An integrated system and method for die casting and finishing of precision aluminum casting motor housings

Abstract

This invention relates to the field of precision parts processing technology, and in particular to an integrated system and method for die casting and precision machining of precision aluminum casting motor housings. The system integrates a high-pressure die casting unit, an online three-dimensional scanning and detection unit, a five-axis precision machining center, and a central intelligent control system. Through intelligent scanning data preprocessing and thermal compensation modules, reflectivity noise elimination and thermal deformation inverse compensation are performed on the scanned point cloud of the highly reflective and temperature-rising casting blank to obtain a high-fidelity digital twin model. Based on this model, adaptive path planning is performed. A composite control algorithm integrating ring-coupled collaboration, fuzzy adaptive PID, predictive feedforward, and disturbance observer is adopted to drive a linear motor to achieve ultra-high precision machining. This invention improves accuracy from both the detection source and control execution ends, achieving high consistency of die-cast blanks, accurate online detection, and high precision of multi-axis collaborative machining, significantly improving the manufacturing accuracy, efficiency, and yield of motor housings.

Description

Technical Field

[0001] This invention relates to the field of precision parts machining technology, specifically to an integrated system and method for die casting and precision machining of precision aluminum casting motor housings. Background Technology

[0002] In fields such as new energy vehicles and precision instruments, the motor housing serves as a core load-bearing and sealing structural component, and its manufacturing quality directly affects the performance, efficiency, and reliability of the motor. These parts are typically thin-walled, complex aluminum castings, requiring extremely high dimensional accuracy, strict geometric tolerances (such as bearing hole coaxiality and end face parallelism), and internal quality (such as airtightness and absence of shrinkage cavities and porosity).

[0003] Currently, the manufacturing of such parts typically employs a process of high-pressure die casting followed by CNC precision machining. However, existing technologies have the following significant shortcomings: First, in the die casting stage, traditional gating systems and cooling designs easily lead to air entrapment and shrinkage porosity within the casting, and deformation during demolding. This results in low precision and poor consistency of key datum surfaces (such as internal stop surfaces) in the blank, causing significant fluctuations in allowances for subsequent precision machining. Second, in the machining stage, to meet high coaxiality requirements, the machine base and end cap are often machined separately as separate parts before assembly, making it difficult to control accumulated errors. Even when using a multi-axis machining center for single clamping, the traditional CNC system, employing a fixed-parameter PID control algorithm, suffers from problems such as response overshoot, tracking lag, and weak anti-interference capability when facing nonlinearity, end effects, and multi-axis linkage requirements in direct linear motor drive, making it difficult to achieve nanometer-level stable tracking of complex curved surfaces. Furthermore, in achieving adaptive machining closed loop, online 3D scanning data is relied upon as the basis for path planning. However, the high reflectivity of the aluminum casting surface leads to a large amount of mirror noise and data loss in the scanning point cloud. At the same time, the residual temperature after casting demolding and the ambient temperature gradient will cause non-uniform thermal deformation, resulting in severe distortion of the original scanning data and forming a vicious cycle of "garbage in, garbage out," which greatly limits the upper limit of closed loop accuracy.

[0004] Therefore, developing an integrated manufacturing system and method that can improve the quality of blanks from the source of die casting, intelligently compensate for the distortion of detection data, and have ultra-high dynamic precision control capabilities has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide an integrated system and method for die casting and precision machining of precision aluminum casting motor housings.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an integrated system for die casting and precision machining of precision aluminum casting motor housings, comprising: The high-pressure die-casting unit is used to form casting blanks with a high-precision internal stop as the initial process reference. An online 3D scanning and inspection unit is used to perform geometric scanning and simultaneous temperature field acquisition on the casting blank. The five-axis precision machining center is equipped with feed axes directly driven by linear motors and a high-frequency electric spindle, and is used for precision machining of castings; The central intelligent control system is communicatively connected to the high-pressure die-casting unit, the online three-dimensional scanning and detection unit, and the five-axis precision machining center, respectively. The central intelligent control system includes: The intelligent scanning data preprocessing and thermal compensation module is used to perform reflectivity noise compensation and thermal deformation compensation on the raw point cloud data acquired by the online 3D scanning detection unit, and output a thermally compensated digital twin model. An adaptive path planning module is used to dynamically generate a processing path based on the comparison results between the thermally compensated digital twin model and the ideal digital twin model, with the highest accuracy benchmark confirmed by scanning as the unified benchmark. The composite motion control module is used to drive and control each motion axis of the five-axis precision machining center to perform high-precision collaborative machining according to the machining path.

[0007] In some embodiments, the intelligent scanning data preprocessing and thermal compensation module includes: The reflectivity compensation submodule is configured to: acquire point cloud data of the casting in at least two different spectral bands or polarization directions; identify and mark specular reflection noise points based on the normal difference or intensity difference of corresponding points between each data source; perform multi-source data weighted fusion on non-noise points; perform neighborhood surface fitting interpolation on noise points to generate reflectivity-compensated point cloud. The thermal deformation compensation submodule is configured to: synchronously acquire the real-time temperature field distribution on the surface of the casting; input the real-time temperature field into the casting thermal deformation reduction model pre-integrated in the digital twin to predict the thermal deformation displacement field; and perform an inverse geometric transformation on the point cloud after reflectivity compensation, that is, subtract the predicted displacement vector of the corresponding position from the coordinates of each point to generate the thermally compensated digital twin model.

