Vibration mitigation with active damping

Abstract

The invention relates to a tooling system (1) for the machining of metal comprising a module (10) for active damping of vibrations with a sensor unit (11), an actuators unit (16) and a control unit (13) communicating data between the sensor unit (11) and the actuator unit (16).

Description

[0001] Vibration mitigation with active damping

[0002] Description

[0003] The invention relates to a tooling system for machining metal with a module for active damping of vibrations.

[0004] The machining of metal can be realized, for example, by a cutting process. A machining process is a manufacturing process in metalworking in which material is removed from a workpiece by removing small chips of material in order to achieve the desired shape, surface or dimensions.

[0005] This is done by bringing a tool with a cutting edge into contact with the workpiece in such a way that material can be removed. In this manufacturing process, either the tool or the workpiece moves, or both move relative to each other. When the cutting edge comes into contact with the workpiece, the movement results in chip formation and thus material removal. Machining processes are counted as subtractive manufacturing.

[0006] For a machining process, machine tools are used to control the relative movement between the cutting tool and the workpiece in a targeted and defined manner.

[0007] Such a machine tool is, for example, a lathe. In a lathe, the workpiece rotates and the cutting tool moves parallel and / or perpendicular to the axis of rotation. In the process, the distance between the cutting tool and the workpiece can be reduced in order to bring the tool into contact with the workpiece for chipping.

[0008] In a milling machine, the cutting tool rotates and can move along the xyz axes to create the shaping of the workpiece. Drilling machines are used to create cylindrical recesses in metal workpieces. In this process, the tool rotates and moves linearly into the workpiece. Grinding machines use wheels to very finely remove material from the surface of the workpiece and to achieve a high surface finish.

[0009] State-of-the-art CNC machine tools, short for Computerised Numerical Control machine tool, are used as highly automated machines to precisely machine and shape workpieces. These machines are equipped with a control system to control the movements of tools and workpieces.

[0010] The tool holder is the component that securely holds the cutting tool, such as an end mill or drill, during machining and realizes the connection to the machine tool. The toolholder is the first component to directly experience the vibrations caused by the cutting process between the cutting tool and the workpiece.

[0011] Passive damping toolholders are already used today to reduce these vibrations. Such passive damping is based on the inherent mechanical properties of the toolholder or on the addition of passive components to reduce vibrations. Passively damping toolholders can contain materials such as vibration damping alloys, tuned mass dampers or special geometries that dissipate the vibration energy.

[0012] However, passive damping toolholders also have limitations in terms of adaptability and effectiveness under different conditions. They do not include active control systems and they usually do not include sensors or actuators.

[0013] EP 1 677 932 B1 discloses a device for damping vibration and / or controlling deflection of an object during machining, wherein the object is a tool, a tool holder or a workpiece. The device comprises at least one force exchange device arranged outside a surface of the object, the force exchange device being attached to a locating device surrounding the object and either exchanging a force having a force component perpendicular to the surface of the object or exchanging a moment between the object and the device directly or via a mechanical lever.

[0014] DE 10 2015 122 758 A1 describes a tool holder having a cutting tool mounted on a head attached to a neck at a first end of the tool holder. A shank is arranged at a second, opposite end of the tool holder. A central cavity extends inwardly from the first end to the shank. A passive vibration damping arrangement is disposed in the central cavity. The passive vibration damping arrangement comprises an absorber body and an absorber cover attached to the absorber body. The absorber body has a first end, a second end opposite the first end, and one or more cavities formed in the second end.

[0015] However, with passive damping, the degree of damping is limited. Also, the reaction speed to sudden vibrations cannot be influenced and is dependent on the type of vibration damping arrangement installed.

[0016] The object of the present invention is to provide a tooling system for machining metal which can damp vibrations actively. The tool system should be able to react actively to the type and extent of the vibrations. In addition, the tool system should also be able to dampen different vibrations very quickly.

[0017] According to the invention, this task is ensured by a tool system for machining metal with a module for active damping of vibrations. Preferred variants can be found in the subclaims, the description and the drawing.

[0018] According to the invention, the module for active damping of vibrations comprises a sensor unit, an actuator unit and a control unit which communicates data between the sensor unit and the actuator unit. For example, the sensor unit comprises at least one sensor for determining the bending force and / or the speed and / or the acceleration and / or the temperature and / or the vibration and / or the cutting force.

