5031: Uncertainty of digital twin model after quantization calibration.

[0095] It should be noted that, based on the parameter estimation variance-covariance matrix provided by the calibration process, the confidence interval of the predicted temperature is estimated by Monte Carlo sampling or the first-order error propagation method.

[0096] S5032: Assess the certainty of forward directives.

[0097] It should be noted that the determinism of look-ahead instructions is comprehensively evaluated from three dimensions: instruction continuity, instruction stability, and instruction complexity.

[0098] Specifically, regarding instruction continuity, it is determined whether the instruction sequence in the lookahead buffer is continuous and uninterrupted in the future prediction time domain. If instruction is missing due to data transmission delays or other reasons, the determinism of instructions after the interruption point is reduced. In actual evaluation, a binary marking method can be used, with continuous instructions marked as "1" and interruptions marked as "0".

[0099] Specifically, regarding command stability, the command sequence read in the current control cycle is compared with the command sequence at the same time in the previous control cycle. If the change in command parameters at the same time index exceeds the stability threshold, the command is deemed unstable and the deterministic score is deducted accordingly. Command parameters include planned caliber and planned flow rate. When the caliber command changes by more than 0.2 mm, the deterministic nature of the command at that moment is considered to have decreased.

[0100] Specifically, regarding instruction complexity, the system analyzes and identifies whether the look-ahead instructions contain non-deterministic execution instructions such as conditional jumps, loops, or waits. Conditional jumps are instructions executed based on real-time sensor feedback, while the actual number of executions or duration of loop or wait instructions depend on the real-time system state. If the above non-deterministic instructions are identified, it is determined that the determinism of the instructions in the corresponding future time period is significantly reduced, and such instruction segments can be assigned a low baseline determinism value of 0.3.

[0101] Furthermore, after completing the above three aspects of evaluation, a time-series deterministic vector or a comprehensive scalar index is generated for the instructions in the prediction time domain based on the evaluation results. The comprehensive scalar index has a value range of 0-1 and can be obtained by assigning deterministic scores to each time point in the prediction time domain or by taking the lowest deterministic value in the entire time domain as the overall index.

[0102] S5033: Generate a comprehensive credibility score based on uncertainty and certainty.

[0103] It should be noted that a comprehensive credibility score with a value between 0 and 1 is generated by combining four factors: model calibration status, prediction dynamic rationality, input instruction determinism, and prediction uncertainty band width.

[0104] The model calibration status is obtained based on the output of calibration quality monitoring. Specifically, it is mapped based on the root mean square error of recent prediction errors and its stability. When the root mean square error is consistently below the good threshold, the model calibration status is set to 1.0. When the root mean square error is between the good and bad threshold and the abnormal threshold, the model calibration status is linearly interpolated between 0.5 and 0.9. When a calibration abnormality is triggered, the model calibration status is set to 0.2.

[0105] The predictive dynamic rationality is obtained by checking the physical rationality of the predicted temperature change trajectory. Specifically, the rate of change of the predicted temperature in each simulation step is calculated. If all rates of change are within the physical limits determined by the maximum power of the nozzle heater and the system heat capacity, the predictive dynamic rationality is set to 1.0. If there are any cases of exceeding the limits, the score is reduced accordingly based on the proportion and degree of exceeding the limits.

[0106] Among them, the input instruction determinism is formed based on the instruction determinism index.

[0107] The predicted uncertainty band width is obtained by mapping the calculated predicted uncertainty band width. Specifically, the average uncertainty band width is converted into a corresponding score. When the average width is less than 2°C, the predicted uncertainty band width is 1.0. When the average width is between 2 and 5°C, it is linearly mapped to 0.7-1.0. When the average width is greater than 5°C, the predicted uncertainty band width is less than 0.7.

[0108] Furthermore, the four scores are weighted and averaged to generate the final overall credibility score: in, C score This indicates the overall credibility score. w 1 represents the weighting coefficient. S cal Indicates the model calibration status score. w 2 represents the weighting coefficient. S dyn This indicates the score for the predictive dynamic reasonableness. w 3 represents the weighting coefficient. S ins Indicates the deterministic score of the input command. w 4 represents the weighting coefficient. S unc This indicates the score for the width of the prediction uncertainty band.

