A dynamic balance control method for dual-spindle cooperative machining

Abstract

The application discloses a kind of dynamic balance control methods of dual spindle collaborative processing, it is related to numerical control equipment technical field. Including: synchronous evaluation is based on the high-frequency transient impact mode of future instruction information prediction and the low-frequency steady-state unbalance mode perceived based on current state information, and generate the bimodal control confidence vector of the reliability thereof;Based on the confidence vector, the impact cancellation instruction and energy balance instruction preliminarily generated are risk weighted;Based on the impact readiness factor of quantifying impact event time urgency, and in combination with the confidence vector, the instruction after risk weighting is dynamically fused to generate final collaborative control instruction;And the final collaborative control instruction is synchronously executed.The application constructs a kind of control architecture capable of identifying and collaborative processing bimodal interference, improves the systematic dynamic balance under all working conditions, and improves processing precision and stability.

Description

Technical Field

[0001] This invention relates to the field of CNC equipment technology, and in particular to a dynamic balance control method for dual-spindle collaborative machining. Background Technology

[0002] Twin-spindle machining centers are widely used in modern manufacturing to improve production efficiency. During collaborative machining, the machine tool system is simultaneously subjected to two dynamic disturbances with drastically different physical characteristics and time scales: one is the high-frequency transient impact generated by the instantaneous interaction between the tool and the workpiece; the other is the low-frequency steady-state imbalance caused by the long-term uneven load on the two spindles. The high-frequency transient impact demands extremely high instantaneous response capability from the control system, while the low-frequency steady-state imbalance requires good smoothness and robustness. Current technologies typically employ control methods based on a single control law or a fixed parameter set. This control architecture lacks an inherent mechanism to identify and address the inherent dual-modal characteristics of these two combined disturbances. Therefore, any system based on a single control paradigm is forced to compromise between these two conflicting performance requirements, fundamentally preventing the achievement of true, full-condition dynamic balance. This has become a core technological bottleneck limiting further breakthroughs in the ultimate accuracy and stability of twin-spindle machining equipment. Summary of the Invention

[0003] This invention provides a dynamic balance control method for dual-spindle collaborative machining to solve the performance compromise problem caused by the inherent limitations of the control architecture in the prior art, which cannot coordinately cope with two dual-mode disturbances: high-frequency transient shocks and low-frequency steady-state imbalances.

[0004] In view of the above problems, the present invention provides a dynamic balance control method for dual-spindle collaborative machining, comprising the following steps:

[0005] Simultaneously evaluate the high-frequency transient impact mode predicted based on future command information and the low-frequency steady-state imbalance mode perceived based on current state information, and generate the impact prediction confidence and energy balance confidence corresponding to the two modes, which together constitute the dual-mode control confidence vector;

[0006] For the high-frequency transient impact mode and the low-frequency steady-state imbalance mode, preliminary impact cancellation commands and energy balance commands are generated respectively. The preliminary commands are then weighted using the corresponding confidence levels in the dual-mode control confidence vector to generate risk-weighted impact cancellation commands and risk-weighted energy balance commands.

[0007] Based on the impact preparedness factor used to quantify the time urgency of the high-frequency transient impact event, and combined with the dual-axis control confidence, the impact control weight and energy balance weight are dynamically calculated, and the dynamic weights are used to fuse the risk-weighted impact offset command and the risk-weighted energy balance command to generate the final coordinated control command.

[0008] The final coordinated control command is decomposed into drive signals for the dual spindle actuators and executed synchronously.

