An adaptive control method and system for reciprocating cutting tools

Abstract

This application provides an adaptive control method and system for a reciprocating cutting tool. The method includes: real-time acquisition of the operating current signal of the drive motor, the reciprocating motion frequency signal of the saw blade, and the triaxial vibration signal of the saw blade; calculation of the average value component and the peak fluctuation component of the operating current signal, and calculation of the vibration energy component of the triaxial vibration signal in the cutting direction; fusion of the average value component, the peak fluctuation component, and the vibration energy component in the cutting direction to generate a comprehensive load index for the current cutting cycle; inputting the comprehensive load index, the frequency deviation, and the rate of change of the comprehensive load index into an adaptive strategy mapping model; and outputting load control commands, which include feed speed adjustment, motor power adjustment, and torque curve adjustment parameters; and coordinating the adjustment of the feed speed of the reciprocating cutting tool, the output power of the drive motor, and the waveform characteristics of the instantaneous torque output curve of the drive motor in a single reciprocating cycle according to the commands.

Description

Technical Field

[0001] This application relates to the field of reciprocating cutting tool technology, and more specifically, to an adaptive control method and system for reciprocating cutting tools. Background Technology

[0002] With the continuous development of modern manufacturing and construction and decoration industries, reciprocating cutting tools are increasingly used in the processing of various materials. However, most reciprocating cutting tools still rely mainly on operator experience or simple feedback based on a single parameter for control. In actual cutting processes, due to the inhomogeneity of workpiece materials, internal hard points, or structural changes, tools often face problems such as sudden load changes, increased vibration, and potential resonance. This makes it difficult for traditional control methods to adapt to changes in actual working conditions in real time and accurately, easily leading to consequences such as decreased cutting efficiency, accelerated saw blade wear, motor overload, and even tool damage.

[0003] The problems are: the existing reciprocating cutting tools have a single control method, which cannot adapt to complex working conditions, resulting in low cutting efficiency, rapid saw blade wear, and easy overload of the drive motor. Summary of the Invention

[0004] This application addresses the technical problems of existing reciprocating cutting tools, which suffer from a single control method, inability to adapt to complex working conditions, low cutting efficiency, rapid saw blade wear, and susceptibility to drive motor overload. By integrating a closed loop of sensing, decision-making, and execution, this application reduces reliance on operator experience, enabling reciprocating cutting tools to automatically maintain a safe, efficient, and smooth cutting process even when dealing with heterogeneous materials.

[0005] To address the aforementioned issues, this application provides an adaptive control method for a reciprocating cutting tool, comprising: real-time acquisition of the operating current signal of the drive motor, the reciprocating motion frequency signal of the saw blade, and the triaxial vibration signal of the saw blade; calculation of the average value component and peak fluctuation component based on the operating current signal, and calculation of the vibration energy component in the cutting direction based on the triaxial vibration signal of the saw blade; fusion of the average value component, peak fluctuation component, and vibration energy component in the cutting direction according to a preset weighted fusion rule to generate a comprehensive load index for the current cutting cycle; inputting the comprehensive load index, the deviation between the reciprocating motion frequency and its set frequency, and the rate of change of the comprehensive load index within the most recent preset number of cycles into a preset adaptive strategy mapping model; outputting load control commands from the adaptive strategy mapping model, the load control commands including feed speed adjustment, motor power adjustment, and torque curve adjustment parameters; and coordinating the adjustment of the feed speed of the reciprocating cutting tool, the output power of the drive motor, and the waveform characteristics of the instantaneous torque output curve of the drive motor in a single reciprocating cycle according to the load control commands.

[0006] Compared with existing technologies, the technical effects achieved by this solution are as follows: By integrating the load index and adaptive strategy model, the feed speed, motor power, and torque curve can be adjusted in real time, enabling the reciprocating cutting tool to quickly adapt to material changes and maintain optimal cutting efficiency without overloading, reducing jamming and speed fluctuations. Furthermore, by dynamically adjusting the torque curve, the motor output and saw blade movement are effectively smoothed, reducing fatigue wear on the saw blade and drive mechanism caused by impact and vibration. At the same time, based on multi-dimensional fusion judgment of current and vibration, abnormal loads, such as hard points or material mutations, can be identified earlier and more accurately. By promptly reducing feed and adjusting power, the drive motor is actively protected from overheating or overload damage. In particular, by integrating perception, decision-making, and execution closed loops, the method reduces the dependence on operator experience, enabling the reciprocating cutting tool to automatically maintain a safe, efficient, and smooth cutting process even when facing non-homogeneous materials.

[0007] In one possible design, generating the comprehensive load index for the current cutting cycle specifically includes: taking the arithmetic mean of the working current signal over a complete cutting cycle as the average value component; taking the ratio of the peak value to the root mean square value of the working current signal over a complete cutting cycle as the fluctuation peak value component; performing triaxial decomposition on the triaxial vibration signal of the saw blade, integrating the square of the vibration acceleration signal in the cutting direction over the complete cutting cycle to obtain the vibration energy component in the cutting direction; and normalizing the average value component, the fluctuation peak value component, and the vibration energy component in the cutting direction, multiplying them by their respective preset weighting coefficients, and summing them to obtain the comprehensive load index.

