Field of the Invention

The present invention relates to a detecting device and method used for ensuring secure lock of quick-release type electrical cable connectors in various industries such as automotive, white goods, aerospace, etc. designed to minimize the human error in various factories for the secure lock of cable connectors.

Background of the Invention

Most modem vehicles are equipped with electrical and electronics systems, and wiring harness assemblies, which are set of organized wires, connectors, and terminals, are used for relaying power and information. Connectors are essential elements of wiring harness assemblies, and their major function is connecting wires or other electronics in a manner to establish intended circuits and functionality. The connectors used in the automotive industry require a high level of reliability and robustness. As a reliability measure against relative movements of the connected bodies and vibration exposed, the connectors often have secure locking systems built into the plug and socket. These systems are designed to prevent unwanted disconnects and to avoid the risk of loss of contact between wires and transmitting surfaces. Stainless steel latch, bayonet coupling systems, reverse bayonet, push-pull systems are some of the examples of the types that produce audible click and/or tactile feel confirming the intended mating.

Although robots and automation can be used for some connector- assembly tasks, many connectors are still assembled by hand. Assembly instructions that foresee confirmation of intended mating by operators based on their check using the audible click and/or the tactile feel are very common. Other techniques are not preferred because of their challenges or drawbacks. Another technique of confirming proper assembly can be pulling the cables in the opposite direction; however, this method involves considerable damage risks for the harness. Employing in-line techniques is generally not feasible as the assembly order of the harness has to follow other constraints, such as the assembly order of the other parts, rather than already assembled parts configurations required for functional testing. A Complicated, crowded, and closed pack environment obstructs the optical control methods.

Although the method based on the audible click to be recognized by the operator and/or vibration to be felt like a tactile feel is common, it is open to operator faults. The environment is generally noisy and sometimes the noise level necessitates the operators to wear ear protectors. The operator may have to wear thick gloves to protect their finger skins. Furthermore, operators generally assemble many connectors during a shift, which may also vary by type.

It is normal for operators to be fatigued and their perceptions to weaken after many assemblies. As the type, size, location, access, and part stiffness of the connectors vary, the level and nature of the sound and tactile feel also vary in a wide range. Operators can also totally miss to handle any of the connectors that they are supposed to assemble. Sometimes assembly order of different connectors can be important but may not be followed properly by the operator. In short, the operator may miss the indicators of correct assembly and damage the connection during the post-check or may fall into misperception of indicators, or, as another failure mode, matching the two sides (parts) of a connector can be missed totally, or realized in the wrong order.

Once the proper mating is missed in the relevant station, depending on how much does it close to proper assembled position, it causes expensive test and repair operations either in the production plant or at the field, where the customer will experience the failure, first, after a certain mileage that is enough to lose the intended contact. Therefore, there is a need for a device to reduce considerably or eliminate the human error in the common method of confirming proper mating of the connector male and female pairs to achieve an improved level of reliability and robustness. A feasible solution is based on an audible click and/or produced vibration. Such a device can also support traceability, which is infeasible when left to humans only, i.e. without any supporting equipment, and support proper execution of the assembly order.

EP3360206 discloses that a connector mating assurance system includes a user- worn sensor unit configured to be worn on or near a user's hand. The sensor unit has an audible sensor configured to be in a mating zone for electrical connectors. The audible sensor is configured to detect an audible sound when the electrical connectors are mated. However, it has become evident that an audible sensor is very sensitive and susceptible to ambient noise. Considering the extremely noisy environment of the automotive industry, the audible sensor may detect false positives and provide fallacious feedback to the user. Furthermore, an audible sensor-based mating detection system may provide erroneous results in case of mating connectors that are lined up next to each other in the same order if the order of these connectors is changed mistakenly while mating, the audible sensor will still report the successful mating status of the connectors.