[0009] The technical solution provided in this application has at least the following technical effects or advantages:

[0010] The technical solution provided in this application delivers holistic benefits by constructing a control architecture capable of fundamentally identifying, decoupling, and collaboratively handling the dual-mode characteristics of dynamic imbalance. This method no longer passively suppresses vibration but actively performs concurrent, differentiated, and uncertainty-aware feedforward control of the two disturbances. The architecture's inherent collaborative decision-making mechanism resolves the performance compromise issues faced by single control strategies, enabling the system to automatically operate in optimal dynamic equilibrium under all operating conditions. Ultimately, this method significantly reduces both the high-frequency vibration peaks caused by transient impacts and the low-frequency structural torsion caused by steady-state imbalances, thereby achieving qualitative improvements in both microscopic surface quality and macroscopic part geometric accuracy, and enhancing equipment robustness and processing efficiency. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the dynamic balance control method for dual-spindle collaborative machining according to the present invention. Detailed Implementation

[0012] The above technical solutions will now be described in detail with reference to the accompanying drawings and specific embodiments to provide a better understanding of them. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments used only to explain the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. Furthermore, it should be noted that, for ease of description, only the parts related to the present invention are shown in the drawings, not all of them.

[0013] Please see Figure 1 A dynamic balance control method for dual-spindle collaborative machining includes the following steps:

[0014] Simultaneously evaluate the high-frequency transient impact mode predicted based on future command information and the low-frequency steady-state imbalance mode perceived based on current state information, and generate the impact prediction confidence and energy balance confidence corresponding to the two modes, which together constitute the dual-mode control confidence vector;

[0015] For the high-frequency transient impact mode and the low-frequency steady-state imbalance mode, preliminary impact cancellation commands and energy balance commands are generated respectively. The preliminary commands are then weighted using the corresponding confidence levels in the dual-mode control confidence vector to generate risk-weighted impact cancellation commands and risk-weighted energy balance commands.

[0016] Based on the impact preparedness factor used to quantify the time urgency of the high-frequency transient impact event, and combined with the dual-axis control confidence, the impact control weight and energy balance weight are dynamically calculated, and the dynamic weights are used to fuse the risk-weighted impact offset command and the risk-weighted energy balance command to generate the final coordinated control command.

[0017] The final coordinated control command is decomposed into drive signals for the dual spindle actuators and executed synchronously.

[0018] The specific implementation process of the dynamic balance control method for dual-spindle collaborative machining disclosed in this invention is uniformly scheduled and commanded by an executable dual-modal collaborative control engine deployed within the CNC system. The entire process of this dual-modal collaborative control engine begins with real-time evaluation and quantification of the dynamic imbalance characteristics during machining. The goal of this stage is to provide subsequent collaborative control decisions with high-dimensional composite state information that includes future predictions, current state, and a quantitative assessment of the reliability of both information.

[0019] Evaluation and Quantification of Dynamic Features: The specific implementation path of this evaluation and quantification stage begins with the synchronous acquisition and preprocessing of information sources. Within a control cycle, the dual-modal collaborative control engine first executes two parallel information acquisition operations. The first operation is to acquire a G-code prefetch data stream containing machining instructions for at least 100 milliseconds in the future through the information interface with the CNC system kernel. The acquired G-code prefetch data stream is then sent to a built-in syntax parser, which, according to the ISO6983 standard, parses the ASCII code format program segment into a structured instruction queue containing information such as motion instruction type (e.g., G00, G01, G02 / G03), target coordinates, and feed rate. The second operation is to synchronously acquire the real-time instantaneous power signals P1(t) and P2(t) from the dual-spindle servo drives through a high-speed data bus (e.g., EtherCAT) at a sampling frequency of not less than 1 kHz. To eliminate interference from high-frequency electrical noise, the acquired instantaneous power signals P1(t) and P2(t) will be processed by a second-order Butterworth low-pass filter with a cutoff frequency set to, for example, 100Hz, to retain the effective frequency components related to the cutting load variation.

[0020] After successfully acquiring and preprocessing the information source, the dual-modal collaborative control engine then performs parallel extraction and prediction of dual-modal features. Specifically, one branch processing unit of the engine continuously scans the aforementioned structured instruction queue to identify discontinuous changes in the machining state, thereby predicting high-frequency transient impact events. A specific identification rule is that when a switch occurs in the instruction sequence from G00 (rapid positioning) to G01 / G02 / G03 (cutting feed), the target coordinates of that switch point are determined as a predicted impact position Pos. next Based on the distance between the current location and the target coordinates, as well as the operating speed of G00, the predicted impact time T can be accurately calculated. next Meanwhile, another branch processing unit of the engine processes the filtered instantaneous power signal and obtains a real-time updated power difference signal to characterize the low-frequency steady-state imbalance trend by calculating ΔP(t)=P1(t)-P2(t).