[0109] S504: Process the prediction results of the comprehensive reliability score to obtain the nozzle heat load change trend and nozzle temperature change trend.

[0110] Specifically, different prediction result application strategies are adopted according to different ranges of the comprehensive confidence score: when the comprehensive confidence score is greater than the high confidence threshold, the prediction trajectory is adopted with full confidence. The predicted temperature trajectory and predicted heat load are smoothed and directly transmitted to the control module. Based on the predicted heat load curve, the optimal heating power feedforward curve can be calculated in advance; when the comprehensive confidence score is between the low confidence threshold and the high confidence threshold, the prediction results are conservatively processed. This includes applying stronger low-pass filtering to the prediction trajectory, multiplying the predicted heat load change amplitude by an attenuation factor of 0.7, and only using predictions within a shorter time frame, such as 0.5 seconds in the future. As a result, model calibration can be performed simultaneously. When the overall confidence score is less than the low confidence threshold, a conservative control strategy is implemented. Specifically, the detailed results of this open-loop simulation are discarded, and only a very conservative feedforward is provided based on extremely simple empirical rules (e.g., a fixed power step based on the direction of aperture change within the next 0.2 seconds). The gain of the feedforward channel in the controller is significantly reduced, or even the feedforward function is temporarily disabled. The system stability is maintained mainly by feedback PID control. At the same time, fault events are recorded, and a degraded printing speed or a maintenance alarm is requested. It is understandable that this situation usually occurs when the model calibration continues to diverge, the sensor is abnormal, or the printing plan changes abruptly. In addition, when the look-ahead analysis detects special events such as pullback or idling, a dedicated compensation mode pre-calibrated for such events is activated. For example, when pullback is predicted to occur, the model will predict a sudden drop in flow and a rising temperature trend. At this time, the control module will proactively reduce the heating power in advance and introduce a reverse temperature overshoot setting to effectively prevent the occurrence of stringing.

[0111] Furthermore, the prediction results are encapsulated and output, packaged into a standard data structure.

[0112] The standard data structure specifically includes: a reference temperature trajectory for feedforward compensation, a basic feedforward power compensation curve, a prediction confidence level, and special event flags.

[0113] The reference temperature trajectory used for feedforward compensation is directly derived from the adjusted temperature prediction results. When the overall confidence score is high, the reference temperature trajectory is the original temperature prediction curve after smoothing; when the overall confidence score is medium, the reference temperature trajectory is a conservative trajectory after the original temperature prediction curve has been subjected to stronger filtering and amplitude attenuation (such as multiplying by an attenuation factor); when the overall confidence score is low, the reference temperature trajectory is set to a constant value (such as the current measured temperature) or a smooth transition trajectory, essentially abandoning the aggressive model prediction.

[0114] The basic feedforward power compensation curve is calculated based on the predicted heat load change trend and adjusted for confidence level. It is derived from the predicted heat load combined with parameters such as nozzle heater efficiency and system heat loss, reflecting the theoretical power change required to maintain the predicted temperature. Its amplitude is dynamically adjusted according to the prediction confidence level. For example, at a medium confidence level, the overall curve will scale proportionally; at a low confidence level, the curve amplitude may be significantly reduced or reduced to zero.

[0115] The predicted confidence level discretizes and qualitatively classifies the calculated comprehensive confidence score. Specifically, when the comprehensive confidence score is greater than or equal to 0.8, the corresponding predicted confidence level is determined to be high; when the comprehensive confidence score is greater than or equal to 0.5 and less than 0.8, the corresponding predicted confidence level is determined to be medium; and when the comprehensive confidence score is less than 0.5, the corresponding predicted confidence level is determined to be low.

[0116] The special event flag is directly derived from specific instruction codes or event markers identified during the parsing of lookahead instructions. When parsing the lookahead instruction sequence, if preset keywords or instruction codes such as pullback, idle shift, or layer switching are identified, this event type and the expected time of occurrence are encapsulated in the special event flag. For example, the special event flag can be set to a format such as RETRACT@t=1.2s, indicating that a pullback event will occur after 1.2 seconds.