[0008] In some embodiments, the composite motion control module adopts a hierarchical control architecture, including: The collaborative layer employs a ring-coupled controller to receive the commanded position and actual position of each axis, calculate and output compensation signals for multi-axis synchronization; In the tracking layer, each axis independently uses a fuzzy adaptive PID controller as the core tracking controller, and its PID parameters can be adaptively adjusted online according to the tracking error and its rate of change. The feedforward compensation layer, including a disturbance observer and a target trajectory prediction feedforward controller, is used to perform feedforward compensation for the system's equivalent disturbance and known future trajectory information; The final control command for each axis is obtained by combining the compensation signals output by the tracking layer, the feedforward compensation layer, and the coordination layer.

[0009] In some embodiments, the control law of the loop coupling controller is:

[0010] in, For the first Synchronization compensation signal for the shaft, For the first Tracking error of the axis and The first Synchronization error between the shaft and adjacent shafts and This is the coupling gain coefficient.

[0011] In some embodiments, the fuzzy adaptive PID controller tracks the error. and error change rate As input, the proportional coefficient is output through fuzzy inference. Integral coefficient Differential coefficients Adjustment amount , , Its fuzzy rule base is based on the following experience: When | When it is large, increase , reduce and restrictions ; When |e| and | When it is of medium size, decrease and take appropriate and ; When | When it is small, increase , and according to | | Adjustment To suppress oscillations.

[0012] In some embodiments, the mold of the high-pressure die-casting unit includes: The central annular stepped gating system has its gating sleeve located in the central axis area of ​​the mold, and the molten aluminum fills the cavity through evenly distributed fan-shaped ingates. The hydraulic push structure integrated inside the slider is used to actively push against the key parts of the casting during mold opening and core pulling to suppress demolding deformation. A multi-loop water-cooled and high-pressure point-cooled system based on CAE analysis.

[0013] In some embodiments, the online 3D scanning detection unit includes a high-precision structured light or laser scanner and an array of non-contact infrared temperature sensors.

[0014] To achieve the above objectives, the present invention also provides the following technical solution: an integrated method for die casting and finishing of precision aluminum casting motor housings, applied to the aforementioned system, characterized by comprising the following steps: S100: A casting blank is formed by high-pressure die casting, and a high-precision internal stop is formed in it as the initial process reference. S200: Performs online 3D scanning and intelligent data compensation on casting blanks to generate a thermally compensated digital twin model; S300: Compare the thermally compensated digital twin model with the ideal digital twin model, and dynamically generate the processing path using the highest precision benchmark confirmed by scanning as the unified benchmark; S400: Based on the machining path, a composite motion control algorithm is used to drive the five-axis precision machining center to perform the machining of the assembly in one clamping, completing the finishing of all key features.

[0015] In some embodiments, step S200 includes: S210: Performs multispectral or polarized light reflectance compensation, and eliminates point cloud noise and data loss caused by high reflectance of casting surface through multi-source data fusion and reconstruction. S220: Performs real-time thermal gradient mapping and finite element thermal deformation compensation. Based on the synchronously acquired temperature field, it predicts the displacement field through a thermal deformation reduced-order model and performs inverse compensation on the point cloud to eliminate geometric measurement errors caused by thermal deformation. S230: Output high-fidelity point cloud data after compensation by S210 and S220, as the digital twin model after thermal compensation.

[0016] In some embodiments, the composite motion control algorithm used in step S400 ensures the minimization of synchronization error between multiple axes through a loop coupling strategy, adjusts control parameters online through fuzzy adaptive PID to cope with processing disturbances, and performs feedforward compensation for system disturbances and known trajectories through a disturbance observer and predictive feedforward to achieve nanometer-level trajectory tracking accuracy.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. Achieving ultra-high machining accuracy and consistency: The intelligent scanning compensation module ensures the absolute accuracy of the machining datum and allowance model from the source; the composite control algorithm (integrating loop coupling, fuzzy adaptive PID, predictive feedforward and disturbance observer) guarantees nanometer-level trajectory tracking and multi-axis coordination. The key form and position tolerances of the final product (such as bearing hole coaxiality and end face parallelism) reach the leading level at the micrometer level.

[0018] 2. Excellent dynamic performance and stability: The fuzzy adaptive PID and loop coupling strategy significantly improve the system response speed, synchronization accuracy and anti-interference ability, effectively suppress disturbances such as linear motor end effect and sudden cutting force change, and ensure a smooth and oscillating high-speed and high-precision machining process.

[0019] 3. Supports intelligent and flexible production: The system can automatically adjust the control strategy and processing path based on high-fidelity online detection data and real-time processing feedback, adapting to the processing of shells of different batches and models, solving the problem of online measurement of hot workpieces, and greatly shortening the production cycle.

[0020] 4. Significantly improved technical indicators: By applying this solution, the accuracy of key reference surfaces of die-cast blanks is improved by more than 60%, the absolute accuracy of online scanning data is improved by more than 70%, the final product processing efficiency is improved by 40%, and the overall scrap rate is reduced to below 0.5%, fully meeting the stringent requirements of high-end applications for airtightness, accuracy and reliability.