US2018263314A1 discloses a device and a method for monitoring the assembly of two components to be connected through a clip fastening using a combination of an audible and a force sensor. The detection sensors are arranged on a user worn glove and configured to detect the audible sound and the force while performing assembly of the connectors. The device reports a successful assembly of the two components when either of the sound or force signal crosses a predefined threshold value. However, in the automotive industry, there are numerous types of connectors, having signatory audible and exerted force response while mating. Therefore, using a threshold-based detection method may result in false detection. Furthermore, the audible and a force sensor may not provide enough information to detect the incorrect mating of the connectors that are lined up next to each other in the same order.

US2016249147 discloses a mating assurance system that includes two microphones that are in the vicinity of a mating zone for electrical connectors. The microphones are configured to detect the audible sound when the electrical connectors are mated. The audible sound signals from the first microphone are compared with the second microphone to attenuate the effect of ambient noise and to detect the successful mating of the connectors. However, the microphones are very sensitive and are prone to ambient noises significantly. Therefore, interferences may occur for example and the detection of the mating status of the connectors may become erroneous. Additionally, a microphone-based mating detection system may provide false results in case of incorrect mating of the connectors that are lined up next to each other in the same order.

DE10308403B4 discloses a monitoring system for the manual fixing of electrical connectors used in automobiles, comprises a user-worn glove with a built-in acoustic sensor that responds to sound generated when manual force is applied. The monitoring system reports a successful connection between the electrical connectors when the recorded sound signal from the acoustic sensor crosses a predefined threshold. The detection of the successful mating of the electrical connectors using the acoustic sensor in the extremely noisy environment of the industry may end up fallacious. Furthermore, an acoustic sensor-based system may not provide decisive information about the orientation, the fitting order, and the type of electrical connector.

Summary of the Invention An object of the invention is to provide a device and method to reduce potential human error in confirming proper mating of male and female parts of connector assemblies of wiring harness systems.

Another object of the present invention is to provide a feedback function on the detecting device that, produces relevant feedback, which, enables easier recognition of proper mating of both sides of the connector assembly by the operator, even in challenging environments.

Another object of the present invention is to equip the detecting device and method such that it can understand the connector identity or type.

Another object of the present invention is to ornament the detecting device and method to capture altered order of assembly and provide predefined feedback.

Another object of the present invention is to realize the detecting device and method such that it still recognizes the failures for different types of connectors even when the signature of the proper mating is altered due to different assembly styles of different operators.

Another object of the present invention is to improve the robustness of the detection operation against several noise factors originate from different operators and their assembling styles, differences in ambient noise level and characteristics, differences in the location of the assembly region and operator position.

Detailed Description of the Invention

A method realized to fulfill the objective of the present invention is illustrated in the accompanying figures, in which: Figure 1 is the perspective view of the detecting device

Figure 2 is the perspective exploded view of the body

Figure 3 is the perspective exploded view of sensors

Figure 4 is the perspective view of another embodiment of the detecting device

Figure 5 is the perspective exploded view of another embodiment of the body

The parts illustrated in the figures are individually numbered where the numbers refer to the following:

1. Detecting device

2. Body

21. Wrist linkage

22. Finger linkage

3. Accelerometer

4. Gyroscope

5. Acoustic sensor

6. Feedback module

7. Power source

8. Controller

9. Central computation unit

10. Communication Unit

A detecting device (1) comprises

- at least a gyroscope (4), which is placed near the finger to detect the orientation of the hand,

- at least an acoustic sensor (5), which is placed near the finger to detect the sound of cable connection,

- at least a feedback module (6) to provide visual or haptic feedback to the user in response to the improper assembly operation,

- at least a controller (8) configured for the gyroscope (4) and acoustic sensor (5), Detecting device (1) comprises a hybrid sensing module which is comprising an acoustic sensor (5) and a tri-axial gyroscope (4). A controller (8) is configured for the acoustic sensor (5) and the tri-axial gyroscope (4), and evaluate the input signals from the acoustic sensor (5) and the triaxial gyroscope (4) to determine and/or calculate the orientation signature and the click sound signature. In another embodiment of the invention, these signatures are evaluated either by edge computing or by a central computation unit (9) to decide the operator id and/or operator style and/or orientation of the connector at the instant of assembly, and/or type of the connector, and with the help of all those, to rate the propemess of the assembly and determine the correctness of assembly order.