[0019] In one variant, the sensor unit is integrated directly in the tool holder. The detection of the vibrations can thus take place directly and particularly quickly, which is advantageous for real-time influencing by means of active damping. In an alternative variant a sensor unit can additionally be arranged on the drive of the tool holder.

[0020] In another variant, the sensor unit is arranged on the drive in which the tool holder is installed.

[0021] The sensor unit has, for example, at least one sensor that is placed in the machining environment. This at least one sensor continuously determines, for example, the vibrations on the tool holder and / or on the cutting tool and / or on the workpiece. In doing so, the sensor provides, for example, real-time data that is transmitted directly and without interruption and in real-time to the control unit.

[0022] The actuator unit receives control signals from the control unit and makes precise adjustments to the drive elements and / or damping mechanisms to optimize the machining process for the lowest possible vibrations.

[0023] The actuator unit of the module is a component responsible for the control and movement of, for example, the tool holder. Exemplarily, the actuator unit is adapted to the CNC control of the machine tools to perform precise movements and actively dampen vibrations.

[0024] In one variation of the invention, the actuator unit comprises servo motors for precise and accurate control of the tool holder. In another variant of the invention, the active damping module has direct control access to the servomotors of the drive unit of the CNC machine tool.

[0025] In an alternative variant, the actuator unit comprises at least one hydraulic cylinder to control the movement of the tool holder and / or the movement of the workpiece. In this case, the control unit can actively dampen vibrations on the tool holder by controlling the hydraulic pressure, for example.

[0026] In another alternative variant of the invention, the actuator unit comprises at least one pneumatic actuator. In this case, the control unit can control the actuator unit in such a way that, for example, compressed air is used to actively dampen vibrations of the tool holder and / or the workpiece.

[0027] In a variant of the invention, the actuator unit comprises at least one linear guide. In this case, vibrations can be actively damped, for example, by adapting the linear movement.

[0028] For example, the actuator unit may also comprise at least one electromagnetic actuator and at least one piezoelectric element for actively damping vibrations.

[0029] For example, the actuator unit may also comprise a combination of said motors, cylinders or guides.

[0030] Advantageously, the active damping module suppresses unwanted oscillations or vibrations during the machining of metal by cutting machine tools. The symbiotic communication of the sensor unit with the control unit and the actuator unit aims to control and reduce the energy or motion generated by the vibrations, rather than simply passively absorbing or isolating them.

[0031] The active damping module works by means of a feedback loop in which the sensor unit continuously determines data and transmits it to the control unit, the control unit analyses the data and, for active damping, adjusts the actuator unit accordingly to minimise or eliminate the vibrations.

[0032] In contrast to passive damping, in which energy is absorbed in the form of heat or movement, active damping realizes a targeted reduction of the vibrations or achieves a suppression of the vibrations almost completely, which at the same time leads to a higher energy efficiency and an improvement in the performance of the machining machine tools.

[0033] In one variant of the invention, the module for active damping of vibrations is fully integrated into the tool holder.

[0034] In an alternative variant of the invention, the module for active damping is at least partially integrated into the tool holder. For example, part of the module is implemented in the machine stand, thereby realizing short signal paths to the drive of the cutting tool.

[0035] For example, the module for active damping of vibrations comprises a preprocessing unit for modulating signals from the sensor unit to the control unit.

[0036] In this process, the modulation of the signals changes certain properties of a carrier signal depending on a low-frequency information source. This serves to encode the information content of the signal for transmission to the control unit and for storage and processing in the control unit.

[0037] For this purpose, the carrier signal can be in the form of a high-frequency sinusoidal wave or in the form of a continuous signal. The frequency, the current strength and / or the voltage of the carrier signal are defined in such a way that the transmission by means of the transmission channel is particularly fast and reliable. For example, the modulation changes one or more characteristics of the signal based on the characteristics of the sensor unit. For this purpose, preferably the amplitude and / or the frequency and / or the phase is modulated.

[0038] In a variant of the invention, demodulation of the signal is performed in the control unit to extract the modulated information from the carrier signal. The demodulation thereby reconstructs the original information of the sensor unit.

[0039] In another embodiment of the invention, the active vibration damping module comprises a post-processing unit for modulating signals from the control unit to the actuators.

[0040] For example, the module for active damping of vibrations comprises a programming unit for the control unit.