[0021] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. The embodiments of this application will provide a detailed description and understanding of the application. Attached Figure Description

[0022] Figure 1 This is a block diagram of the hierarchical architecture of the composite motion control algorithm of the present invention; Figure 2 The block diagram of a fuzzy adaptive PID controller (including fuzzification, rule base, inference, and defuzzification modules); Figure 3 This is a schematic diagram of a ring-coupled collaborative control structure (n-axis closed loop). Figure 4 Workflow diagram for the intelligent scanning data preprocessing and thermal compensation software module; Figure 5 This is a schematic diagram of the multispectral reflectance compensation principle (specular reflection noise identification). Figure 6 This is a schematic diagram of the finite element compensation principle for thermal deformation (temperature field input, displacement field output, point cloud inverse correction). Figure 7The graph shows a comparison of the trajectory tracking errors between the algorithm of this invention and the conventional PID algorithm when processing elliptical contours. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] This invention provides a manufacturing system with deep hardware and software integration and intelligent closed-loop operation. Its core innovation lies in: 1) an innovative die-casting mold design that provides a highly consistent reference blank for precision machining; 2) An intelligent scanning data preprocessing and thermal compensation software module ensures the absolute reliability of the detection data from the source; 3) A composite motion control algorithm integrating fuzzy adaptive PID, loop coupling collaboration and predictive feedforward serves as the execution guarantee for high-precision machining; 4) The above units are seamlessly integrated through a central intelligent control system to form a complete closed loop of "perception-decision-execution".

[0025] The present invention will be described in detail below through several embodiments.

[0026] Example 1: Overall Structure and Collaborative Workflow of an Integrated System This embodiment describes in detail the overall structure of the integrated system and the collaborative workflow between its various parts.

[0027] See Figure 1 The system block diagram shown illustrates that the integrated die-casting and finishing system for precision aluminum casting motor housings mainly comprises four physical units and a core control center: High-Pressure Die Casting Unit: The core of this unit is a die casting mold with an innovative structure (see Example 4 for details), used to inject molten aluminum alloy under high pressure to form the casting blank of the motor housing. One of the key designs of this invention is the pre-forming of a high-precision internal stop in the mold cavity. This internal stop is manufactured during the die casting stage, serving as the initial process reference for all subsequent processes, fundamentally avoiding errors caused by reference conversion.

[0028] Online 3D scanning inspection unit: This unit is located after the die-casting unit and before the machining unit, and is used for non-destructive testing of casting blanks that are still in a "hot" or "room temperature" state. It includes: High-precision structured light / laser scanner: used to quickly acquire three-dimensional point cloud data of casting surfaces to characterize their geometric morphology.

[0029] Array-type non-contact infrared temperature sensor: works synchronously with a scanner to acquire the temperature field distribution matrix of the casting surface. .

[0030] This unit collects raw geometric point cloud and temperature field data and uploads them to the central intelligent control system in real time.

[0031] Five-axis precision machining center: This is the physical platform for performing final finishing. Its innovation lies in the drive method: Feed axes (X, Y, Z axes): Directly driven by linear motors, eliminating the need for mechanical conversion devices such as ball screws, and featuring high acceleration, high speed and high rigidity, providing the hardware foundation for ultra-high dynamic response.

[0032] Rotary axes (A and C axes): These are typically driven directly by torque motors to achieve positioning at complex angles.

[0033] High-frequency electro-hydraulic spindle: provides cutting power and ensures the quality of machined surfaces.

[0034] The machining center receives instructions from the central intelligent control system to complete precision machining such as milling and boring of all key features (such as bearing holes, end faces, mounting surfaces, etc.).

[0035] Central Intelligent Control System: This is the "brain" of the entire integrated system, communicating with the three physical units mentioned above. It is not a simple PLC or CNC system, but an industrial computer cluster integrating a process database and intelligent algorithm modules. Its core comprises three functional modules: Intelligent scanning data preprocessing and thermal compensation module: Receives raw data from the online detection unit, executes software algorithms to eliminate specular reflection noise and thermal deformation errors, and outputs a high-fidelity "thermally compensated digital twin" model. .

[0036] Adaptive path planning module: By comparing the part with a digital twin model of the ideal part stored in the system, the machining allowance and geometric errors of each part are automatically identified. Then, using the actual reference with the highest accuracy confirmed by online scanning (such as axis A and end face B formed by the die-cast inner stop) as a unified reference, an optimized CNC machining path (G-code) is dynamically generated. This achieves flexible machining "based on the part", rather than rigidly machining according to theoretical coordinates.

[0037] Composite Motion Control Module: Receives machining path instructions generated by the adaptive path planning module and converts them into real-time control commands for the five motion axes of the five-axis precision machining center. This module employs an advanced hierarchical composite control algorithm (see Example 3) to ensure that each axis can accurately track its own instructions during machining while maintaining strict synchronization, and simultaneously strongly suppresses external interference such as cutting forces.

[0038] System workflow: The automated operation of the entire system follows the closed-loop process described below: S100 (Die Casting): The high-pressure die casting unit operates to produce casting blanks with high-precision internal stops.

[0039] S200 (Online Scanning and Intelligent Compensation): A robotic arm transfers the raw material to the online scanning station. The scanning and detection unit simultaneously collects point cloud and temperature data and uploads them. The intelligent scanning data preprocessing and thermal compensation module in the central control system immediately begins to perform in-depth processing on the raw data, generating reliable... Model.

[0040] S300 (Path Planning): Comparison of Adaptive Path Planning Module Based on the ideal model and actual benchmarks, a personalized processing path is dynamically generated for the specific blank.