[0117] In this embodiment of the invention, by using forward command parsing, forward simulation of the digital twin model, quantitative evaluation of uncertainty and credibility, and hierarchical adaptive post-processing, the thermal load and temperature change trend of the nozzle can be accurately predicted 1–4 seconds in advance. At the same time, a comprehensive credibility score is generated by combining multi-dimensional information such as command continuity, stability, complexity, and model calibration quality, so as to realize the hierarchical use of prediction results: at high credibility, the entire predicted trajectory is used to achieve accurate feedforward; at medium credibility, conservative processing is used to ensure stable control; at low credibility, downgraded reliance on feedback is used to ensure safety. A special compensation mode can also be enabled to suppress stringing for special events such as pullback and drift. Finally, the prediction results are encapsulated into a standard data structure for output. This not only gives full play to the advanced prediction capability of the digital twin model, but also avoids the control risks caused by invalid or erroneous predictions through the credibility mechanism, significantly improving the response speed, stability and robustness of temperature control, and providing reliable feedforward support for high-quality and high-consistency output of variable-diameter 3D printing.

[0118] S6: Generate heating power command based on the nozzle heat load change trend and nozzle temperature change trend.

[0119] In one possible implementation, S6 specifically includes: S601: Calculate the heating power feedforward compensation amount based on the nozzle heat load change trend and nozzle temperature change trend.

[0120] S602: Dynamically adjust the feedforward compensation amount of heating power based on confidence level prediction.

[0121] Specifically, based on the prediction confidence level, the heating power feedforward compensation is dynamically adjusted to obtain the final feedforward compensation: when the prediction confidence level is high, the final feedforward compensation directly adopts the basic feedforward compensation; when the prediction confidence level is medium, the final feedforward compensation is equal to the product of the basic feedforward compensation and a dynamic attenuation factor less than 1; when the prediction confidence level is low, the final feedforward compensation is approximately 0, at which point the system essentially degenerates into a control mode dominated by feedback control.

[0122] S603: Input the deviation between the target temperature and the measured temperature to the feedback controller and output the feedback power correction amount.

[0123] S604: The dynamically adjusted heating power feedforward compensation and feedback power correction are superimposed to obtain the heating power command.

[0124] In this embodiment of the invention, the precise and adaptive regulation of the nozzle heating power is achieved through the organic combination of feedforward compensation and feedback control. This fully utilizes the advantages of look-ahead prediction while effectively compensating for prediction deviations and system disturbances, ensuring temperature control accuracy. By calculating the feedforward compensation amount based on the nozzle heat load and temperature change trend, temperature changes can be predicted in advance and the power can be actively adjusted to avoid temperature overshoot or undershoot caused by lag. The feedforward compensation amount is dynamically adjusted based on the prediction confidence level. At high confidence levels, the feedforward is fully activated to achieve precise control, while attenuation control ensures stability at medium confidence levels. At low confidence levels, degradation-dependent feedback ensures safety and effectively avoids control risks caused by ineffective predictions. Then, the feedback controller outputs a correction amount based on the deviation between the target temperature and the measured temperature, compensating for uncertainties such as feedforward prediction deviations and environmental disturbances in real time. Finally, the feedforward compensation amount and the feedback correction amount are superimposed to generate a heating power command, forming a closed-loop control of "advance prediction and real-time correction". This significantly improves the stability and control accuracy of the nozzle temperature during variable-diameter 3D printing, reduces the impact of temperature fluctuations on extrusion uniformity and printing quality, and enhances the system's robustness to changes in operating conditions.

[0125] In one possible implementation, after S6 and before S7, the following is also included: The safety and physical constraints of the heating power command are checked.

[0126] Specifically, the safety constraint is the prevention of extruder stall.

[0127] The extruder stall prevention mechanism involves comprehensively assessing the risk of extruder stall by combining the predicted viscosity surge trend with real-time motor current feedback. If the predicted viscosity surges and the motor current exceeds the threshold, a short-duration temperature pulse command is superimposed on the final heating power command to actively and briefly increase the nozzle temperature to reduce melt viscosity. Simultaneously, the extruder is appropriately decelerated through a fine-tuning command for extrusion speed.

[0128] The physical constraints specifically include: power hard limiting and temperature rise rate limiting.