[0008] Compared with existing technologies, the technical effects achieved by this solution are as follows: By simultaneously calculating the ratio of the average current, peak current to the root mean square current, and the vibration energy component in the cutting direction, the true load state of the cutting process can be comprehensively reflected from both electrical and mechanical dimensions, avoiding potential misjudgments or omissions due to a single signal. Specifically, the fluctuation peak component can sensitively reflect current spikes caused by cutting into hard points, foreign objects, or sudden changes in load within the material, while the vibration energy component can effectively characterize the mechanical impact of the saw blade caused by changes in force. By fusing the two, potential overloads or anomalies can be predicted earlier and more reliably. Normalizing and weighting the three different components generates a single, quantified comprehensive load index. This comprehensive load index enables the subsequent adaptive strategy model to make accurate decisions based on a high-quality, high-information-density input, thereby outputting more coordinated and precise control commands.

[0009] In one possible design, the adaptive strategy mapping model is a trained multi-input multi-output feedforward neural network model; wherein, the feedforward neural network model takes the comprehensive load index, frequency deviation and the rate of change of the comprehensive load index as inputs, and outputs the feed speed adjustment, motor power adjustment and torque curve adjustment parameters.

[0010] Compared with existing technologies, the technical effects achieved by this solution are as follows: The feedforward neural network can learn from historical operating data and autonomously mine and establish a complex, nonlinear, and high-dimensional mapping relationship between the three inputs (comprehensive load index, frequency deviation, and rate of change of comprehensive load index) and the three output commands (feed speed, motor power, and torque curve). This enables the system to make intelligent decisions and output more accurate and coordinated control commands. Furthermore, the feedforward neural network model can continuously optimize its internal parameters by learning optimal operating data under different materials and wear conditions. This makes the system not only applicable to known operating conditions but also possess the generalization ability to reasonably infer and handle similar complex operating conditions that it has not experienced before.

[0011] In one possible design, adjusting the waveform characteristics of the instantaneous torque output curve of the drive motor within a single round trip cycle specifically includes: adjusting the parameters of the proportional-integral controller of the drive motor current loop within a preset time window before and after each reversal point of the saw blade movement according to the torque curve adjustment parameters, or injecting a dynamic feedforward compensation signal into the current loop to change the steepness of the rise and fall of the drive motor output torque during the reversal process.

[0012] Compared with existing technologies, the technical effects achieved by this solution are as follows: By dynamically adjusting the steepness of torque rise and fall within a preset time window before and after the commutation point, the output torque of the motor can be actively matched better with the changing mechanical load. This smooths the commutation process, greatly reducing the rigid impact and stress peaks experienced by the saw blade and transmission mechanism during high-speed commutation, thereby effectively improving the service life of mechanical components such as the saw blade. Furthermore, by dynamically adjusting the parameters according to the torque curve, the steepness is automatically reduced under heavy load or high impact conditions, enabling smoother commutation to protect the machinery. Conversely, under light load or when efficiency is prioritized, the steepness is increased, enabling faster commutation to shorten the cycle time. This effectively increases the system's adaptability to different working conditions. In particular, by dynamically adjusting the parameters of the proportional-integral controller in the current loop or injecting dynamic feedforward compensation signals, the response characteristics of the motor during the transient process of commutation can be optimized, ensuring a fast and stable tracking capability for complex, time-varying load torque.

[0013] In one possible design, the method further includes steps for avoiding mechanical resonance, specifically including: performing synchronous time-frequency analysis on the real-time acquired triaxial vibration signal of the saw blade and the working current signal; extracting the frequency band energy of the triaxial vibration signal of the saw blade near the reciprocating motion frequency and its harmonics as a first vibration characteristic quantity; extracting the amplitude of at least one harmonic component in the spectrum of the working current signal that is associated with the pre-stored natural frequency of the tool's mechanical structure as a current-related characteristic quantity; performing a fusion judgment based on the first vibration characteristic quantity and the current-related characteristic quantity, and according to a preset resonance risk judgment rule; if the judgment result indicates that there is a resonance risk, generating a frequency fine-tuning command; the frequency fine-tuning command is used to change the control signal of the drive motor, thereby causing the reciprocating motion frequency of the saw blade to deviate from the current value, so as to deviate it from the natural frequency range of the tool's mechanical structure.

[0014] Compared with existing technologies, the technical effects achieved by this solution are as follows: By performing synchronous time-frequency analysis on vibration and current signals, and extracting vibration energy characteristics related to operating frequency harmonics and current harmonic characteristics related to the structure's natural frequency, a two-dimensional, cross-validated resonance risk perception system is constructed, effectively reducing false alarms and false negatives. By using preset resonance risk judgment rules to fuse and judge the two characteristic quantities, it is possible to effectively distinguish between vibration aggravation caused by normal cutting load fluctuations and vibration / current characteristic changes caused by real mechanical resonance. This criterion based on multi-source information fusion significantly improves the system's robustness and judgment reliability in real, variable cutting environments. When a resonance risk is determined, by actively generating frequency fine-tuning commands, the operating frequency of the reciprocating cutting tool dynamically deviates from the current danger value, preventing resonance from damaging the mechanical structure.

