Dual-mode three-dimensional force sensor billiard hitting system
The dual-mode three-dimensional force sensor system with static capacitance and dynamic piezoelectric films addresses the limitations of existing table tennis training systems by providing precise, real-time feedback on stroke parameters, improving training effectiveness.
Patent Information
- Application Number
- CN202510331064.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-20
- Publication Date
- 2025-07-15
AI Technical Summary
The existing billiards training system has significant limitations in accuracy and real-time feedback, and cannot capture changes in static and dynamic forces at the same time. The data processing and feedback mechanism are insufficient, making it difficult to meet the needs of high-level athletes and enthusiasts.
The dual-mode three-dimensional force sensor system is adopted, combined with a static capacitive dielectric layer and a dynamic piezoelectric film, and real-time analysis is carried out through neural network models, supporting the detection and feedback of a variety of basic rod methods, including linear hitting, rotary hitting, rebound hitting, etc.
It realizes high-precision multi-dimensional batting action detection, provides real-time feedback, significantly improves training efficiency and competitive level, and is suitable for billiards enthusiasts of different levels. The hardware structure is simple and easy to install and maintain.
Smart Images
Figure CN120315577A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of billiards training, and specifically provides a dual-mode three-dimensional force sensor billiards hitting system. Background Art
[0002] Traditional billiards training systems usually rely on visual recognition or simple mechanical sensors to detect hitting actions. Although these methods can provide basic information about hitting to some extent, they have significant limitations in terms of accuracy and real-time feedback. Visual recognition systems are easily affected by environmental light and background interference, resulting in inaccurate detection results; while simple mechanical sensors cannot provide multi-dimensional hitting data, such as the force, angle, and rotation of hitting. In addition, existing systems often lack the ability to provide real-time feedback on hitting actions, and users cannot adjust hitting strategies in a timely manner, resulting in limited training effects. With the popularization of billiards and the improvement of competitive levels, traditional training methods can no longer meet the needs of high-level athletes and enthusiasts.
[0003] In recent years, with the rapid development of sensor technology and artificial intelligence algorithms, intelligent training systems based on multi-dimensional force detection have gradually become a research hotspot. However, most existing multi-dimensional force detection systems adopt a single sensor mode, such as only using capacitive sensors or piezoelectric sensors, and cannot capture the changes of static force and dynamic force simultaneously, resulting in incomplete detection results. In addition, existing systems also have deficiencies in data processing and feedback mechanisms, often relying on complex algorithms and expensive hardware devices, and are difficult to be widely applied in actual training scenarios. Summary of the Invention
[0004] Aiming at the deficiencies of the prior art, the present invention provides a dual-mode three-dimensional force sensor billiards hitting system, which solves the problem that most existing multi-dimensional force detection systems adopt a single sensor mode and cannot capture the changes of static force and dynamic force simultaneously, resulting in incomplete detection results.
[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A dual-mode three-dimensional force sensor billiards hitting system includes the following structures:
[0006] System structure: It has PDMS upper encapsulation and PDMS lower encapsulation; is provided with a PI substrate as the sensor support layer; includes a static capacitive dielectric layer for detecting static force; includes a dynamic piezoelectric thin film for detecting dynamic force; and lower electrodes and upper electrodes for collecting capacitive and piezoelectric signals.
[0007] Neural network structure: The activation function uses ELU, and the loss function selects the Huber loss function; Stochastic gradient descent combined with the cosine annealing strategy is used for training; An early stopping strategy is set during the training process, and the early stopping patience value is 200.
[0008] Hardware structure: It includes a billiard table, a cue, a three-dimensional force sensor installed at the hitting end of the cue, and a data processing unit. The three-dimensional force sensor is used to collect hitting data in real time and transmit it to the data processing unit.
[0009] Preferably, the signals of the static capacitive dielectric layer and the dynamic piezoelectric film are transmitted to the data processing unit through the lower electrode and the upper electrode. The data processing unit analyzes the signals using a trained neural network model to obtain the hitting force, angle, and rotation parameters.
[0010] Preferably, the system supports the detection of multiple basic cue techniques, including straight hitting, spin hitting, and bank shot.
[0011] Preferably, the data processing unit analyzes the hitting action in real time, provides feedback information, and gives feedback suggestions on the hitting action.
[0012] Preferably, the PI substrate has flexibility and mechanical strength, and can support the sensor and adapt to the mechanical changes during hitting.
[0013] Preferably, the upper PDMS encapsulation and the lower PDMS encapsulation are used to protect the internal sensors and circuits, and buffer and protect the impact force through their own materials and structures.
[0014] Preferably, the early stopping strategy with an early stopping patience value of 200 is adopted during the training process of the neural network structure.
[0015] Preferably, the system combines a static capacitive dielectric layer and a dynamic piezoelectric film to detect and feedback on the hitting action.
[0016] The present invention provides a dual-mode three-dimensional force sensor billiard hitting system, which has the following beneficial effects:
[0017] 1. High-precision detection: By combining a static capacitive dielectric layer and a dynamic piezoelectric film, the system can simultaneously detect the static pressure and dynamic impact force during hitting, achieve multi-dimensional precise detection of the hitting action, and ensure the comprehensiveness and accuracy of the data.