[0129] Among them, the power hard limit is: check whether the heating power exceeds the maximum allowable power of the nozzle heater, if... ,but And record the amplitude limiting event.

[0130] in, P represents the heating power at time t. max Indicates the maximum permissible power. This represents the final heating power at time t, and min indicates taking the minimum value.

[0131] The temperature rise rate limit is set based on the thermal inertia of the heating block, establishing a maximum safe temperature rise rate. The predicted temperature rise rate corresponding to the current control command is estimated through the model. If the predicted temperature rise rate... Greater than Then, it enters the slope limiting mode, and dynamically smooths the change curve of the final power command.

[0132] It should be noted that when a special event is identified as retraction, the system specifically performs the following coordinated control actions: 100-200 milliseconds before the start of the retraction action, the final heating power command is significantly reduced, for example, to 30%-50% of the steady-state value, to quickly cool the molten material at the nozzle tip and suppress stringing; at the same time, the end time of the retraction is predicted, and the heating power is gradually increased starting 50-100 milliseconds before the end of the retraction to ensure that the nozzle temperature is close to the target set point when printing resumes; in addition, the extrusion speed command and the retraction action maintain precise timing coordination to achieve synchronous and coordinated control of temperature, power and motion.

[0133] Furthermore, the system executes the following process for issuing, executing, and monitoring work control commands: The final heating power command and extrusion speed command, after multi-level processing, are issued to the nozzle heater power controller (usually using PWM drive) and the extrusion motor servo driver, respectively; at the same time, the system monitors the actual response of the actuators in real time, including parameters such as the nozzle heater operating current and the actual speed of the extrusion motor, and compares the actual response with the issued commands. If an actuator fault such as an open circuit in the nozzle heater or motor step loss is detected, the safety shutdown protocol is immediately triggered to stop the heating and extrusion actions, and the fault information such as the fault type and occurrence time is fully recorded. During operation, the system continuously monitors the effectiveness of feedforward compensation and makes adaptive adjustments. Specifically, after quantifiable planned disturbances such as a step change in caliber are completed, the feedforward effectiveness index is calculated. If the feedforward effectiveness index is greater than 70% for three consecutive evaluations, the feedforward compensation is considered to be effective, and the current feedforward strategy is maintained. If the feedforward effectiveness index is less than 30% for three consecutive evaluations, it indicates that the feedforward compensation is ineffective. At this time, the system will reduce the feedforward weight and send a "request for enhanced calibration" signal to the model calibration module to optimize the subsequent feedforward compensation effect.

[0134] In this embodiment of the invention, multiple constraint checks, special event collaborative control, and full-process command monitoring provide comprehensive protection for the safe, compliant, and effective execution of heating power commands, while achieving dynamic optimization of control effects: Extruder stall prevention constraints, combined with viscosity prediction and motor current feedback, actively superimpose temperature pulses and fine-tune extrusion speed to effectively avoid stall risks; power hard limiting and temperature rise rate limitation, two major physical constraints, prevent heating power from exceeding hardware limits and excessively rapid temperature rise from damaging the equipment, ensuring the physical safety of system operation; For special events like pullback, precise power timing control and extrusion speed coordination effectively suppress stringing and improve printing quality; furthermore, through command issuance, actuator response monitoring, and fault emergency handling, actuator faults are investigated in real time and a safe shutdown is triggered to prevent the fault from escalating; simultaneously, the effectiveness of feedforward is continuously monitored and the feedforward strategy is dynamically adjusted, and enhanced calibration is requested, continuously optimizing the control effect. This ensures that heating power commands comply with safety and physical specifications while improving the reliability and robustness of control, providing a solid guarantee for the stable and high-quality operation of variable-diameter 3D printing.

[0135] S7: Controls the temperature of variable diameter 3D printing according to the heating power command.

[0136] Furthermore, a status monitoring and adaptive iteration process is set up to monitor the actual operating performance and printing quality of the system in real time, quantitatively evaluate the control effect, and determine whether it is necessary to iterate to the previous process to optimize the model accuracy or adjust the control strategy based on the actual deviation, thereby forming a complete adaptive closed loop of perception-decision-execution-evaluation.