[0021] After feature extraction and prediction are completed, the entire evaluation and quantification phase proceeds to the calculation of the dual-modal control confidence vector (DMCC). The purpose of this step is to assign a weight to the two modal features obtained in the previous step, quantifying their reliability. The specific calculation model is as follows. First, the impact prediction confidence vector... imp The calculation aims to quantify T. next and Pos nextThe accuracy of the prediction can be determined by a weighted model that includes multiple influencing factors:

[0022] Conf imp ×w1·f(interpolation type) + w2·g(model accuracy) where f(interpolation type) is a lookup function. For G01 linear interpolation, its output value is 1.0; for G02 / G03 circular interpolation, its output value is 0.9; and for the more complex NURBS curve interpolation, its output value is 0.8, thus reflecting the predictive determinism of different instruction types. w1 and w2 are preset weights. Next, the energy balance confidence level... eng The calculation, aimed at quantifying the necessity and effectiveness of current energy balance control, can be given by a nonlinear function:

[0023] Conf eng ×tanh(γ·|ΔP(t)|) where γ is an adjustable gain coefficient, for example, 0.1. This function ensures that when the absolute value of the power difference ΔP(t) is small, Conf eng The value of approaches 0, while when the power difference ΔP(t) increases significantly, Conf eng The value then smoothly saturates and approaches 1.0. Finally, the dual-modal cooperative control engine will calculate the Conf... imp and Conf eng Combined into a two-dimensional vector DMCC = [Conf imp Conf eng As an integral part of a composite state information package containing quantified uncertainty information, it is provided to subsequent control command generation and decision-making processes.

[0024] Initial Control Command Generation and Weighting: After the dual-modal collaborative control engine completes the evaluation and quantification of the dynamic characteristics of the processing, the entire control process enters the initial control command generation and weighting stage. The technical objective of this stage is to transform the composite state information package with uncertain quantification information generated in the previous stage into two parallel, executable action plans that contain risk assessments.

[0025] The specific implementation path at this stage begins with the independent generation of the initial control command. Two dedicated computing units within the dual-modal cooperative control engine will respectively process the different modal characteristics in the composite state information packet. One computing unit is responsible for calculating the initial impact cancellation command C based on the predicted impact event and using a pre-established inverse model of machine tool dynamics. imp_rawThis machine tool dynamics inverse model is a mathematical transfer function matrix that describes the relationship between the force / torque applied to the spindle actuator and the acceleration generated at the spindle endpoint. The specific execution process of this calculation unit is as follows: taking the predicted impact event (including direction and estimated amplitude) as input, it solves this inverse model to calculate a theoretically equivalent instantaneous compensating torque or force that can generate an equivalent reverse acceleration at the spindle endpoint. This compensating torque or force is defined as the initial impact cancellation command C. imp_raw Meanwhile, another computing unit is responsible for calculating the preliminary energy balance command C based on the energy difference surge, i.e., the real-time updated power difference signal ΔP(t), through a feedforward controller. eng_raw The feedforward controller can employ a proportional-integral (PI) control algorithm, with its input being the power difference ΔP(t) and its output being a feed rate adjustment coefficient. The initial energy balance command C... eng_raw This is defined as a set of differential fine-tuning operations performed on the current feed rate of the dual spindles, aimed at bringing the power difference ΔP(t) close to zero.