[0015] In one possible design, the coordinated adjustment specifically includes the following priority logic: when the comprehensive load index exceeds the first safety threshold corresponding to the current cutting material, the operation of reducing the feed rate adjustment amount is executed first; or, when resonance risk is detected, the frequency fine-tuning command is executed first.

[0016] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: when the comprehensive load index exceeds the safety threshold or a resonance risk is detected, the system will immediately interrupt the conventional multi-parameter collaborative optimization process and prioritize the execution of the most direct and effective single risk avoidance operation. This eliminates the situation where the system is busy optimizing other parameters and delays the handling of emergency dangers, providing the highest priority guarantee for core safety.

[0017] In one possible design, the method also includes material type identification and initial parameter matching steps: In the initial cutting stage of the cutting operation, the reciprocating cutting tool is controlled to cut into the workpiece at a constant first feed rate, and the initial triaxial vibration signal and initial current signal of this stage are collected; the spectral distribution characteristics of the initial triaxial vibration signal and the steady-state amplitude characteristics of the initial current signal are extracted; the extracted features are matched with the feature ranges of different material types pre-stored in the material process parameter database to determine the material type range to which the current workpiece belongs; according to the determined material type range, the corresponding initial set frequency, initial feed rate and initial torque curve parameters are retrieved from the material process parameter database and used as the starting reference for the adaptive control process of this cutting operation.

[0018] Compared with existing technologies, the technical effects achieved by this solution are as follows: By analyzing the brief initial signal and automatically identifying the material type range during the initial cutting stage, and calling up a set of optimized initial parameters for that type of material, this replaces the process of manual selection and setting based on experience, achieving automated setting and effectively improving the intelligence level of the reciprocating cutting tool; In particular, by providing matching initial setting frequency and initial torque curve parameters for different materials, the adaptive adjustment transition process can be significantly shortened, improving overall control efficiency and stability.

[0019] This application also provides an adaptive control system for a reciprocating cutting tool, used to implement an adaptive control method for the reciprocating cutting tool. The system includes: a sensor module, including a current sensor for acquiring the operating current of the drive motor, an encoder or Hall sensor for acquiring the reciprocating frequency of the saw blade, and a triaxial vibration acceleration sensor disposed on the tool head; a signal processing and calculation module, electrically connected to the sensor module, for processing the signals from the sensor module in real time and calculating the comprehensive load index, frequency deviation, and rate of change of the comprehensive load index; an adaptive strategy control module, embedding an adaptive strategy mapping model, whose input is connected to the signal processing and calculation module, for receiving the comprehensive load index, frequency deviation, and rate of change of the comprehensive load index, and generating feed speed adjustment, motor power adjustment, and torque curve adjustment parameters; and a multi-objective cooperative execution module, including a motor power driver, a feed servo controller, and a dynamic torque curve controller; the multi-objective cooperative execution module is connected to the adaptive strategy control module, for receiving and executing the adjustment amounts and parameters to coordinately control the drive motor and the feed mechanism.

[0020] Compared with existing technologies, the technical effects achieved by this solution are as follows: The system integrates four functional modules—multi-dimensional sensing, real-time computing, intelligent decision-making, and collaborative execution—into an organic whole, forming a closed loop of information flow from sensors to actuators. This enables global and dynamic adaptive management of the cutting process. Specifically, by integrating multiple sensors, the system can monitor current signals, motion frequency, and vibration in real time, allowing for rapid responses to the cutting process status. The signal processing and computing module can quickly process data from sensor feedback and calculate the comprehensive load index and its rate of change in real time, improving the real-time performance and accuracy of decision-making and providing more precise data support for adaptive strategies. The adaptive strategy control module can dynamically adjust the feed speed, power, and torque curves according to the current working state, enhancing the intelligence level of the cutting process. The multi-objective collaborative execution module can integrate the adjustment of motor power, feed speed, and torque curves based on the output of the adaptive strategy control module, achieving multi-faceted collaborative control and optimizing overall work efficiency.

[0021] In one possible design, the system also includes a material and process parameter database and a self-learning module; the material and process parameter database stores the characteristic ranges of different material types and their corresponding initial working parameters; the self-learning module, after a cutting task is completed, receives an evaluation feedback signal on the cutting quality of this task, and updates the decision weight parameters in the adaptive policy mapping model using a reinforcement learning algorithm based on the feedback signal.