[0041] A programming unit for a control unit is a component used in many technical systems and devices to develop, edit or update the software or programmed logic that enables the control of the system. The function of such a programming unit is to allow the developers to create the desired programmes or instructions for the control unit and transfer them to the control unit. Here are the basic functions and features of a programming unit for a control unit:

[0042] For example, the programming unit comprises an integrated development environment within which the programming code of the adaptive damping module for the control unit can be edited, adapted and / or further developed. This is advantageous, for example, in the case of an adaptive embodiment of the module, as this allows the adaptations to be implemented directly.

[0043] In principle, a complete code can be written with the programming unit of the control unit to define the desired functions of the active damping. For example, the programming unit is advantageously designed for autonomous adaptations of the code.

[0044] For example, adaptations can be stored in a version chronology in the programming unit of the control unit. In a variant of the invention, the code adaptations can additionally be transmitted via various interfaces such as USB, Ethernet or serial connections on a development platform.

[0045] The programming unit may also comprise, for example, functions for versioning and storing codes.

[0046] In one variant of the invention, the programming unit of the control unit is used to perform firmware updates, which are made available via external interfaces, for example.

[0047] For example, the programming unit can also implement security mechanisms and access controls to ensure exclusively authorized access to the control unit.

[0048] In one variant of the invention, the control unit processes the data asynchronously.

[0049] In another variant of the invention, the control unit processes the data synchronously.

[0050] The term asynchronous data processing refers to a data processing approach in which tasks or processes do not need to be synchronized. The processing does not take place in a fixed temporal rhythm. Instead, the control unit of the active vibration damping module processes the data based on events, spontaneous requests or conditions, without the influence of a clock signal. This makes it possible to react directly and immediately to vibrations that occur and to initiate the active damping of these vibrations. In the asynchronous data processing of the module for active damping of vibrations, the tasks or the processes take place independently of each other and can directly initiate damping when certain conditions are met or also when external events occur.

[0051] The asynchronous data processing of the module is advantageously flexible and adaptable, especially compared to synchronized data processing, as it can react better and directly to an unexpected vibration.

[0052] At the same time, the asynchronous data processing of the module for active damping of vibrations enables the simultaneous execution of several processes for active damping, for example a speed change and a position correction of the tool holder, without the processes interfering with each other. In this respect, the module for active damping is capable of multitasking due to the asynchronous data processing.

[0053] Due to the asynchronous processing of the data, the control unit of the module is advantageously efficient and energy-saving, since resources are only used when vibrations are to be damped.

[0054] The module for active damping of vibrations, in particular the sensor unit, the control unit and the actuator unit, communicate on the basis of asynchronous data processing on request, i.e. when an event occurs, in particular when vibrations occur.

[0055] For example, the control unit processes the data depending on the event.

[0056] Event-driven processing of data in the control unit of the active vibration damping module refers to a data processing approach in which the processing of information or data is triggered upon the occurrence of specific events or conditions, namely vibrations. In this context, data is not processed continuously or periodically, but processing is done in response to specific, predefined events, in particular the vibrations on the cutting tool of a machine tool.

[0057] In contrast to continuous or periodic processing, where data is constantly monitored, the processing of data in the event-dependent module for active damping of vibrations usually remains idle until a triggering vibration occurs. The processing of the control unit then reacts immediately to the vibration, which allows the reaction to be particularly fast and immediate.

[0058] For example, the control unit of the module for active damping of vibrations processes data suitable for real-time data streams.

[0059] The real-time data streams of the active damping module are immediate data streams in which information on vibrations is acquired, transmitted and processed into active damping measures in real time. These data streams are transmitted from one unit of the module to another without noticeable delay or with minimal latency.

[0060] In one variation of the invention, the control unit of the module for active damping of vibrations processes data suitable for dynamic environments.

[0061] The dynamic environment of data processing refers to a situation or context in which data is acquired, processed and used in a constantly changing or adapting manner. The vibrations during machining of workpieces can be of a very different nature and subject to permanent change.

[0062] For example, the sensor unit comprises at least one combination of one or more memristors and one or more piezoelectric elements. In one example the control unit comprises a system and / or structure of programable resistive elements. These elements can be memristors for example.

[0063] In one embodiment of the invention, the control unit comprises one or a plurality of memristors.

[0064] In one embodiment of the invention, the control unit is formed as a neuromorphic machine based on memristors.

[0065] For example, memristors in the neuromorphic machine enable the control unit of the active vibration damping module to learn and adapt autonomously to optimize the damping performance.

[0066] A memristor is an electrical component designed as a type of non-linear resistive element. Memristors have the special property that they can store information about the electrical charge or current flowing into them and retain this information even if the current supply is removed. This behavior advantageously forms a kind of synaptic memory.