[0026] After independently generating two preliminary control commands, the dual-modal cooperative control engine then performs a step of risk-weighted processing based on confidence levels. The purpose of this step is to perform risk-weighted processing on the dual-modal control confidence vector DMCC×[Conf] generated in the previous stage, which is part of the composite state information packet. imp Conf eng As a key modulation factor, it is used to adjust the amplitude of the two preliminary commands, thereby generating the final command that is passed to the decision module and has an embedded uncertainty risk assessment. The specific implementation process of this risk-weighted processing is as follows: the engine first extracts the impact prediction confidence level (Conf) from the dual-modal control confidence vector. imp And Energy Balance Confidence eng Subsequently, a risk-weighted shock offset instruction C is generated by performing two parallel scalar multiplication operations. imp_risk and risk-weighted energy balance instruction C eng_risk

[0027] The first operation is:

[0028] C imp_risk =C imp_raw Conf imp ;

[0029] The second operation is:

[0030] C eng_risk =C eng_raw Conf eng ;

[0031] Through this weighting operation, the theoretical command value generated in the previous step is effectively scaled by its own reliability level. For example, a C that theoretically requires a compensation torque of 10 Nm... imp_raw If its corresponding Conf imp If it is only 0.8, then the generated C imp_risk It will be automatically adjusted to 8 Nm. This process ensures that only high-confidence control modes can generate high-amplitude control commands, thus providing a safer and more reliable input for subsequent optimal fusion decision-making.

[0032] Fusion and Decision-Making of Cooperative Control Commands: After the dual-modal cooperative control engine generates two parallel commands containing risk assessments, the entire control process proceeds to the fusion and decision-making stage of cooperative control commands. The technical objective of this stage is to fuse the two potentially conflicting action plans with uncertain risks output from the previous stage into a single, globally optimal final command that can be directly issued to the executing agency, through an intelligent, multi-dimensional decision-making algorithm.

[0033] The implementation path at this stage begins with the quantitative calculation of a key intermediate variable—the Impact Preparedness Factor (IAF). The Impact Preparedness Factor aims to quantify the temporal urgency of the high-frequency transient impact event using a dimensionless value between 0 and 1. Specifically, the calculation involves the dual-modal cooperative control engine utilizing the predicted impact time T obtained in the preceding steps. next and the current time T now This is achieved through an exponentially decaying function:

[0034] IAF=exp(-k·(T next -T now ));

[0035] In this formula, k is a preset time decay constant, which physically defines the width of the time window for the system to enter a "ready state." The larger the value of k, the more significantly the IAF will increase only when the system is extremely close to the impact. The calculated IAF will serve as an input for the subsequent generation of dynamic weights.

[0036] Next, the dual-modal cooperative control engine executes the core algorithm of this invention: dynamic weight generation. The underlying logic of this algorithm is to construct a dual modulation mechanism capable of simultaneously responding to both time urgency (IAF) and command reliability (DMCC). Its specific implementation involves two steps: First, a baseline weight value, disregarding command reliability, is generated through a soft switching function that takes the IAF as input. This baseline weight value aims to achieve a smooth transition of control focus between impact cancellation and energy balance. Second, the Conf in the dual-modal control confidence vector (DMCC) generated in the preceding steps is then utilized... imp and Conf eng Each component is scaled relative to the baseline weight value to obtain the final impact control weight w. imp and energy balance weight w eng A preferred specific formula implementation of this algorithm is as follows:

[0037] w imp ×(0.5·(1+tanh(β·(IAF-α))))·Conf imp

[0038] w eng =(0.5·(1-tanh(β·(IAF-α))))·Conf eng

[0039] In this formula, tanh() is the hyperbolic tangent function, which forms the core of the soft switching function and provides a smooth S-shaped switching curve. α and β are preset parameters, where α serves as the switching center point, defining the value of IAF at which the baseline weights of the two modes are equal (both are 0.5), typically ranging from 0.5 to 0.8; β, as the switching steepness, defines the rate of weight switching; the larger the β value, the closer the switching process is to an ideal step. imp and Conf eng This serves as the final scaling factor, ensuring that even under very tight time constraints (IAF approaching 1), the confidence level of the shock prediction itself remains high. imp Very low, the final w imp It will also be suppressed accordingly, thus effectively avoiding the control risks introduced by unreliable instructions.