[0137] It should be noted that the system monitors multiple performance metrics in parallel during the control execution process.

[0138] The specific performance indicators include temperature tracking error, prediction accuracy, feedforward effectiveness, actuator health, sensor consistency, material extrusion uniformity, interlayer bonding state, and acoustic emission signal.

[0139] Among them, the temperature tracking error is calculated as the root mean square error and maximum deviation between the actual temperature and the target temperature within the past 2-second sliding time window.

[0140] Among them, the prediction accuracy index is the degree of agreement between the predicted temperature and the actual temperature after the fact, and the mean and variance of the prediction residuals are calculated.

[0141] Among them, the feedforward effectiveness index is to compare the actual temperature fluctuation amplitude with the predicted fluctuation amplitude under the condition of no feedforward after a known disturbance such as a step change in aperture occurs, and calculate the suppression percentage.

[0142] Among them, the actuator health status is to monitor whether the nozzle heater current is within the normal range and whether the motor load rate is abnormal.

[0143] Among them, sensor consistency is to check whether the differences between the readings of multiple temperature sensors are within a reasonable range.

[0144] Among them, material extrusion uniformity is assessed indirectly by integrating a vision sensor or by evaluating whether the extruded wire is uniform and continuous through the torque fluctuation of the extrusion motor.

[0145] Among them, the interlayer bonding state is measured by a high-precision ranging sensor attached to the printhead to measure the height consistency of the deposited layers. Abnormal fluctuations may reflect uneven layer height caused by temperature instability.

[0146] Among them, acoustic emission signal monitoring analyzes the audio signals generated during the printing process. Abnormal sounds at specific frequencies may correspond to stringing, material blockage, or interlayer delamination.

[0147] Furthermore, based on the aforementioned monitoring indicators, a comprehensive evaluation and grading process is conducted, along with preliminary root cause analysis. When the temperature tracking error or prediction accuracy exceeds the excellent threshold but falls below the warning threshold (e.g., the root mean square error of temperature is between 3°C and 5°C), the model is deemed to require fine-tuning. This situation is typically caused by slight drift in model parameters or batch differences in materials. In this case, the printing process is not interrupted; instead, the calibration frequency is temporarily increased, and the forgetting factor of the recursive least squares algorithm is reduced to make the calibration algorithm more sensitive to new data. Simultaneously, the feedforward gain in the control loop is appropriately reduced, and the proportion of feedback control is increased, allowing the system performance to recover to an excellent level within seconds to tens of seconds. When the temperature tracking error or prediction error continuously exceeds the warning threshold (e.g., the root mean square error of temperature is greater than 5°C and lasts for more than 5 seconds), or the feedforward effectiveness index remains below 50%, it is determined that execution is required. A more thorough model update is typically required, as this is often caused by model parameter mismatch, sensor performance degradation, or changes in actuator characteristics. In such cases, the model calibration process is re-executed, using the latest excitation and response data for batch least squares estimation or enhanced unscented Kalman filtering. Emphasis is placed on re-identifying suspected mismatch parameters such as rheological parameters and heat loss coefficients. The updated parameters are used to reconstruct the digital twin model, and the recalibration effect is verified using subsequent normal print data. Model mismatch is corrected and control performance is restored within 10-30 seconds. When severe temperature overshoot or undershoot occurs, prediction fails completely, or a hardware fault indicator is detected, it is determined that the fault has exceeded the adaptive adjustment range. A safety shutdown procedure is immediately initiated, cutting off heating, extrusion, and motion outputs. A fault cause prompt and detailed diagnostic report are generated by integrating multi-channel sensor logs and system status data, facilitating rapid fault location and troubleshooting.

[0148] Reference manual attached Figure 2 The diagram shows a schematic of the temperature control system for variable aperture 3D printing provided in an embodiment of the present invention.

[0149] The present invention also provides a temperature control system 20 for variable aperture 3D printing, applied to the above-mentioned temperature control method for variable aperture 3D printing, comprising: Processor 201.

[0150] The memory 202 stores computer-readable instructions, which, when executed by the processor 201, implement the temperature control method for variable-diameter 3D printing as described in the method embodiment.