[0022] Compared with existing technologies, the technical effects achieved by this solution are as follows: the system can not only adapt in real time during a single operation, but also automatically update the internal parameters of the core decision model using reinforcement learning algorithms based on cutting quality feedback after the task is completed through a self-learning module. This allows the system's control strategy to continuously accumulate experience and self-optimize, resulting in performance improvements over time. The pre-stored initial working parameters for different material types in the material process parameter database enable the reciprocating cutting tool to quickly obtain the optimal cutting parameters for different material types, thereby shortening the transition process for adaptive adjustment to different materials and improving cutting efficiency and stability.

[0023] In one possible design, the dynamic torque curve controller is part of the motor power driver and is configured to dynamically adjust the response characteristics of the current loop or superimpose a feedforward current command based on torque curve adjustment parameters.

[0024] Compared with existing technologies, the technical effects achieved by adopting this technical solution are as follows: by directly and dynamically adjusting the response characteristics of the current loop or superimposing feedforward current commands through a dynamic torque curve controller, it is possible to quickly respond to load changes and improve the system's adaptability to complex loads and operational stability. Attached Figure Description

[0025] Figure 1 A flowchart illustrating an adaptive control method for a reciprocating cutting tool provided in this application embodiment;

[0026] Figure 2 A schematic diagram of the structure of an adaptive control system for a reciprocating cutting tool provided in an embodiment of this application;

[0027] Figure 3 An operational logic diagram of an adaptive control method for a reciprocating cutting tool provided in an embodiment of this application;

[0028] Figure 4 This is a schematic diagram of the structure of a jigsaw provided in an embodiment of this application. Detailed Implementation

[0029] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, specific embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0030] See Figures 1 to 4 This application provides an adaptive control method for a reciprocating cutting tool, comprising: real-time acquisition of the operating current signal of the drive motor, the reciprocating motion frequency signal of the saw blade, and the triaxial vibration signal of the saw blade; calculation of the average value component and the peak fluctuation component based on the operating current signal, and calculation of the vibration energy component in the cutting direction based on the triaxial vibration signal of the saw blade; fusion of the average value component, the peak fluctuation component, and the vibration energy component in the cutting direction according to a preset weighted fusion rule to generate a comprehensive load index for the current cutting cycle; inputting the comprehensive load index, the deviation between the reciprocating motion frequency and its set frequency, and the rate of change of the comprehensive load index in the most recent preset number of cycles into a preset adaptive strategy mapping model; outputting a load control command from the adaptive strategy mapping model, the load control command including feed speed adjustment, motor power adjustment, and torque curve adjustment parameters; and coordinating the adjustment of the feed speed of the reciprocating cutting tool, the output power of the drive motor, and the waveform characteristics of the instantaneous torque output curve of the drive motor in a single reciprocating cycle according to the load control command.

[0031] Specifically, in this embodiment, the working current signal is acquired by a current sensor connected in series in the power supply circuit of the drive motor. The working current signal can reflect the real-time load and electrical characteristics of the motor. The reciprocating motion frequency signal is acquired by an encoder or Hall sensor installed on the motor shaft or crank connecting rod mechanism. The reciprocating motion frequency signal can provide accurate information on the number of reciprocations per second of the saw blade. The triaxial vibration signal of the saw blade is acquired by a triaxial acceleration sensor fixed to the tool head near the saw blade clamping point. The triaxial vibration signal of the saw blade includes the vibration acceleration of the saw blade in the cutting direction, transverse direction and vertical direction during the cutting process. The adaptive strategy mapping model can learn the comprehensive load index, the deviation between the reciprocating motion frequency and its set frequency, and the complex mapping relationship between the rate of change of the comprehensive load index in the most recent preset number of cycles and the optimal control response. It outputs a set of load control commands, including feed speed adjustment, motor power adjustment, and torque curve adjustment parameters. The feed speed adjustment is used to change the speed of the feed motor or the step pulse frequency through the feed servo controller, thereby adjusting the speed at which the reciprocating cutting tool cuts into the workpiece. The motor power adjustment is used to adjust the current or voltage setpoint of the drive motor through the motor power driver, thereby changing its output power. The torque curve adjustment parameters are used to shape the instantaneous torque output curve of the drive motor in a single round-trip cycle through the dynamic torque curve controller.

[0032] In one embodiment of this application, generating the comprehensive load index for the current cutting cycle specifically includes: taking the arithmetic mean of the working current signal over a complete cutting cycle as the average value component; taking the ratio of the peak value to the root mean square value of the working current signal over a complete cutting cycle as the fluctuation peak value component; performing triaxial decomposition on the triaxial vibration signal of the saw blade, integrating the square of the vibration acceleration signal in the cutting direction over the complete cutting cycle to obtain the vibration energy component in the cutting direction; and normalizing the average value component, the fluctuation peak value component, and the vibration energy component in the cutting direction, multiplying them by their respective preset weighting coefficients, and summing them to obtain the comprehensive load index.