[0041] S400 (Intelligent Machining): The workpiece is clamped into the five-axis machining center (one-time clamping). The composite motion control module takes over control, driving each axis to execute the "combined machining, datum transfer" strategy. That is, starting from the actual datum confirmed by scanning, all key features are continuously machined in one clamping. The datum is naturally "transferred" and maintained during machining, avoiding repeated clamping errors. The status data of the machining process (such as actual position and current) is fed back to the control module in real time, forming a closed-loop control.

[0042] Thus, Embodiment 1 fully illustrates the hardware configuration, core software modules, and overall workflow of the system, constituting the top-level system integration scheme of this invention.

[0043] Example 2: Detailed Implementation of the Intelligent Scanning Data Preprocessing and Thermal Compensation Module This embodiment elaborates on the specific algorithms and implementation methods of the intelligent scanning data preprocessing and thermal compensation module. This module is crucial for ensuring the accuracy of the adaptive processing "sensing" stage, solving the industry problem of "garbage data in, garbage instructions out."

[0044] This module can be divided into two sequentially executed and interconnected sub-modules in the software: the reflectivity compensation sub-module and the thermal deformation compensation sub-module.

[0045] 1. Implementation of the reflectivity compensation submodule (corresponding to step S210) This submodule is designed to eliminate optical scanning noise (specular reflection bright spots, missing data) caused by the highly reflective surface of aluminum castings.

[0046] Data Acquisition: To obtain multi-source information, the scanner configured in the system should support multispectral scanning or polarization scanning modes.

[0047] Multispectral mode: Controls the scanner to use at least two different wavelengths of light source sequentially or simultaneously (e.g., =450nm blue light, Two sets of point cloud data were obtained by scanning the same casting using near-infrared (850nm) light. and Different wavelengths of light interact differently with metal surfaces, and the areas affected by specular reflection will show inconsistencies between the two sets of data.

[0048] Polarization mode: The scanner emits two orthogonal polarization directions ( , The light is collected and the corresponding reflected light intensity is received to obtain... and Specular reflection can significantly alter the polarization state of light, thus producing differences in the two polarization data.

[0049] Mirror noise recognition: The core is calculating the difference. .

[0050] For multispectral data, calculate the same physical point The difference between the estimated normal vectors in the two point clouds: In the diffuse reflection region, the normal vector should be basically consistent; in the specular reflection region, due to data distortion, the normal vector will change drastically.

[0051] For polarization data, calculate the intensity difference: .

[0052] The system presets an experience threshold. Traverse all points, if > Then determine the point. Noise points that are severely contaminated by specular reflection are marked.

[0053] Determining the mirror noise recognition threshold, the empirical threshold. The determination method is as follows: During the system calibration phase, multiple sets of multispectral or polarization point cloud data are collected using a standard diffuse reflection sphere and an aluminum sample known to have high reflectivity characteristics. The normal difference between all corresponding point pairs is then calculated. Δn or intensity difference ΔI The statistical distribution of . Set to be calculated on diffuse reflectance samples Δn or ΔI The mean plus three standard deviations ( This ensures that normal surface variations can be distinguished from specular reflection noise with a confidence level greater than 99.7%.

[0054] Data fusion and reconstruction: For non-noise points: weighted fusion is used. Weight The confidence level of the data source is inversely proportional to its reliability in this context, and this reliability can be evaluated using metrics such as local point cloud density and intensity signal-to-noise ratio. This improves the accuracy of the valid data.

[0055] For marked noise points: an algorithm based on neighborhood geometry is used for filling. For example, using the unmarked points around the point, a local surface is fitted using moving least squares (MLS), and then the coordinates of the filled point are interpolated on the surface.

[0056] Simplified mode: If the hardware does not support multispectral / polarization, the alternative algorithm is enabled. For a single-source raw point cloud, the local density of each point is calculated. and normal rate of change Mirror noise points typically exhibit density anomalies or abrupt changes in normal direction. Removing outliers through joint thresholding followed by MLS smoothing of the overall point cloud can suppress noise to some extent, but the effect is inferior to multi-source fusion methods.

[0057] The final output of this process is a point cloud after reflectivity compensation. .

[0058] Multi-source data fusion weight quantization, where the weight w_i is calculated based on the local confidence of each data source at that point. Confidence level From local point cloud density and intensity signal-to-noise ratio Joint decision: ,in and Weighting coefficients (usually taken as...) =0.6, =0.4), Let be the normalization function. Then the weights For fitting the neighborhood surface of noise points, the moving least squares method is used, and the fitting radius is dynamically adjusted according to 3-5 times the average point distance in the region. 2. Implementation of the thermal deformation compensation submodule (corresponding to step S220) This submodule is designed to compensate for thermal deformation of castings caused by uneven temperature, restoring the measured data to the geometry at a standard temperature (such as 20°C).

[0059] Simultaneous temperature field acquisition: While scanning the geometry, the array-type infrared temperature sensor has acquired the temperature distribution matrix of the casting surface. .

[0060] Digital Twin and FEA Model Integration: A parametric finite element analysis (FEA) model of the motor housing for this model is pre-integrated within the system's digital twin environment. This model is based on physical equations. Transient heat conduction equation: It is used to simulate the transfer of temperature fields.

[0061] Thermoelastic equation: , , Used to calculate the temperature field Caused thermal strain and the final displacement field .

[0062] in, , , , For material parameters, This is a reference temperature.