[0067] Memristors persist in a certain state and store information about the electrical charge conducted into them. This makes them a non-volatile memory in the module for active damping of vibrations.

[0068] The electrical conductivity of a memristor does not depend linearly on the applied voltage or current. Instead, it varies depending on the previous state, in particular on the number as well as the intensity of the input systems e.g. vibrations.

[0069] For example, the control unit of the module for active damping of vibrations is designed as a neuromorphic machine for example in-memory computing architecture. In memory computing is an alternative to von Neumann architecture. The neuromorphic architecture of the control unit is particularly energy-efficient, so that it can also advantageously dampen vibrations in the energy-limited environment of a tool holder.

[0070] The module's control unit processes data in an event-driven and spiking manner, comparable to biological neural networks. The module's control unit responds to vibrations in real time, which allows for immediate adjustment of the damping process.

[0071] In one embodiment, the control unit is a neuromorphic machine that uses memristors as synaptic elements. These memristors exhibit adaptive learning and synaptic plasticity, allowing the system to adapt itself on the basis of the vibrations occurring in the tool holder or cutting tool or workpiece.

[0072] The memristive neuromorphic control unit of the module for active damping of vibrations also adapts to changing machining conditions during cutting of the workpiece without manual intervention, thus overcoming the limitations of conventional control systems of machine tools.

[0073] In one variant of the invention, the control unit of the module for active damping of vibrations enables adaptive learning. As a result, the module, and thus also the tool holder of the cutting tool, continually adapts to the workpiece to be machined and the vibrations that occur with it.

[0074] For example, the control unit of the module for active damping of vibrations enables autonomous adjustments. As a result, the active damping module automatically adapts to the vibrations during the machining of workpieces without any externally initiated activity.

[0075] The innovative aspect of this invention lies, for example, in the event-driven and spiking behavior of the module's control unit. Unlike conventional control units based on fixed algorithms, the control unit according to the invention processes information asynchronously and reacts to vibrations as soon as they occur.

[0076] The event-driven and spike-like behavior of the control unit enables immediate adaptation to changing machining conditions.

[0077] Lower vibrations during the machining of workpieces result in better surfaces and higher metal removal rates, which improves the quality and efficiency of workpiece machining.

[0078] Lower vibrations during the machining of workpieces lead to better surfaces and higher metal removal rates, which improves the quality and efficiency of workpiece machining.

[0079] The module for active damping of vibrations, especially in metal machining, represents a significant advance in the field of workpiece machining. By integrating e.g. an event-driven or e.g. spiking neuromorphic control unit with memristors, this system overcomes the listed limitations of conventional passive damping methods. It offers real-time adaptability and improved machining quality, making it a valuable tool in the manufacturing industry.

[0080] In certain embodiments, the control unit may include a neuromorphic engine that mimics biological neural networks. This neuromorphic engine could utilize analog signal processing and / or spiking neural networks to detect and interpret patterns relevant to machining processes. The neuromorphic engine might operate on an event-driven basis, processing data asynchronously to reduce power consumption and enhance response speed.

[0081] The tooling system described includes a neuromorphic engine that features multiple parallel channels. These channels are each specifically designed to monitor distinct frequency bands, which allows for the precise detection of vibrational anomalies across a broad spectrum of signal frequencies.

[0082] The neuromorphic engine may also implement a Leaky Integrate-and-Fire (LIF) model, which processes incoming signals by integrating them over time and generating spike outputs when specific thresholds are reached. This approach enables the system to efficiently handle and respond to complex, dynamic machining environments.

[0083] The control unit may also interface directly with piezoelectric actuators, which could be positioned in a symmetrical configuration around the tool holder. This arrangement allows for uniform control of the tool holder in all three spatial dimensions, providing real-time adjustments that counteract vibrations and improve machining precision.

[0084] Optionally, feedback sensors might be integrated with the piezoelectric actuators to continuously monitor the position of the tool holder. This configuration could allow the control unit to fine-tune actuator responses in real-time, ensuring effective vibration damping and maintaining the optimal position of the tool holder during machining.

[0085] The system may also include a divisive normalization device within the control unit. This device might comprise an input module that receives an input spike train, a threshold calculation module that yields a threshold parameter based on the average values of the input spike train in an averaging window, and a normalization module that enables an integrate-and-fire counter. The counter could produce a single spike at the output and reset when the threshold parameter is reached.