[0040] After the dynamic weights are generated, this stage of the process proceeds to the final step: the weighted fusion of the final collaborative control commands. The dual-modal collaborative control engine will then use the dynamic weights w generated in the previous step... imp and w eng Compared with the risk-weighted instruction C generated in the previous stage imp_risk and C eng_riskThe final coordinated control command C is obtained through a weighted summation operation. final :

[0041] C final =w imp ·C imp_risk +w eng ·C eng_risk

[0042] To further illustrate this embodiment, a specific numerical example is provided: Assume that during a certain control cycle, the system acquires the following inputs and intermediate calculation results:

[0043] The time difference between the predicted impact time and the current time: T next -T now =5ms = 0.005s;

[0044] Dual-modal control confidence vector: DMCC = [Conf imp =0.9, Conf eng =0.6];

[0045] Initial instructions:

[0046] C imp_raw = [torque 10 Nm],

[0047] C eng_raw =[Feed adjustment -1.0%];

[0048] Preset parameters: k = 0.5, α = 0.7, β = 5.0. The calculation process of the dual-modal cooperative control engine is as follows:

[0049] Risk-weighted instruction calculation:

[0050] C imp_risk =10Nm·0.9=9Nm

[0051] C eng_risk =-1.0%·0.6=-0.6%;

[0052] Impact alert factor calculation:

[0053] IAF=exp(-0.5·0.005)≈0.9975;

[0054] Dynamic weight calculation: First, calculate the core part of the soft handover function:

[0055] tanh(β·(IAF-α));

[0056] Substituting the values, we get:

[0057] tanh(5.0·(0.9975-0.7))≈0.903;

[0058] Based on this result, the impact control weight w is calculated. imp and energy balance weight w eng :

[0059]

[0060] w eng = [0.5·(1-0.903)]·0.6≈0.029;

[0061] Final instruction fusion:

[0062] C final It will be a composite command containing a torque component of 0.856.9 Nm and a feed adjustment component of 0.029 (-0.6%).

[0063] This example demonstrates that, due to the imminent nature of the impact event, the system allocates the majority of control weights to the impact control mode; simultaneously, because the energy balance mode has a low confidence level, its final weight is further suppressed. This process fully embodies the intelligent decision-making mechanism proposed in this invention, which comprehensively considers both time urgency and command reliability.

[0064] Decomposition and Synchronous Execution of Drive Signals: After the dual-modal cooperative control engine completes the optimal fusion and decision-making of the cooperative control commands, the entire dynamic balance control process enters its final stage of decomposition and synchronous execution of drive signals. The technical objective of this stage is to unify the final cooperative control command C generated in the previous stage—a culmination of the upper-level algorithm's wisdom—into a unified final cooperative control command. final This is precisely and accurately converted into multi-channel drive signals that can be executed by the underlying physical hardware, ensuring the synchronization of these signals in time, thereby completing the closed loop from intelligent decision-making to physical response. The implementation path of this stage begins with the decomposition and issuance of the final instruction.

[0065] The C final It is a composite instruction containing multiple weighted and fused components for different control modes. The dual-modal cooperative control engine needs to decompose this composite instruction into specific drive signals for different physical actuators based on its physical meaning. A specific implementation of this decomposition process is: the engine first parses C... final The internal data structure is analyzed to extract the "impact compensation" component and the "energy balance" component. The "impact compensation" component (e.g., an instantaneous compensation torque with a target of 7.7 Nm) is routed to the spindle servo drive responsible for applying the compensation torque. This instruction is then translated into an additional torque instruction that the servo drive can recognize and sent to the drive via a high-speed bus (e.g., CANopen or EtherCAT).

[0066] Simultaneously, the "energy balance" component (e.g., a feed rate fine-tuning targeting -0.017%) is routed to the feed axis servo drive controlling the corresponding spindle feed motion. This command is translated into a small adjustment to the current speed loop setpoint of that feed axis drive and is also sent via the high-speed bus. After the command decomposition and sending are completed, this stage enters the step that plays a decisive role in the final control accuracy: the synchronous execution of the drive signals. To ensure that the multiple decomposed drive signals can be applied accurately and synchronously at predetermined times, the dual-modal cooperative control engine can employ a time synchronization mechanism based on Distributed Clocks (DC). This mechanism is a standard feature of industrial Ethernet protocols such as EtherCAT.