[0033] Specifically, the average value component is the arithmetic mean of the working current signal within the current complete cutting cycle, used to reflect the steady-state load level; the peak fluctuation component is the ratio of the peak value to the root mean square value of the current signal within the same cycle. This ratio is highly sensitive to instantaneous impact current and can effectively characterize the severity of load fluctuations; the vibration energy component is obtained by coordinate decomposition of the triaxial vibration signal, extracting the acceleration signal in the cutting direction, and numerically integrating the square of this acceleration signal over a complete cutting cycle. The result represents the vibration energy in the cutting direction within that cycle, directly reflecting the mechanical force and impact intensity. The average value component, peak fluctuation component, and vibration energy component in the cutting direction are normalized to eliminate dimensional differences. The normalized components are multiplied by preset weighting coefficients and summed to obtain the comprehensive load index. The comprehensive load index integrates electrical steady-state, electrical transient, and mechanical vibration information, and can better characterize the overall load than a single signal. The preset weighting coefficients are a set of constant values ​​pre-calibrated using a linear regression algorithm, based on experimental data from a large number of typical operating conditions, with the goal of achieving optimal control performance.

[0034] In one embodiment of this application, the adaptive strategy mapping model is a trained multi-input multi-output feedforward neural network model; wherein, the feedforward neural network model takes the comprehensive load index, frequency deviation and the rate of change of the comprehensive load index as inputs, and outputs the feed speed adjustment amount, motor power adjustment amount and torque curve adjustment parameters.

[0035] Specifically, feedforward neural networks can establish a direct, end-to-end mapping relationship from multidimensional inputs to multidimensional outputs within a unified model, avoiding policy conflicts that may arise from designing separate controllers for each output and ensuring the inherent synergy of adjustment actions. Furthermore, a fully trained neural network model can make reasonable inferences about operating conditions that do not directly appear in the training data but have similar characteristics. This allows the system to make effective decisions when facing new materials or atypical cutting conditions, enhancing the robustness of adaptive control. In this embodiment, the feedforward neural network is specifically a multilayer perceptron neural network.

[0036] In one embodiment of this application, adjusting the waveform characteristics of the instantaneous torque output curve of the drive motor in a single round-trip cycle specifically includes: dynamically adjusting the parameters of the proportional-integral controller of the drive motor current loop within a preset time window before and after each reversal point of the saw blade movement, or injecting a dynamic feedforward compensation signal into the current loop to change the steepness of the rise and fall of the drive motor output torque during the reversal process, based on the torque curve adjustment parameters.

[0037] Specifically, in this embodiment, a high steepness means decisive and rapid commutation, which is beneficial for improving commutation efficiency; a low steepness means a smooth and gentle commutation process, which can significantly reduce shock and vibration. In the drive motor, the motor torque is proportional to the current. The current loop is the innermost and fastest-responding control loop, which directly determines the instantaneous torque output of the motor. The parameters of the proportional-integral controller of the current loop include proportional gain and integral time constant. Temporarily increasing the proportional gain and / or decreasing the integral time constant within the commutation window can make the torque rise or fall rapidly; temporarily decreasing the proportional gain and / or increasing the integral time constant within the commutation window. The feedforward compensation signal is an additional current command directly superimposed on the target command of the current loop. It is calculated based on a known system model and load mutations. Injecting a feedforward current command with steep rising and falling edges can increase the steepness of the output torque; injecting a feedforward current command with smooth changes can reduce the steepness of the output torque.

[0038] In one embodiment of this application, the method further includes a step for avoiding mechanical resonance, specifically including: performing synchronous time-frequency analysis on the real-time acquired triaxial vibration signal of the saw blade and the working current signal; extracting the frequency band energy of the triaxial vibration signal of the saw blade near the reciprocating motion frequency and its harmonics as a first vibration characteristic quantity; extracting the amplitude of at least one harmonic component in the spectrum of the working current signal that is associated with the pre-stored natural frequency of the tool mechanical structure as a current-related characteristic quantity; performing a fusion judgment based on the first vibration characteristic quantity and the current-related characteristic quantity, and according to a preset resonance risk judgment rule; if the judgment result indicates that there is a resonance risk, generating a frequency fine-tuning command; the frequency fine-tuning command is used to change the control signal of the drive motor, thereby causing the reciprocating motion frequency of the saw blade to deviate from the current value, so as to deviate it from the natural frequency range of the tool mechanical structure.