[0086] The threshold calculation module could dynamically adjust the threshold parameter based on the variability of the input spike train within the averaging window, allowing the normalization module to adapt to different signal conditions. This enhances the accuracy and reliability of the system in processing signals from various machining environments.

[0087] The divisive normalization device may be integrated within the neuromorphic processor of the control unit to enhance real-time signal processing, particularly in applications such as feature extraction and classification.

[0088] In one variant of the invention the feature extraction is including vibration signals. In one special variant of the invention the vibration signals occur as audio signal.

[0089] Vibration signals can be characterized in three parts, as low-frequency vibrations. Below 20 Hz, vibration for example in slow oscillations in large structures or machinery. Also, as acoustic vibrations. In the audible range 20 Hz to 20 kHz, for example including sounds and noises perceivable by the human ear. There are also ultrasonic vibrations, which are above 20 kHz.

[0090] The control unit could include a normalized vibration feature extractor that processes an original vibration signal collected for example by a microphone, yielding a pre-normalized spike train for each channel. The divisive normalization device could then process the pre-normalized spike train to yield a postnormalized spike train.

[0091] A classifier within the control unit might execute classification tasks based on the output spikes of the normalized vibration feature extractor, utilizing a spiking neural network architecture for efficient and accurate classification. The classifier may be designed to support multi-modal input, allowing the system to integrate vibration features with other sensory data for enhanced pattern recognition and decision-making. Additionally, the tooling system may incorporate a programmable neuromorphic chip, which excels in low-dimensional signal processing. The neuromorphic chip combines an analog front end (AFE) that efficiently provides pre-processing functionality to input analog signals. This chip is highly re-configurable and scalable, supporting feed-forward, recurrent, reservoir, and other complex neural network structures. It can seamlessly integrate with a wide range of sensors, including MEMS microphones, thermal sensors, pressure sensors, vibration sensors, IMUs, gyros, and PPG sensors.

[0092] Additionally, the control unit of the tooling system incorporates a classifier that performs classification tasks based on the output spikes from a normalized vibration feature extractor. This classification process is achieved using a spiking neural network (SNN) architecture, ensuring both efficiency and accuracy in classification.

[0093] Moreover, the control unit is optimized for low-power operation and is designed to be integrated into real-time audio processing systems. These systems could include applications such as speech recognition devices, hearing aids, or other neuromorphic computing applications where low latency and high accuracy are essential.

[0094] Further advantages and features of the invention can be seen from the description of an example of an embodiment on the basis of a drawing and from the drawing itself.

[0095] The following is shown

[0096] Fig. 1 is a schematic representation of a machine tool. Fig. 1 shows a schematic representation of a machine tool 1 in the exemplary embodiment of a milling machine. In this embodiment, the frame 2 as the lowest assembly carries all other components. The frame 2 carries the machine stand 3, in which the motors and the gears for the work spindle 4 and feed are integrated.

[0097] The workpiece 5 is placed on the machine table 6 and is clamped there. In the version shown as a conventionally constructed milling machine 1, the machine table 6 is arranged on a cross slide 7 and a console 8 with guides of the frame 2. The console 8 moves in a vertical direction and the cross slide 7 in a horizontal direction.

[0098] The vertical work spindle 4 is arranged in the milling head, into which the tool holder 9 with the cutting tool is inserted. In the version shown, the module 10 for active damping is integrated in the tool holder 9.

[0099] The module 10 for active damping of vibrations comprises a sensor unit 11 , which in the embodiment shown comprises a force sensor, a speed sensor, an acceleration sensor, a temperature sensor and a vibration sensor. The sensor unit 11 sends the data to the pre-processing unit 12.

[0100] The pre-processing unit 12 modulates the data of the sensor unit 11 for processing in the control unit 13. The control unit 13 is designed as a neuromorphic control unit as a system of plurality of programable resistive elements in the embodiment of a plurality of memristors. The control unit 13 has a programming unit 14 which adapts the memristors to changed or suddenly occurring external boundary conditions, in particular vibrations during machining of the workpiece 5, and thus maps the adaptive element of the control unit 13.

[0101] The event-controlled, asynchronous and spike-shaped data processing of the control unit 13 realizes the immediate adaptation, i.e. in real time, to changing machining conditions in the form of vibrations and thus forms the core element for active damping.

[0102] The module 10 for active damping of vibrations comprises a post-processing unit 15, which modulates the data of the control unit 13 for the actuator unit 16. The control unit 13 communicates between the sensor unit 11 and the actuator unit 16 in order to actively dampen any vibrations that occur.