In other embodiments of the invention, the detecting device (1) comprises a body (2), which is positioned on the hand. Detecting device (1) comprises a hybrid sensing module which is comprising an acoustic sensor (5), a tri-axial accelerometer (3), and a tri-axial gyroscope (4). The accelerometer (3), which is placed near the finger to detect the vibrational movement of the hand. The gyroscope (4), which is placed near the finger to detect the orientation of the hand. The acoustic sensor (5), which is placed near the finger to detect the sound of the cable connection. The feedback module (6), which is placed on the body (2) to provide visual or haptic feedback to the user. The power source (7), which is placed on the body (2). The controller (8), which is placed on the body (2) and configured for the accelerometer (3), gyroscope (4), and acoustic sensor (5). The controller (8) is configured for the accelerometer (3), the acoustic sensor (5), and the tri-axial gyroscope (4) inputs, and evaluation of the input signals from the accelerometer (3), the acoustic sensor (5) and the triaxial gyroscope (4) to determine and/or calculate the orientation signature, acceleration signature, and the click sound signature. These signatures are then evaluated either by the controller (8) or by the central computation unit (9) to decide the operator id and/or operator style and/or orientation of the connector at the instant of assembly, and/or type of the connector, and with the help of all those, to rate the propemess of the assembly and determine the correctness of assembly order. Interfaced of the controller (8) interpolates the output signals from the acoustic sensor (5), the tri-axial accelerometer (3), and the tri-axial gyroscope (4) and returns feedback to the user in terms of a visual and a physical stimulus. The working principle behind the proposed solution depends on the hybrid sensing module that encapsulates three different types of sensors (acoustic sensor (5), accelerometer (3), and gyroscope (4)).

The accelerometer (3) is placed near the finger. The tri-axial accelerometer (3) monitors the vibrational movement of the hand in the X, Y, and Z-axis during the locking mechanism. The accelerometer (3) is mounted in such a way that the y-axis of the accelerometer (3) is parallel to the movement of the human hand whereas the x-axis and z-axis are in the perpendicular direction to the movement of the hand. The advantage of recording the data from all three axes allows for the better detection of lock protrusion snapping in contact with the locking arm. It also helps to identify and remove erroneous data that may be produced due to the no motion values produced due to the collision of the connectors at a wrong angle or the half immersion of the cables without the securing of the locks. The motion of accelerometer (3) is measured in x, y, z Cartesian coordinates for the detection of the click- lock in quick-release type cable connectors. The triaxial accelerometer (3) is a MEMS-based capacitive sensing module that responds to the gravitational force on the body in motion or at rest.

The gyroscope (4) is placed near the finger. The tri-axial gyroscope (4) monitors the orientation of the hand holding connector while locking in terms of roll, yaw, and pitch rotation angles. The orientation of the connector while locking plays a significant role in the successful locking. The misorientation of the male connector to the female connector would result in an unsuccessful cable connection. Furthermore, the gyroscope data also helps in eradicating the fallacious connection between the connectors that are lined up next to each other. The acoustic sensor (5) is placed near the finger. The acoustic sensor (5) records the sound wave that originates because of the locking of the connector. There are different types of electrical cable connectors and each type of connector generates a specific type of sound waves. The acoustic sensor (5) is mounted on the finger which brings it in proximity with the cable connectors and allows the shielding of the device from ambient noise in the work environment. The acoustic sensor (5) hears the clicking sound and generates a response in terms of voltages. The acoustic sensor (5) is based on the diaphragm which vibrates as a result of the incoming sound waves of different frequencies and amplitudes. These differences in frequencies and amplitudes allow the audible sensor to differentiate between sounds. The snapping of the locking mechanism in quick-lock cable connectors produces a distinct sound that has physical properties different from other audible signals. The acoustic sensor (5) achieved through a change in the voltage output of the auditory sensor, the auditory sensor may be comprised of the diaphragm with magnets, a piezoelectric crystal, and/or piezo-resistive element.