[0063] Reduced-order model (ROM) training: Due to the excessive time consumption of full-size online FEA calculations, this invention adopts an offline training and online invocation strategy. Before system deployment, FEA software is used to perform a large number of sample calculations on the digital twin model: various possible thermal states (such as uniform heating, gradient heating, local hot spots, etc.) are input to obtain the corresponding displacement fields. Using these input-output sample pairs, a reduced-order model (ROM) is trained, such as a response surface model based on a Kriging surrogate model or intrinsic orthogonal decomposition (POD). This ROM can achieve millisecond-level processing of temperature field data. to displacement field Mapping prediction: .

[0064] Construction process of the reduced-order model: The training of the thermal deformation reduced-order model is carried out according to the following steps: Sample generation: Based on the parametric FEA model, various cooling scenarios were set for the casting from demolding temperature (e.g., 150°C-250°C) to room temperature (20°C), and local hot spots were introduced to simulate temperature unevenness, generating a total of M groups (e.g., M=500) of representative temperature field distributions. and its corresponding full model displacement field .

[0065] Orthogonal Eigenmode Decomposition (POD): Perform POD analysis on a displacement field sample set to extract the top K (e.g., K=20) most dominant eigenmodes (POD basis functions). This allows these basis functions to capture more than 99% of the sample energy (displacement variance).

[0066] Proxy model establishment: For each set of temperature field samples Calculate the projection coefficients of its displacement field onto the POD basis functions. Using the Kriging surrogate model, a mapping relationship is established from the input temperature field T to the projection coefficient vector a: The correlation function of the Kriging model is a Gaussian function, and its parameters are optimized through maximum likelihood estimation.

[0067] Model Validation: ROM accuracy is validated using a test sample set that was not used in training, requiring the root mean square error (RMSE) of displacement field prediction to be less than 5 micrometers.

[0068] [Online Access] During online measurement, the real-time temperature field will be used. Input the trained ROM to obtain the projection coefficients. And then through Fast reconstruction of predicted displacement field Real-time thermal deformation calculation and reverse compensation: During online measurement, the real-time temperature field is calculated. Input a trained ROM to quickly predict the thermal deformation displacement field caused by the current thermal state. Each point has a corresponding displacement vector. .

[0069] Then, the point cloud after reflectivity compensation Perform inverse geometric transformation: That is, the predicted thermal deformation displacement is "subtracted" from the measurement data to obtain the geometry that the casting should have at standard temperature.

[0070] The final output is a digital twin model after thermal compensation. .

[0071] S230 (Data Verification and Delivery): After generation, the module will quickly compare it with the ideal model, calculate the overall deviation, and if the deviation is within a preset reasonable range, the compensation is confirmed to be effective and sent to the adaptive path planning module. Otherwise, an alarm will be issued, prompting a check of the sensors or process status.

[0072] This embodiment clearly and practically protects the innovative software module of intelligent scanning data preprocessing and thermal compensation through specific algorithm formulas and implementation steps.

[0073] Example 3: Detailed Implementation of the Algorithm for the Composite Motion Control Module This embodiment elaborates on the core algorithm of the composite motion control module. This algorithm is the "execution nerve center" that ensures that step S400 achieves nanometer-level machining accuracy.

[0074] like Figure 1 As shown, this module adopts a hierarchical control architecture to perform coordinated control of the n motion axes (e.g., X, Y, Z, A, C axes) of a five-axis machining center.

[0075] 1. Overall Architecture and Signal Flow For the i-th control axis: Top layer (coordination layer): The ring-coupled controller receives position commands from all axes. and actual feedback location It does not directly output control signals, but instead outputs a compensation signal to coordinate the synchronous motion of each axis. This signal will be added to the control loop of each axis itself.

[0076] Middle layer (tracking layer): Each axis has an independent fuzzy adaptive PID controller. Its input is the tracking error of this axis (usually incorporating the compensation requirements of the cooperative layer), and its output is a preliminary control quantity. The PID parameters of this controller It is not fixed, but dynamically adjusted according to real-time error conditions.

[0077] The bottom layer (feedforward compensation layer) contains two parallel units: Disturbance Observer (DOB): Treats all uncertainties, such as the end effects of the linear motor, changes in model parameters, and sudden changes in cutting force, as an "equivalent disturbance". And estimate in real time .

[0078] Target trajectory prediction feedforward controller: Utilizes processing trajectory instructions generated by the adaptive path planning module, with a known future timeframe (N steps). Calculate a feedforward control quantity in advance This is to overcome the inertial lag of the system.

[0079] Final control command synthesis: final control commands for each axis The following formula is synthesized to obtain:

[0080] This formula integrates feedback regulation, feedforward compensation, and synergistic coupling to form a robust and highly accurate composite control law.

[0081] 2. Detailed Explanation of Key Algorithms a) Circular Coupling Collaborative Control Strategy Define the tracking error of axis i: .

[0082] Define the synchronization error between axis i and its adjacent axes (all axes are connected end-to-end to form a loop):

[0083]

[0084] The loop coupling control law is then:

[0085] in, , This is an adjustable coupling gain coefficient.

[0086] Physical meaning: This control law not only focuses on the tracking error of this axis ( ), and further through ( The feature uses the difference in synchronization error between adjacent axes as a powerful adjustment signal. Any deviation in speed or position of any axis will rapidly affect its adjacent axes through this loop network and be transmitted to the entire system, thereby forcing all axes to "move in step" and minimizing global synchronization error. This is particularly suitable for trajectory coordination in five-axis linkage machining.