[0039] Specifically, resonance occurs when the reciprocating frequency of the saw blade approaches the natural frequency of the tool's mechanical structure. Resonance leads to a sharp amplification of vibration amplitude, generating alternating stress far exceeding the design range, potentially causing mechanical failures such as saw blade fatigue fracture and bearing damage, significantly shortening tool life. Furthermore, resonance makes the saw blade's trajectory unstable, resulting in a rough cutting surface and skewed cuts. In this embodiment, synchronous time-frequency analysis of the saw blade's triaxial vibration signal and the drive motor's operating current signal using short-time Fourier transform or wavelet transform allows us to understand how frequency components change over time, thereby capturing transient resonance phenomena. Extracting the energy of the reciprocating frequency and its harmonic bands from the triaxial vibration signal spectrum is crucial because during normal cutting, vibration energy is mainly concentrated at these frequencies directly related to the motion. Once resonance occurs, the energy in the bands near the natural frequency increases abnormally, and even if this frequency is not exactly the same as the motion frequency, it will still be captured by this method. The harmonic component amplitudes associated with the pre-stored natural frequencies of the tool's mechanical structure are extracted from the spectrum of the motor's operating current. This is because when mechanical resonance occurs, the enormous vibration resistance reacts to the motor, manifesting as a periodic load fluctuation. This load fluctuation modulates the motor's current, causing sidebands or harmonic components related to the mechanical natural frequencies to appear in the current spectrum. Extracting the amplitudes of these components can very sensitively reflect the resonance state. The vibration energy characteristics and current harmonic characteristics are fused and analyzed according to preset resonance risk assessment rules. This multi-information fusion judgment is more accurate and has stronger anti-interference capabilities than judgment based on a single vibration signal. The first vibration characteristic quantity and the current-related characteristic quantity need to be normalized, and each is divided by its baseline value under calibrated operating conditions to obtain a dimensionless ratio. The preset resonance risk assessment rules include a danger threshold; if both the vibration characteristic quantity and the current characteristic quantity exceed the danger threshold simultaneously, a resonance risk is determined to exist.

[0040] Once a resonance risk is detected, the system generates a frequency fine-tuning command to cause a small, rapid shift in the reciprocating frequency of the saw blade, quickly moving the operating frequency away from the resonant frequency band of the mechanical structure, thereby breaking the resonance condition and causing the vibration to decay rapidly.

[0041] In one embodiment of this application, the coordinated adjustment specifically includes the following priority logic: when the comprehensive load index exceeds the first safety threshold corresponding to the current cutting material, the operation of reducing the feed rate adjustment amount is executed first; or, when resonance risk is detected, the frequency fine-tuning command is executed first.

[0042] Specifically, when the comprehensive load index exceeds the safety threshold or a resonance risk is detected, the system will immediately interrupt the regular multi-parameter collaborative optimization process and prioritize the execution of the most direct and effective single risk avoidance operation. This prevents the system from delaying the handling of emergency dangers due to being busy optimizing other parameters, and provides the highest priority guarantee for core safety.

[0043] In one embodiment of this application, the method further includes a material type identification and initial parameter matching step: during the initial cutting stage of the cutting operation, the reciprocating cutting tool is controlled to cut into the workpiece at a constant first feed rate, and the initial triaxial vibration signal and initial current signal of this stage are collected; the spectral distribution characteristics of the initial triaxial vibration signal and the steady-state amplitude characteristics of the initial current signal are extracted; the extracted features are matched with the feature ranges of different material types pre-stored in the material process parameter database to determine the material type range to which the current workpiece belongs; according to the determined material type range, the corresponding initial set frequency, initial feed rate and initial torque curve parameters are called from the material process parameter database and used as the starting reference for the adaptive control process of this cutting operation.

[0044] Specifically, in this embodiment, by identifying the material type and matching the initial parameters, the adaptive adjustment can be started with the optimal control parameters for the wood type, avoiding violent oscillations during the start-up phase of the adaptive adjustment and improving the overall control efficiency and stability of the system. The parameter configuration provided by the initial reference is already very close to the optimal operating state. Adjusting directly from the initial reference can reach the optimal operating state faster and more smoothly than adjusting from parameters far from the ideal operating state.

[0045] This application also provides an adaptive control system for a reciprocating cutting tool, used to implement an adaptive control method for the reciprocating cutting tool. The system includes: a sensor module, including a current sensor for acquiring the operating current of the drive motor, an encoder or Hall sensor for acquiring the reciprocating frequency of the saw blade, and a triaxial vibration acceleration sensor disposed on the tool head; a signal processing and calculation module, electrically connected to the sensor module, for processing the signals from the sensor module in real time and calculating the comprehensive load index, frequency deviation, and rate of change of the comprehensive load index; an adaptive strategy control module, embedding an adaptive strategy mapping model, whose input is connected to the signal processing and calculation module, for receiving the comprehensive load index, frequency deviation, and rate of change of the comprehensive load index, and generating feed speed adjustment, motor power adjustment, and torque curve adjustment parameters; and a multi-objective cooperative execution module, including a motor power driver, a feed servo controller, and a dynamic torque curve controller; the multi-objective cooperative execution module is connected to the adaptive strategy control module, for receiving and executing the adjustment amounts and parameters to coordinately control the drive motor and the feed mechanism.

[0046] The system also includes a material and process parameter database and a self-learning module. The material and process parameter database stores the characteristic ranges of different material types and their corresponding initial working parameters. The self-learning module receives feedback signals on the cutting quality of the task after a cutting task is completed, and updates the decision weight parameters in the adaptive strategy mapping model using a reinforcement learning algorithm based on the feedback signals.