The feedback module (6), positioned on the body (2), provides visual and haptic feedback to the user in response to the improper assembly operation. The visual feedback is comprised of a display screen that is placed on the body (2) that exhibits the correct orientation of the connector, the locking status of the connector, and the connector type. In another embodiment, the visual feedback may also contain status LEDs, is positioned on the body (2), which indicates the locking status with multiple colors. The visual feedback to the user about the status of the cable connection in the form of a positive response or a negative response based on the detection algorithm. The haptic feedback is comprised of a vibrating actuator which is positioned on the body (2), embedded in the detecting device (1). The haptic feedback notifies the user about individual connector types by varying vibration patterns. The haptic feedback informs the user not only about the security of the lock but also the cable connector type. In another embodiment of the invention, the feedback module (6) is positioned near to the working space, provides visual feedback to the user in response to the improper assembly operation. When the controller (8) detects the user’s improper assembly operation, the controller (8) sends these data to the feedback module (6) and the user is warned.

In another embodiment of the invention, the detecting device (1) comprising at least one communication unit (10) to communicate with the controller (8). The communication unit (10) is a wireless communication transceiver. Thanks to the communication unit (10), collecting data is sent to a central computation unit (9) by the controller (8). In another embodiment of the invention, the feedback module (6) is positioned near to the working space, provides visual feedback to the user in response to the improper assembly operation. When the controller (8) detects the user’s improper assembly operation, the controller (8) sends these data to the feedback module (6) by using the communication unit (10), and the user is warned.

In other embodiments of the invention, the body (2) is positioned on the wrist of the user, because it can be carried easily, and the effect of the weight of the body (2) does not feel from the user. In another embodiment of the invention, there is a wearable glove, and body (2) is placed on the wearable glove. Another embodiment of the invention, the body (2) is placed on a wrist linkage (21) which has a wrist fitting ring shape. The power source (7) is placed on the body (2), and the controller (8), which is placed on the body (2) and configured for the accelerometer (3), gyroscope (4), and acoustic sensor (5).

Another embodiment of the invention is that at least a finger linkage (22) which has a ring-shaped on the finger. Acoustic sensor (5) is mounted on the finger linkage (22). The finger linkage (22) can be customized according to the range of finger size. The accelerometer (3), gyroscope (4), and acoustic sensor (5) are placed on the finger linkage (22), close to the connectors during the locking mechanism.

The detecting device (1) is achieved through a change in vibration response such that: the change in vibrational response may be based on the change in resistance of the accelerometer (3), change in capacitance of the accelerometer (3). The change in uniaxial direction, coaxial direction, and triaxial coordinates may be used to detect the vibrations produced in the hand during click-lock. The change in the directions of the coordinates will also identify the change in the gyroscopic (4) values of the system, therefore, indicating the direction and orientation of the cable being held in the hand while matting connectors

The controller (8), performs signal analysis on the incoming sensor data to evaluate the correct lock connection of the cables based on the threshold values set. The controller (8) provides feedback about the connector locking status (proper or not proper). The controller (8) provides feedback about the connection type (different types of cables) by using machine learning algorithms. The controller (8) will be able to identify different cables with different levels of acoustic and vibrational values and set custom thresholds for each cable. With each iteration recorded from the controller (8), the proposed controller (8) algorithm will minimize the error percentage of the system and provide corrective feedback to the user. The controller (8) taking input from sensors which will store the audio, accelerometer (3), and gyroscope (4) data for each cable. A machine learning methodology will be implemented in the system to learn the correct cable lock values from each cable system.