[0103] The actuator unit 15 actively intervenes in the drive of the milling machine and, for example, actively adjusts parameters such as speed and feed.

Claims

Vibration mitigation with active dampingClaims1. Tooling system (1 ) for the machining of metal comprising a module (10) for active damping of vibrations with a sensor unit (11 ), an actuators unit (16) and a control unit (13) communicating data between the sensor unit (11 ) and the actuator unit (16).

2. Tooling system according to claim 1 , characterized in that, the module (10) comprises a pre-processing unit (12) for modulating signals from the sensor unit (11 ) to the control unit (13), and a post-processing unit (15) for modulating signals from control unit (13) to the actuator unit (16).

3. Tooling system according to claim 1 or 2, characterized in that, the module (10) comprises a programming unit (14) for configuring the control unit (13) to process data asynchronously or synchronously as events occur, enabling real-time adjustments to the machining process.

4. Tooling system according to at least one of the preceding claims, characterized in that, the control unit (13) processes data suitable for realtime data streams in dynamic environments, enabling adaptive learning and autonomous adjustments to improve the tooling system's response to vibrational anomalies over time.RECTIFIED SHEET (RULE 91) ISA / EP5. Tooling system according to at least one of the preceding claims, characterized in that, the control unit (13) comprises a neuromorphic engine that mimics biological neural networks, utilizing spiking neutral networks (SNN), possibly combined with analog signal processing (ASP) to detect and interpret patterns relevant to machining processes.

6. Tooling system according to claim 5, characterized in that, the neuromorphic engine comprises multiple parallel channels, each designed to monitor specific frequency bands, enabling precise detection of vibrational anomalies across a wide range of signal frequencies.

7. Tooling system according to claim 5 or 6, characterized in that, the neuromorphic engine operates on an event-driven basis, processing data asynchronously to reduce power consumption and enhance response speed and preferably implementing a Leaky Integrate-and-Fire (LIF) model to process incoming signals by generating spike outputs when specific thresholds are reached.RECTIFIED SHEET (RULE 91) ISA / EP8. Tooling system according to at least one of the preceding claims, characterized in that, the control unit (13) directly interfaces with piezoelectric actuators, providing real-time control signals that modulate the actuators' response to detected vibrational anomalies, with actuators positioned in a symmetrical configuration around the tool holder for uniform control in all three spatial dimensions (X, Y, Z).

9. Tooling system according to claim 8, characterized in that, the piezoelectric actuators are integrated with feedback sensors that continuously monitor the position of the tool holder, allowing the control unit (13) to fine-tune actuator responses in real-time.

10. Tooling system according to at least one of the preceding claims, characterized in that, the system includes a divisive normalization device within the control unit (13), comprising:- an input module receiving an input spike train,- a threshold calculation module yielding a threshold parameter according to the average values of the input spike train in an averaging window, and- a normalization module that enables an integrate-and-fire (IAF) counter to produce a single spike at the output and reset the counter when the IAF counter reaches the threshold parameter.RECTIFIED SHEET (RULE 91) ISA / EP11 . Tooling system according to claim 10, characterized in that, the threshold calculation module dynamically adjusts the threshold parameter based on the input spike train's variability within the averaging window, allowing the normalization module to adapt to different signal conditions.

12. Tooling system according to claim 10 or 11 , characterized in that, the divisive normalization device is integrated within the neuromorphic processor of the control unit (13) to enhance real-time signal processing in applications such as feature extraction and classification.

13. Tooling system according to claim 10 to 12, characterized in that, the control unit (13) includes a normalized vibration feature extractor comprising an audio front end and the divisive normalization device, where the vibration front end processes an original vibration signal collected and yields a prenormalized spike train for each channel, with the divisive normalization device processing the pre-normalized spike train to yield a post-normalized spike train.RECTIFIED SHEET (RULE 91) ISA / EP14. Tooling system according to claim 12 or 13, characterized in that, the control unit (13) includes a classifier that executes a classification task based on the output spikes of the normalized audio feature extractor, utilizing a spiking neural network (SNN) architecture for efficient and accurate classification.

15. Tooling system according to at least one of the preceding claims, characterized in that, the control unit (13) is optimized for low-power operation and is designed for integration into real-time vibration processing systems, such as speech recognition devices, hearing aids, or other neuromorphic computing applications where low-latency and high accuracy are critical.RECTIFIED SHEET (RULE 91) ISA / EP

Images