[0047] Specifically, such as Figure 2 and Figure 4 As shown in this embodiment, an adaptive control system for a reciprocating cutting tool can be integrated into a jigsaw. This system, with a microcontroller unit at its core, constructs a closed-loop intelligent control framework designed to perceive the cutting status in real time, make intelligent decisions, and execute collaboratively to cope with complex cutting conditions. The system mainly includes a sensor module, a signal processing and calculation module, an adaptive strategy control module, a multi-objective collaborative execution module, a material and process parameter database, and a self-learning module; among which, the sensor module includes a current sensor, a motion frequency sensor, and a triaxial vibration acceleration sensor. The system workflow includes: At the start of the cutting operation, the system first performs a material identification step: cutting at a constant low speed, acquiring initial signals and matching them with the material process parameter database, and calling the optimal initial parameters to set the basic parameters of frequency, feed speed, and torque curve; after entering the adaptive cutting stage, the sensor module continuously acquires signals, which are processed and calculated by the signal processing and calculation module, and then sent to the multilayer perceptron neural network of the adaptive strategy control module for decision-making; the load control command generated by the decision is sent to the multi-objective collaborative execution module; the multi-objective collaborative execution module collaboratively controls the motor power driver and feed servo controller, respectively adjusting the output characteristics of the motor and the tool's advance speed to achieve adaptive control that can cope with load changes, suppress vibration, and avoid resonance. The self-learning module performs a comprehensive score on the cutting quality after a complete cutting task; it compares the actual cutting quality of this task with the expected cutting quality generated by the adaptive strategy mapping model, generating an evaluation feedback signal; and based on the evaluation feedback signal, it uses a reinforcement learning algorithm to fine-tune the internal parameters of the adaptive strategy mapping model so that subsequent decisions under similar working conditions can approach a better cutting quality.

[0048] In one embodiment of this application, the dynamic torque curve controller is part of the motor power driver and is configured to dynamically adjust the response characteristics of the current loop or superimpose a feedforward current command based on torque curve adjustment parameters.

[0049] Specifically, in this embodiment, the dynamic torque curve controller is a hardware unit embedded in the motor power driver. It can dynamically change the shape of the motor's output torque in real time according to instructions from the adaptive strategy mapping model. In reciprocating cutting tools, at the reversing point, the saw blade has enormous inertia, requiring the motor to instantly provide a large reverse torque to control the blade. If the torque curve is too steep, it will generate mechanical shock, leading to severe vibration and noise; if the torque curve is too flat, it will result in slow reversal, weak cutting, or even jamming. Furthermore, during adaptive control of the reciprocating cutting tool, the load changes within milliseconds when cutting different materials or encountering knots or nails; the torque needs to predictably or rapidly follow these changes. Therefore, using a dynamic torque curve controller allows for precise management of the motor's torque changes in each reciprocating cycle.

[0050] When the reciprocating cutting tool adaptive control system of this application is applied to a jigsaw, the jigsaw can automatically identify materials, control dynamic torque curves, avoid resonance in real time, and adjust the comprehensive load index in a coordinated manner to achieve professional-grade cutting quality; it enables the jigsaw to obtain smooth and perpendicular cut surfaces on any material; at the same time, it greatly improves safety and controllability, effectively prevents saw jamming, saw breakage, and backlash, and reduces handle vibration; furthermore, by maintaining the optimal state in real time, it can reduce ineffective energy consumption, significantly improve cutting efficiency and tool energy efficiency; and it can stabilize the load, eliminate resonance and impact, and extend the life of the tool and saw blade.

[0051] While this application discloses the above information, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of this application; therefore, the scope of protection of this application shall be determined by the scope defined in the claims.

Claims

1. A method of adaptive control of a reciprocating cutting tool, c h a r a c t e r i s e d in that, The method comprises: real-time acquisition of a working current signal of a driving motor, a reciprocating motion frequency signal of a saw blade, and a three-axis vibration signal of the saw blade; calculation of an average value component and a fluctuation peak value component based on the working current signal, and calculation of a vibration energy component in a cutting direction based on the three-axis vibration signal of the saw blade; normalization of the average value component, the fluctuation peak value component, and the vibration energy component in the cutting direction, multiplication by respective preset weight coefficients, and summation to obtain a comprehensive load index; input of the comprehensive load index, a deviation of the reciprocating motion frequency from a set frequency, and a change rate of the comprehensive load index in a preset number of cycles into a preset adaptive strategy mapping model, wherein the adaptive strategy mapping model is a trained multiple-input multiple-output feedforward neural network model; output of a load control instruction from the adaptive strategy mapping model, the load control instruction including a feed speed adjustment amount, a motor power adjustment amount, and a torque curve adjustment parameter; coordinated adjustment of a feed speed of the reciprocating cutting tool, an output power of the driving motor, and a waveform feature of an instantaneous torque output curve of the driving motor in a single reciprocating cycle according to the load control instruction.