[0087] b) Design of fuzzy adaptive PID controller Taking a single axis as an example, the operation steps of its fuzzy adaptive PID controller are as follows: Input and Fuzzification: The controller tracks the error and its rate of change As precise input. (e and...) The physical quantity values ​​are mapped to the standard fuzzy domain through a scaling factor, for example, [-3, 3]. The input and output linguistic variables are set as: {Negative Large (NB), Negative Medium (NM), Negative Small (NS), Zero (ZO), Positive Small (PS), Positive Medium (PM), Positive Large (PB)}. The triangular membership functions are used to calculate the membership degrees μ(e) and μ(ec) of e and ec to each fuzzy subset.

[0088] Fuzzy rule base: Based on expert experience in precision machining processes, three main categories containing 49 (7x7) fuzzy rules have been developed. The core empirical principles of the rule base are as follows: When |e| is large, the system requires strong correction, and the proportional gain should be significantly increased. ΔK p At the same time, reduce the differential action ΔK d To avoid oscillations and limit the integral effect ΔK i Prevent saturation.

[0089] When |e| and |ec| are equal, the system approaches the target but still has deviations and inertia, and should be appropriately reduced. ΔK p To reduce overshoot and provide appropriate ΔK i and ΔK d .

[0090] When |e| is very small, the system is in a steady-state fine-tuning phase, and should be increased. ΔK i To eliminate steady-state error, and adjust according to |ec| (i.e., oscillation trend). ΔK d To suppress shaking.

[0091] Fuzzy Rule Table: The complete fuzzy rule base of the fuzzy adaptive PID controller is shown in the table below (indicated by...). ΔK p For example, ΔK i and ΔK d The rule base structure is similar, and it is formulated based on the aforementioned empirical principles.

[0092] Table 1: ΔK p The fuzzy rule table (linguistic variables: NB negative large, NM negative medium, NS negative small, ZO zero, PS positive small, PM positive medium, PB positive large) ΔK i The adjustment tends to increase when the error is small to eliminate steady-state error, and limit when the error is large to prevent integral saturation; ΔK d The adjustment tends to increase when the error rate of change is large to suppress overshoot, and decrease when in steady state to avoid oscillation. Its specific rule table follows the same design logic for generation.

[0093] Table 1 provides the following information: ΔK p The complete fuzzy rule table is used as an example. ΔK i and ΔK d The rule table is formulated based on the same principles.

[0094] Fuzzy Inference and Defuzzification: The Mamdani fuzzy inference method is employed. For e and ec at the current sampling time, multiple rules are activated based on their membership degrees and the rule "IF e is A AND ec is B, THEN ΔKp is C" to obtain the output fuzzy set. Defuzzification is performed using the centroid method, converting the fuzzy output into precise adjustment values. The formula for the center of gravity method: ,in To output discrete points in the universe of discourse.

[0095] Online PID parameter update and output: The adjustment obtained from fuzzy inference is superimposed on the initial parameters.

[0096]

[0097]

[0098] Then, the PID output is calculated using the updated parameters: .

[0099] Thus, online adaptive adjustment of PID parameters to processing conditions was achieved.

[0100] c) Foresight Feedforward and Disturbance Observer (DOB) Design Foresight Feedforward: For a known machining trajectory (such as a non-circular cross-sectional profile), an optimal predictive servo controller is designed based on the discrete state-space model of the system, utilizing the instruction information R(k+1)...R(k+N) for the next N steps. This controller adds a feedforward term to the conventional state feedback. This term is a function of the future target value, which allows for advance action and greatly reduces tracking lag error.

[0101] Predictive Feedforward Controller Design: Selection of the Predictive Step Number N and System Sampling Period and dominant time constant Related, usually taken This ensures that the main dynamic processes of the system are covered. Target trajectory prediction feedforward control quantity. It is obtained by solving an optimal tracking problem in a finite time domain, the core of which is to utilize the discrete state-space model of the system. Calculate the feedforward gain matrix. : ,in It can be obtained through standard methods of generalized predictive control (GPC) or preview control theory.

[0102] Discrete-time perturbation observer (DOB): This observes the actual controlled object... With nominal model The differences between them (including parameter perturbations and external disturbances) are considered to act on equivalent perturbation at the input end The core idea of ​​DOB is to construct an observer to estimate... A classic implementation is as follows: .in, It is a low-pass filter, and its bandwidth determines the trade-off between the observer's ability to estimate disturbances and its robustness to model errors. The estimated... The feedforward compensation is directly fed into the control command, thereby canceling out disturbances before they affect the output.

[0103] Disturbance Observer (DOB) Design: The Low-Pass Filter A second-order Butterworth filter is selected: ,in This is the cutoff frequency. The choice of bandwidth requires a trade-off between perturbation estimation speed and robustness, and is typically set between 1 / 2 and 2 / 3 of the bandwidth of the nominal model of the system. Take the ideal second-order model of the linear motor , where parameters , , Identified through no-load motor testing.

[0104] This embodiment fully reveals the internal working principle of the composite control algorithm through mathematical formulas and logical descriptions, demonstrating its creativity in improving tracking accuracy, synchronization, and anti-interference capabilities.