[0151] The temperature control system 20 for variable diameter 3D printing provided by this invention can execute the temperature control method for variable diameter 3D printing described above and achieve the same or similar technical effects. To avoid repetition, this invention will not elaborate further.

[0152] This invention provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the temperature control method for variable-diameter 3D printing as described in the method embodiment.

[0153] The present invention provides a computer-readable storage medium that can implement the steps and effects of the temperature control method for variable aperture 3D printing described in the above method embodiments. To avoid repetition, the present invention will not repeat the details.

[0154] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A temperature control method for variable caliber 3D printing, characterized in that ,include: S1: Collect nozzle status data of the 3D printer; S2: Construct a digital twin model to simulate the internal thermal-fluid coupling state of the nozzle; S3: Based on the nozzle status data, calibrate the parameters in the digital twin model; S4: Obtain forward-looking instructions containing information on future nozzle diameter changes; S5: Input the forward-looking command into the calibrated digital twin model and output the nozzle heat load change trend and nozzle temperature change trend; S6: Generate a heating power command based on the nozzle heat load change trend and the nozzle temperature change trend; S7: Control the temperature of variable diameter 3D printing according to the heating power command.

2. The variable caliber 3D printed temperature control method of claim 1, wherein S2 specifically includes: S201: Based on the nozzle design data of the 3D printer, establish a parametric three-dimensional geometric model of the internal flow channel and heating area; S202: Based on the parametric three-dimensional geometric model, construct a simplified mathematical model for simulating the internal thermal-fluid coupling state of the nozzle; S203: Perform initial calibration and verification on the simplified mathematical model to obtain the digital twin model.

3. The variable caliber 3D printed temperature control method of claim 2, wherein S203 specifically includes: S2031: Under typical operating conditions, the 3D printer is operated to a thermally stable state; S2032: Obtain the initial calibration data of the 3D printer under the thermal steady state; S2033: Based on the initial calibration data, adjust the key adjustable parameters in the simplified mathematical model; S2034: Verify the adjusted simplified mathematical model to obtain the digital twin model.

4. The variable caliber 3D printed temperature control method of claim 1, wherein S3 specifically includes: S301: Define the vector of parameters to be estimated in the digital twin model; S302: Construct the observation equation; S303: Based on the observation equation, the estimated value of the parameter vector to be estimated is calibrated by recursive least squares method.

5. The temperature control method for variable aperture 3D printing according to claim 1, characterized in that... S5 specifically includes: S501: Parse the look-ahead instruction to obtain a key feature sequence; S502: Using the calibrated digital twin model, perform dynamic forward simulation on the key feature sequence to obtain the predicted trajectory of the key state; S503: Perform uncertainty quantification and credibility assessment on the predicted trajectory of the key states to obtain a comprehensive credibility score; S504: Process the prediction results of the comprehensive credibility score to obtain the trend of the nozzle heat load change and the trend of the nozzle temperature change.

6. The temperature control method for variable aperture 3D printing according to claim 5, characterized in that... S503 specifically includes: S5031: Uncertainties in the digital twin model after quantization calibration; S5032: Assess the certainty of the forward-looking instructions; S5033: Generate the comprehensive credibility score based on the uncertainty and the certainty.

7. The temperature control method for variable aperture 3D printing according to claim 1, characterized in that... S6 specifically includes: S601: Calculate the heating power feedforward compensation amount based on the nozzle heat load change trend and the nozzle temperature change trend; S602: The heating power feedforward compensation amount is dynamically adjusted based on confidence level prediction; S603: Input the deviation between the target temperature and the measured temperature to the feedback controller, and output the feedback power correction amount; S604: The dynamically adjusted heating power feedforward compensation amount and the feedback power correction amount are superimposed to obtain the heating power command.

8. The temperature control method for variable aperture 3D printing according to claim 1, characterized in that... After S6 and before S7, it further includes: The safety and physical constraints of the heating power command are checked.

9. A temperature control system for variable aperture 3D printing, characterized in that... ,include: processor; A memory storing computer-readable instructions, which, when executed by the processor, implement the temperature control method for variable-diameter 3D printing as described in any one of claims 1 to 8.

10. A readable storage medium, characterized in that... The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the temperature control method for variable-diameter 3D printing as described in any one of claims 1 to 8.

Images