In another embodiment of the invention, the controller (8) taking input from sensors and the controller (8) send these input data to the central computation unit (9) by using the communication unit (10). The central computation unit (9) stores the audio, accelerometer (3), and gyroscope (4) data for each cable. The central computation unit (9), performs signal analysis on the incoming sensor data to evaluate the correct lock connection of the cables based on the threshold values set. The central computation unit (9) provides feedback about the connector locking status (proper or not proper). The central computation unit (9) provides feedback about the connection type (different types of cables) by using machine learning algorithms. The central computation unit (9) will be able to identify different cables with different levels of acoustic and vibrational values and set custom thresholds for each cable. With each iteration recorded from the central computation unit (9), the proposed central computation unit (9) algorithm will minimize the error percentage of the system. The central computation unit (9) sends the signal to the feedback module (6). In other words, when the central computation unit (9) detects the user’s improper assembly operation, the central computation unit (9) sends these data to the feedback module (6) and the user is warned.

All the sensors (3, 4, 5) send data to the controller (8) that preprocesses the data. Motion artifacts and ambient noises are filtered out in the preprocessing step. Different types of attributes (statistical, time, and frequency domain) are extracted from the data and are fed to the machine learning algorithm for training purpose. After significant training, a new set of data is acquired for testing purposes, and the same set of attributes are extracted and passed as input to the machine learning algorithm. The incoming attributes are compared with the trained model and a decision is generated. This decision is transferred to the feedback module (6). The feedback module (6) visually displays the locking status, the connector type, and the orientation of the hand. The feedback module (6) may also contain an array of LEDs that gives a visual indication to the user about the locking and orientation status of the connector.

The detecting device (1), during one of the connector locking, sounds of the connector is detected with an acoustic sensor (5) and this data is sent to the controller (8). There are different types of cable connectors and each type of connector generates a specific type of sound waves and these specific types of sound data are sent to the controller (8) by the acoustic sensor (5). At the same time, the gyroscope (4) detects the orientation of the hand during one of the connectors locking, and this data is sent to the controller (8). There are different types of electrical cable connectors and hand moves different orientation each type of connector locking, these specific types of hand orientation data are sent to the controller (8) by the gyroscope (4). Thanks to acoustic sensor (5) data and gyroscope (4) data, a detecting device (1) provides a user to change the order of connectors fitting line freely and improve reliability. Depending on the configuration of the system, whether edge computing or computation at a central unit takes place, the results of the evaluation are transferred to the feedback module (6) via information signals from the central computation unit (9) or controller (8).

In another embodiment of the invention, the controller (8) taking input from sensors and the controller (8) send these input data to the central computation unit (9) by using the communication unit (10). The central computation unit (9) stores the audio, accelerometer (3), and gyroscope (4) data for each cable. The central computation unit (9), performs signal analysis on the incoming sensor data to evaluate the correct lock connection of the cables based on the threshold values set. The central computation unit (9) provides feedback about the connector locking status (proper or not proper). The central computation unit (9) provides feedback about the connection type (different types of cables) by using machine learning algorithms. The central computation unit (9) will be able to identify different cables with different levels of acoustic and vibrational values and set custom thresholds for each cable. With each iteration recorded from the central computation unit (9), the proposed central computation unit (9) algorithm will minimize the error percentage of the system. The central computation unit (9) sends the signal to the feedback module (6). In other words, when the central computation unit (9) detects the user’s improper assembly operation, the central computation unit (9) sends these data to the feedback module (6) and the user is warned.

A detecting method for detecting device (1) for connector locking status comprising steps of; a training mode;

- during the connector locking, the controller (8) receives single-channel data from the acoustic sensor (5), three-channel data from three-channel data from the gyroscope (4) to detect the orientation of the hand which is the orientation signature,

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave, - a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted by using a windowing scheme from the controller (8), extracted attributes are then used to train the machine learning model, which is used to decide whether the assembly process is proper or not a testing mode

- during the connector locking, the controller (8) receives and recorded single-channel data from the acoustic sensor (5) and three-channel data from the gyroscope (4),

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the controller (8),

- the extracted set of attributes are evaluated by the trained model then the decision is made whether the operation is successful or not.