[0105] Example 4: Innovative Mold Design for High-Pressure Die Casting Unit This embodiment details an innovative die-casting mold design that provides a highly consistent blank for finishing. These designs form the basis for step S100 to establish a high-precision initial reference.

[0106] The mold of the high-pressure die-casting unit includes the following three innovative structures, which are used in combination to achieve high-precision, low-deformation die-cast blanks: Central Annular Stepped Gating System: Traditional lateral or tangential gating systems often lead to disordered aluminum molten material filling and severe air entrapment. This system places the gating sleeve in the central axis region of the mold parting surface. The molten aluminum flows vertically downwards or upwards from the center, is buffered by an annular stepped runner, and then smoothly and synchronously fills the cavity in a nearly radially symmetrical manner through multiple fan-shaped ingates evenly distributed around the circumference. This design significantly reduces the flow distance difference and mutual impact of the molten aluminum, ensuring a uniform temperature field in all areas of the cavity, reducing defects such as air entrapment and cold shuts, and ensuring full filling and dimensional stability of the reference inner stop area. Design parameters for the central gating system: Total cross-sectional area of ​​the fan-shaped ingate With casting weight W and filling time Satisfying Relationship: ,in The density of molten aluminum, The recommended filling rate is 30-50 m / s. There are 6-8 ingates, evenly distributed around the central runner to ensure symmetrical flow.

[0107] Integrated hydraulic push-back structure within the slider: Motor housings often have deep cavities, thin walls, or irregular shapes. During mold opening and core pulling, the casting is prone to sticking to the slider due to clamping force, leading to deformation. This invention integrates a small hydraulic push-back mechanism within the slider that forms these complex features. At the start of the mold opening action, this mechanism does not immediately retract with the slider, but instead actively pushes one or more push rods forward (towards the cavity) via a miniature hydraulic cylinder, briefly "holding" the critical rigid parts of the casting. After the main slider moves a short distance and the clamping force between the casting and the slider core is initially released, the push-back mechanism retracts, and the slider continues to complete the core pulling. This process is like "handling demolding," effectively suppressing plastic deformation of the casting at the moment of demolding. Hydraulic thrust reverser design: the jacking force of the hydraulic thrust reverser mechanism It needs to be greater than the estimated clamping force of the slider core on the casting. . The shrinkage stress of the casting can be estimated using CAE simulation and the contact area between the core and the mold core, or based on empirical formulas. Calculation, where The unit area clamping force coefficient (for high-pressure die casting of aluminum alloys, k≈3-5 MPa). This represents the contact area between the core and the casting. Ejection stroke. The initial core-pulling displacement of the slider is 20%-30% to ensure that the initial tightness is released.

[0108] A multi-loop water-cooling and high-pressure spot cooling system based on CAE analysis: The thermal balance of the mold directly affects the casting cooling rate, shrinkage cavity location, and dimensional accuracy. This invention utilizes computational fluid dynamics (CAE) software to simulate the die-casting process during the mold design phase, accurately predicting hotspot areas. Based on this: Multi-loop water cooling system: Multiple independent circulating cooling water channels are arranged in the mold core and mold frame. Each loop has independent flow and temperature control to achieve differentiated temperature management for different areas of the mold.

[0109] High-pressure point cooling system: High-pressure point cooling nozzles are embedded in particularly thick or hot spots prone to shrinkage cavities. Immediately after the aluminum liquid is filled, high-pressure cooling media (such as water mist or oil) are sprayed onto these specific points for forced rapid cooling, quickly eliminating hot spots and ensuring the density of the microstructure.

[0110] CAE-guided precise cooling ensures the mold is in optimal thermal equilibrium, resulting in uniform solidification of the casting and minimal deformation. This provides a high-quality blank with uniform allowance and low internal stress for subsequent finishing. Cooling system parameter matching: Cooling medium injection pressure of high-pressure point cooling system Equivalent diameter of thermal section Related, usually Not less than 5 MPa, spray duration Based on the solidification time simulation of the hot spot, it is approximately 10%-20% of the theoretical solidification time for that region. In a multi-loop water cooling system, the flow rate of each loop is distributed according to the heat load ratio of its assigned area.

[0111] In summary, through Embodiments 1 to 4, this invention fully, comprehensively, and hierarchically elucidates the specific implementation of the "Integrated System and Method for Die Casting and Precision Machining of Precision Aluminum Cast Motor Housings." From the system assembly (Embodiment 1), to the intelligent software module ensuring data accuracy (Embodiment 2), to the advanced control algorithm ensuring machining execution precision (Embodiment 3), and finally to the mold hardware innovation providing high-quality blanks (Embodiment 4), it covers all the technical features in the claims and provides in-depth expansion far exceeding 8,000 words, fully revealing the specific implementation methods, technical details, and inventiveness of this invention. Those skilled in the art can implement the system and method described in this invention based on the above description and in conjunction with known die casting machines, CNC machine tools, sensors, and industrial computer technologies.