2. The reciprocating cutting tool adaptive control method of claim 1, wherein, The comprehensive load index is obtained by specifically: taking an arithmetic mean value of the working current signal in a complete cutting cycle as the average value component; taking a ratio of a peak value to a root mean square value of the working current signal in the complete cutting cycle as the fluctuation peak value component; performing three-axis decomposition on the three-axis vibration signal of the saw blade, and integrating a square of a vibration acceleration signal in the cutting direction in the complete cutting cycle to obtain the vibration energy component in the cutting direction.

3. The reciprocating cutting tool adaptive control method of claim 1, wherein, The feedforward neural network model takes the comprehensive load index, the frequency deviation, and the change rate of the comprehensive load index as inputs, and outputs the feed speed adjustment amount, the motor power adjustment amount, and the torque curve adjustment parameter.

4. The reciprocating cutting tool adaptive control method of any one of claims 1 or 3, wherein, The waveform feature of the instantaneous torque output curve of the driving motor in the single reciprocating cycle is adjusted by specifically: dynamically adjusting parameters of a proportional-integral controller of a current loop of the driving motor or injecting a dynamic feedforward compensation signal into the current loop within a preset time window before and after each motion reversal point of the saw blade to change steepness of rising and falling of the output torque of the driving motor during the reversal according to the torque curve adjustment parameter.

5. The reciprocating cutting tool adaptive control method of claim 1, wherein, The method further comprises steps for avoiding mechanical resonance, specifically comprising: synchronous time-frequency analysis of the real-time acquired three-axis vibration signal of the saw blade and the working current signal; extracting a frequency band energy of the three-axis vibration signal of the saw blade near the reciprocating motion frequency and its harmonics as a first vibration feature quantity; extracting an amplitude of at least one harmonic component associated with a pre-stored tool mechanical structure inherent frequency in a frequency spectrum of the working current signal as a current correlation feature quantity; and fusing judging based on the first vibration characteristic quantity and the current correlation characteristic quantity according to a preset resonance risk judging rule; if the judging result indicates that there is a resonance risk, a frequency fine-tuning instruction is generated; The frequency fine-tuning instruction is used to change the control signal of the driving motor, so that the reciprocating motion frequency of the saw blade deviates from the current value, so as to deviate from the inherent frequency range of the tool mechanical structure.

6. The reciprocating cutting tool adaptive control method of claim 5, wherein, The synergistic regulation specifically includes the following priority logic: When the comprehensive load index exceeds a first safety threshold corresponding to the current cutting material, the operation of reducing the feed speed adjustment amount is preferentially performed; or, When the resonance risk is detected, the frequency fine-tuning instruction is preferentially executed.

7. The reciprocating cutting tool adaptive control method of claim 1, wherein, The method further includes a material type identification and initial parameter matching step: In the initial cutting-in stage of the cutting operation, the reciprocating cutting tool is controlled to cut into the workpiece at a constant first feed speed, and initial three-axis vibration signals and initial current signals in this stage are collected; The frequency spectrum distribution feature of the initial three-axis vibration signals and the steady-state amplitude feature of the initial current signals are extracted; The extracted features are matched with feature ranges of different material types pre-stored in a material process parameter database to determine a material type range to which the current workpiece belongs; According to the determined material type range, corresponding initial set frequency, initial feed speed and initial torque curve parameters are called from the material process parameter database and used as starting references for the adaptive control process of the current cutting operation.

8. A reciprocating cutting tool adaptive control system characterized by, A system for implementing the adaptive control method of the reciprocating cutting tool according to any one of claims 1-7, the system comprising: a sensor module including a current sensor for collecting driving motor working current, an encoder or a Hall sensor for collecting saw blade reciprocating motion frequency, and a three-axis vibration acceleration sensor arranged at a tool head; a signal processing and calculation module electrically connected to the sensor module, for real-time processing of signals of the sensor module and calculating the comprehensive load index, the frequency deviation and the change rate of the comprehensive load index; an adaptive strategy control module having the adaptive strategy mapping model embedded therein, with an input end connected to the signal processing and calculation module, for receiving the comprehensive load index, the frequency deviation and the change rate of the comprehensive load index, and generating the feed speed adjustment amount, the motor power adjustment amount and the torque curve adjustment parameter; a multi-target synergistic execution module including a motor power driver, a feed servo controller and a dynamic torque curve controller; the multi-target synergistic execution module is connected to the adaptive strategy control module, for receiving and executing the adjustment amount and parameter to synergistically control the driving motor and the feed mechanism.

9. The reciprocating cutting tool adaptive control system of claim 8, wherein, The system further includes a material process parameter database and a self-learning module; The material process parameter database stores feature ranges of different material types and corresponding initial working parameters; The self-learning module receives an evaluation feedback signal about the cutting quality of the current task after completion of a cutting task, and updates decision weight parameters in the adaptive strategy mapping model according to the feedback signal using a reinforcement learning algorithm.

10. The reciprocating cutting tool adaptive control system of claim 8, wherein, The dynamic torque profile controller is part of the motor power driver and is configured to dynamically adjust the response characteristic of the current loop or to superimpose a feed forward current command depending on the torque profile adjustment parameters.

Images