[0067] The specific implementation process is as follows: the master station in the network (usually a CNC system) broadcasts a unified reference clock signal to all slave stations (including each servo drive). All slave stations calibrate their local clocks according to this reference clock, thereby achieving nanosecond-level time synchronization of the entire system. When the dual-modal cooperative control engine issues the decomposed drive signal, it appends a precise "execution timestamp" based on this synchronization reference clock to the instruction. After receiving the instruction, each servo drive does not execute it immediately, but waits for its local clock to reach the "execution timestamp" specified in the instruction before synchronously and simultaneously applying the corresponding voltage or current signal to the motor, thereby completing the final physical action.

[0068] Through this synchronous execution mechanism, the present invention ensures that the cooperative control strategy calculated by the complex upper-level algorithm can be converted into an accurate response in the physical world with high fidelity and without delay, thereby completing the complete closed loop of the entire dynamic equilibrium control method.

[0069] This detailed embodiment elaborates on the internal operating logic of each major functional module, aiming to provide a detailed basis and explanation for those skilled in the art to understand and implement it. It should be emphasized that the above description constitutes a specific, preferred embodiment, but the concept of the present invention is not limited thereto. Any equivalent transformations, modifications, or improvements based on the core spirit of the present invention, without departing from the technical principles and scope disclosed in this specification, should be considered to fall within the scope of protection claimed by the present invention, as long as they achieve the same or similar technical effects.

Claims

1. A dynamic balance control method for dual-spindle collaborative machining, characterized in that, Includes the following steps: Simultaneously evaluate the high-frequency transient impact mode predicted based on future command information and the low-frequency steady-state imbalance mode perceived based on current state information, and generate the impact prediction confidence and energy balance confidence corresponding to the two modes, which together constitute the dual-mode control confidence vector; For the high-frequency transient impact mode and the low-frequency steady-state imbalance mode, preliminary impact cancellation commands and energy balance commands are generated respectively. The preliminary commands are then weighted using the corresponding confidence levels in the dual-mode control confidence vector to generate risk-weighted impact cancellation commands and risk-weighted energy balance commands. Based on the impact preparedness factor used to quantify the time urgency of the high-frequency transient impact event, and combined with the dual-axis control confidence, the impact control weight and energy balance weight are dynamically calculated, and the dynamic weights are used to fuse the risk-weighted impact offset command and the risk-weighted energy balance command to generate the final coordinated control command. The final coordinated control command is decomposed into drive signals for the dual spindle actuators and executed synchronously.

2. The method according to claim 1, characterized in that, The future instruction information is a G-code pre-read data stream obtained from the CNC system kernel, and the current status information is a real-time instantaneous power signal acquired from the dual-spindle servo drive.

3. The method according to claim 1, characterized in that, The calculation of the impact prediction confidence level depends at least on the interpolation type of the machining command or the prediction accuracy of the machine tool dynamics model; the calculation of the energy balance confidence level depends at least on the amplitude of the instantaneous power difference between the two spindles.

4. The method according to claim 1, characterized in that, The impact preparedness factor is a variable with a value range between 0 and 1, calculated based on the difference between the predicted impact time and the current time.

5. The method according to claim 1 or 4, characterized in that, The steps for calculating the dynamic weights include: A baseline weight value is generated by a soft switching function that takes the impact alert factor as input. The baseline weight values ​​are scaled using the impact prediction confidence and the energy balance confidence, respectively, to obtain the impact control weight and the energy balance weight.

6. The method according to claim 5, characterized in that, The final cooperative control command C final Calculated using the following formula: C final =w imp ·C imp_risk +w eng ·C eng_risk Among them, C imp_risk and C eng_risk These are the risk-weighted shock offsetting command and the risk-weighted energy balance command, respectively; the dynamic weight w imp and w eng Calculated using the following formula: w imp =(0.5·(1+tanh(β·(IAF-α))))·Conf imp w eng =(0.5·(1-tanh(β·(IAF-α))))·Conf eng Wherein, IAF is the impact alert factor, Conf imp For the impact prediction confidence level, Conf eng The energy balance confidence level is given by α and β, which are preset parameters.

Images