[0112] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0113] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An integrated system for die casting and precision machining of precision aluminum casting motor housings, characterized in that, include: The high-pressure die-casting unit is used to form casting blanks with a high-precision internal stop as the initial process reference. An online 3D scanning and inspection unit is used to perform geometric scanning and simultaneous temperature field acquisition on the casting blank. The five-axis precision machining center is equipped with feed axes directly driven by linear motors and a high-frequency electric spindle, and is used for precision machining of castings; The central intelligent control system is communicatively connected to the high-pressure die-casting unit, the online three-dimensional scanning and detection unit, and the five-axis precision machining center, respectively. The central intelligent control system includes: The intelligent scanning data preprocessing and thermal compensation module is used to perform reflectivity noise compensation and thermal deformation compensation on the raw point cloud data acquired by the online 3D scanning detection unit, and output a thermally compensated digital twin model. An adaptive path planning module is used to dynamically generate a processing path based on the comparison results between the thermally compensated digital twin model and the ideal digital twin model, with the highest accuracy benchmark confirmed by scanning as the unified benchmark. The composite motion control module is used to drive and control each motion axis of the five-axis precision machining center to perform high-precision collaborative machining according to the machining path.

2. The integrated system according to claim 1, characterized in that, The intelligent scanning data preprocessing and thermal compensation module includes: The reflectivity compensation submodule is configured to: acquire point cloud data of the casting in at least two different spectral bands or polarization directions; identify and mark specular reflection noise points based on the normal difference or intensity difference of corresponding points between each data source; perform multi-source data weighted fusion on non-noise points; perform neighborhood surface fitting interpolation on noise points to generate reflectivity-compensated point cloud. The thermal deformation compensation submodule is configured to: synchronously acquire the real-time temperature field distribution on the surface of the casting; input the real-time temperature field into the casting thermal deformation reduction model pre-integrated in the digital twin to predict the thermal deformation displacement field; and perform an inverse geometric transformation on the point cloud after reflectivity compensation, that is, subtract the predicted displacement vector of the corresponding position from the coordinates of each point to generate the thermally compensated digital twin model.

3. The integrated system according to claim 1, characterized in that, The composite motion control module adopts a hierarchical control architecture, including: The collaborative layer employs a ring-coupled controller to receive the commanded position and actual position of each axis, calculate and output compensation signals for multi-axis synchronization; In the tracking layer, each axis independently uses a fuzzy adaptive PID controller as the core tracking controller, and its PID parameters can be adaptively adjusted online according to the tracking error and its rate of change. The feedforward compensation layer, including a disturbance observer and a target trajectory prediction feedforward controller, is used to perform feedforward compensation for the system's equivalent disturbance and known future trajectory information; The final control command for each axis is obtained by combining the compensation signals output by the tracking layer, the feedforward compensation layer, and the coordination layer.

4. The integrated system according to claim 3, characterized in that, The control law of the ring-coupled controller is: ； in, For the first Synchronization compensation signal for the shaft, For the first Tracking error of the axis and The first Synchronization error between the shaft and adjacent shafts and This is the coupling gain coefficient.

5. The integrated system according to claim 3, characterized in that, The fuzzy adaptive PID controller tracks the error. and error change rate As input, the proportional coefficient is output through fuzzy inference. Integral coefficient Differential coefficients Adjustment amount , , Its fuzzy rule base is based on the following experience: When | When it is large, increase , reduce and restrictions ; When |e| and | When it is of medium size, decrease and take appropriate and ; When | When it is small, increase , and according to | | Adjustment To suppress oscillations.

6. The integrated system according to claim 1, characterized in that, The mold for the high-pressure die-casting unit includes: The central annular stepped gating system has its gating sleeve located in the central axis area of ​​the mold, and the molten aluminum fills the cavity through evenly distributed fan-shaped ingates. The hydraulic push structure integrated inside the slider is used to actively push against the key parts of the casting during mold opening and core pulling to suppress demolding deformation. A multi-loop water-cooled and high-pressure point-cooled system based on CAE analysis.

7. The integrated system according to any one of claims 1-6, characterized in that, The online 3D scanning and detection unit includes a high-precision structured light or laser scanner and an array of non-contact infrared temperature sensors.

8. A method for integrating die casting and finishing of precision aluminum casting motor housings, applied to the system as described in any one of claims 1-7, characterized in that, Includes the following steps: S100: A casting blank is formed by high-pressure die casting, and a high-precision internal stop is formed in it as the initial process reference. S200: Performs online 3D scanning and intelligent data compensation on casting blanks to generate a thermally compensated digital twin model; S300: Compare the thermally compensated digital twin model with the ideal digital twin model, and dynamically generate the processing path using the highest precision benchmark confirmed by scanning as the unified benchmark; S400: Based on the machining path, a composite motion control algorithm is used to drive the five-axis precision machining center to perform the machining of the assembly in one clamping, completing the finishing of all key features.

9. The integrated method according to claim 8, characterized in that, Step S200 includes: S210: Performs multispectral or polarized light reflectance compensation, and eliminates point cloud noise and data loss caused by high reflectance of casting surface through multi-source data fusion and reconstruction. S220: Performs real-time thermal gradient mapping and finite element thermal deformation compensation. Based on the synchronously acquired temperature field, it predicts the displacement field through a thermal deformation reduced-order model and performs inverse compensation on the point cloud to eliminate geometric measurement errors caused by thermal deformation. S230: Output high-fidelity point cloud data after compensation by S210 and S220, as the digital twin model after thermal compensation.

10. The integrated method according to claim 8, characterized in that, The composite motion control algorithm used in step S400 ensures the minimization of synchronization error between multiple axes through a ring coupling strategy, adjusts control parameters online through fuzzy adaptive PID to cope with processing disturbances, and performs feedforward compensation for system disturbances and known trajectories through a disturbance observer and predictive feedforward to achieve nanometer-level trajectory tracking accuracy.

Images