In another embodiment of the invention, a detecting method for detecting device (1) for connector locking status comprising steps of; a training mode;

- during the connector locking, the controller (8) receives single-channel data from the acoustic sensor (5), three-channel data from the accelerometer (3), and three-channel data from the gyroscope (4) to detect the orientation of the hand which is the orientation signature,

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a bandpass filter is applied to the recorded data of an accelerometer (3),

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts, - after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the controller (8), extracted attributes are then used to train the machine learning model, which is used to decide whether the assembly process is proper or not a testing mode;

- during the connector locking, the controller (8) receives and recorded single-channel data from the acoustic sensor (5), three-channel data from the accelerometer (3), and three-channel data from the gyroscope (4),

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a bandpass filter is applied to the recorded data of an accelerometer (3),

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the controller (8), the extracted set of attributes are evaluated by the trained model then the decision is made whether the operation is successful or not.

The adapted detection criterion in the proposed invention is based on the principles of machine learning. The machine learning -based algorithm operates in two modes, called training mode and the testing mode. In the training mode, the algorithm takes input in the form of a data set with marked class labels (for example data recorded from sensor module for class 1 = successful locking mechanism and class 2 = unsuccessful locking mechanism) and generates a learning curve. Whereas in the testing mode, the algorithm takes blind input in the form of a data set without marked class labels and classifies it into individual classes (to whom they belong) based on the prior learning curve that was developed during training mode.

In this invention, the machine learning-based algorithm is first trained rigorously for a variety of electrical cable connectors while performing the locking mechanism. Initially, the controller (8) receives single-channel data from the acoustic sensor (5), and three-channel data from the gyroscope (4) for a successful locking action of an electrical cable connector. In another embodiment of the invention, the controller (8) receives three-channel data from the accelerometer (3). The controller (8) receives a total of seven channels of time-series data. The acquired data is then passed through a preprocessing step to attenuate noise.

Different types of connectors have different dimensions due to that the output response in terms of locking sound and the hand movement would be different. Moreover, the amplitude and frequency of the locking sound are significantly different from the ambient noises in the industry. Therefore, based on the required frequency range, a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave.

The varying dimensions of different connectors would also affect the output of the accelerometer (3) and the gyroscope (4). There is a significant difference between the acceleration of the cable connector and the random unwanted movements of the hand. Therefore, the data of an accelerometer (3) is also passed through a bandpass filter. However, the orientation angles are not mainly dependent on the individual connector types, therefore a moving average filter is applied to the data of gyroscope (4) to attenuate artifacts.

After preprocessing of the acquired data by using the controller (8), a set of statistical, time-dependent, frequency-dependent, and both time /frequencydependent attributes are extracted from the controller (8). The attributes are extracted by overlapping windows that slide over the length of data in defined increments.

In the training mode, the locking mechanism is performed in a supervised manner in which the extracted attributes are marked with class labels indicating the connector locking status, connector type, and the connector orientation status as individual classes. These datasets are saved in memory. The machine learning algorithm takes in the attributes extracted from the training data and generates a learning curve.

In the testing mode, the locking mechanism is performed in an unsupervised manner in which extracted attributes are not marked with class labels. However, the same types of attributes are extracted as were extracted in training mode. These attributes are fed to the algorithm that was trained during training mode. Each attribute is classified into an individual class to whom it belongs. The decision is made on every incoming data window and corresponding feedback is generated.

In another embodiment of the invention, a detecting method for detecting device (1) for connector locking status comprising steps of; a training mode;

- during the connector locking, the central computation unit (9) receives single-channel data from the acoustic sensor (5), three-channel data from three-channel data from the gyroscope (4) to detect the orientation of the hand which is the orientation signature,

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are racted by using a windowing scheme from the central computation unit extracted attributes are then used to train the machine learning model, which is used to decide whether the assembly process is proper or not a testing mode;

- during the connector locking, the central computation unit (9) receives and recorded single-channel data from the acoustic sensor (5), and three-channel data from the gyroscope (4), - a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the central computation unit (9),

- the extracted set of attributes are evaluated by the trained model then the decision is made whether the operation is successful or not.

In another embodiment of the invention, a detecting method for detecting device (1) for connector locking status comprising steps of; a training mode;

- during the connector locking, the central computational unit (9) receives single-channel data from the acoustic sensor (5), three-channel data from the accelerometer (3), and three-channel data from the gyroscope (4) to detect the orientation of the hand which is the orientation signature,

- a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a bandpass filter is applied to the recorded data of an accelerometer (3),

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the central computational unit (9), extracted attributes are then used to train the machine learning model, which is used to decide whether the assembly process is proper or not a testing mode;

- during the connector locking, the central computational unit (9) receives and recorded single-channel data from the acoustic sensor (5), three-channel data from the accelerometer (3), and three-channel data from the gyroscope (4), - a suitable bandpass filter is applied to the recorded data of the acoustic sensor (5) to attenuate the undesirable frequencies in the sound wave,

- a bandpass filter is applied to the recorded data of an accelerometer (3),

- a moving average filter is applied to the recorded data of gyroscope (4) to attenuate artifacts,

- after preprocessing of the acquired data, a set of statistical, time-dependent, frequency-dependent, and both time /frequency-dependent attributes are extracted from the central computational unit (9), the extracted set of attributes are evaluated by the trained model then the decision is made whether the operation is successful or not.

Initially, the computational unit (9) receives single-channel data from the acoustic sensor (5), and three-channel data from the gyroscope (4) for a successful locking action of an electrical cable connector. In another embodiment of the invention, the computational unit (9) receives three-channel data from the accelerometer (3). The computational unit (9) receives a total of seven channels of time-series data. The acquired data is then passed through a preprocessing step to attenuate noise.

After preprocessing of the acquired data by using computational unit (9), a set of statistical, time-dependent, frequency-dependent, and both time /frequencydependent attributes are extracted from the computational unit (9). The attributes are extracted by overlapping windows that slide over the length of data in defined increments.

The biggest advantage of a machine learning-based algorithm is the ability to train itself on a diverse set of data set that belongs to a varying set of classes. Therefore, with the single algorithm, the locking status, and the orientation status of different types of electrical cable connectors during assembly can be monitored and notified to the user. The gyroscope (4), in conjunction with the acoustic sensor (5) and the tri-axial accelerometer (3), detects the orientation of the hand during the snapping of the quick-release type cable connectors. The tri-axial gyroscope (4) is based on three independent vibratory MEMS rate gyroscopes, which detect rotation about the individual axis. When the gyroscope (4) is rotated about any of the sense axes, the Coriolis effect causes a vibration that is detected by a capacitive pickoff.

The varying dimensions of different connectors would also affect the output of the accelerometer (3) and the gyroscope (4) data. There is a significant difference between the acceleration of the cable connector and the random unwanted movements of the hand. Therefore, the data of an accelerometer (3) is also passed through a bandpass filter. However, the orientation angles are not mainly dependent on the individual connector types, therefore a moving average filter is applied to the data of the gyroscope (4) to attenuate artifacts.

The inclusion of the gyroscope (4) along with the acoustic sensor (5) and the accelerometer (3), not only provides vital information regarding the orientation status of the connector but also ensures the precise measurement of the accelerometer (3) data.

Considering the case of an automobile in an automotive industry, where different types of wire connectors are required to be mated at certain blind locations. Once trained, the machine-learned algorithm notifies the user about the correct orientation of the connector while mating thus minimizes the error.

Every connector type has a specific signatory acceleration data output while mating, which is dominant only in either of the three axes. The orientation data from the gyroscope (4) sensor can be related to the acceleration/acoustic data, which enables the machine learning algorithm to improve the device response over time. The detecting device and method (1) provide to minimize the human error in various factories for the secure lock of cable connectors and also provides accurate detection in multiplex cable configurations. Within the scope of these basic concepts, it is possible to develop a wide variety of embodiments of the inventive wearable detection device (1). The invention cannot be limited to the examples described herein; it is essentially according to the claims.