[0018] 2. Real-time feedback: The system can analyze the hitting action in real time and provide instant feedback through a display screen or voice prompt, helping users quickly adjust their hitting strategies and significantly improving the training efficiency.
[0019] 3. Multifunctional support: The system supports the detection and feedback of five basic cue techniques, including straight hitting, spin hitting, bank shot, etc., is suitable for billiard enthusiasts at different levels, and meets diverse training needs.
[0020] 4. Stability and Generalization Ability: The neural network model adopts the ELU activation function, Huber loss function, and cosine annealing strategy, combined with the early stopping mechanism (the patience value is set to 200), which can effectively prevent overfitting and ensure the stability and generalization ability of the model.
[0021] 5. Easy Installation and Maintenance: The hardware structure of the system is simple, the sensors are easy to install, suitable for various billiard training scenarios, and the maintenance cost is low, which is convenient for popularization and use.
[0022] 6. Improve Training Effect: Through accurate analysis of hitting data and real-time feedback, the system can help users quickly master hitting skills and significantly improve training effects and competitive levels. Brief Description of the Drawings
[0023] Figure 1 It is a module diagram of the dual-mode three-dimensional force sensor billiard hitting system of the present invention;
[0024] Figure 2 It is a structural model diagram of the three-dimensional force sensor of the present invention;
[0025] Figure 3 It is a specific neural network diagram of the billiard hitting system of the present invention;
[0026] Figure 4 It is a loss function and goodness-of-fit diagram regarding the training results of the present invention;
[0027] Figure 5 It is a hardware structure diagram of the billiard training system of the present invention;
[0028] Figure 6 It is a schematic diagram of five basic strokes of the billiard of the present invention;
[0029] Figure 7 (a) It is an interface diagram a showing the force change when hitting with different orientations and forces of the present invention;
[0030] Figure 7 (b) It is an interface diagram b showing the force change when hitting with different orientations and forces of the present invention;
[0031] Figure 7 (c) It is an interface diagram c showing the force change when hitting with different orientations and forces of the present invention. Detailed Embodiment
[0032] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the specification of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0033] Embodiment:
[0034] Please refer to the attached Figure 1 - attached Figure 7 , the embodiment of the present invention provides a dual - mode three - dimensional force sensor billiard hitting system, including the following structures:
[0035] Structure 1. The system structure includes upper PDMS encapsulation and lower PDMS encapsulation, which are used to protect the internal sensors and circuits; the PI substrate serves as the support layer of the sensor, having good flexibility and mechanical strength; the static capacitance dielectric layer is used to detect static forces, such as the pressure distribution when hitting the ball; the dynamic piezoelectric film is used to detect dynamic forces, such as the instantaneous impact force when hitting the ball; the lower electrode and the upper electrode are used to collect capacitance and piezoelectric signals, so as to realize multi - dimensional detection and analysis of the hitting action.
[0036] Structure 2. Neural network structure: In the neural network structure, the activation function adopts ELU (Exponential Linear Unit) to improve the non - linear fitting ability of the model, and the loss function selects the Huber loss function to balance the mean square error and the absolute error. To optimize the training process, Stochastic Gradient Descent (SGD) is combined with the Cosine Annealing strategy (CosineAnnealingLR), so as to improve the training efficiency and the model accuracy. In addition, the early stopping patience value is set to 200 to prevent the model from overfitting and ensure the stability and generalization ability of the training.
[0037] Structure 3. Hardware structure: The system includes a billiard table, a cue, a three - dimensional force sensor, and a data processing unit. The sensor is installed at the hitting end of the cue, and it collects hitting data in real time and transmits it to the data processing unit for analysis and feedback;
[0038] The signals of the static capacitive dielectric layer and the dynamic piezoelectric film are transmitted to the data processing unit through the lower electrode and the upper electrode. The data processing unit analyzes the signals using a trained neural network model and calculates parameters such as the force, angle, and spin of the hit. The system supports the detection and feedback of five basic shot techniques, including straight shots, spin shots, and bank shots. The data processing unit can analyze the hitting action in real time and provide feedback to help the user adjust the hitting action. The PI substrate has good flexibility and mechanical strength, which can effectively support the sensor and adapt to the mechanical changes during hitting. The PDMS upper encapsulation and PDMS lower encapsulation can effectively protect the internal sensors and circuits, preventing interference and damage from the external environment to the system. The training process of the neural network structure adopts an early stopping strategy, and the early stopping patience value is set to 200 to prevent overfitting of the model. By combining the static capacitive dielectric layer and the dynamic piezoelectric film, the system can achieve high-precision detection and real-time feedback of the hitting action. The system is suitable for billiards enthusiasts at different levels, can effectively improve the training effect, and the hardware structure of the system is simple, easy to install and maintain, and is suitable for various billiards training scenarios.
[0039] The usage steps of a dual-mode three-dimensional force sensor billiards hitting system include the following steps:
[0040] Step 1,
[0041] Installation and configuration of the three-dimensional force sensor: First, install the three-dimensional force sensor at the hitting end of the cue. The outside of the sensor is protected by PDMS (polydimethylsiloxane) upper encapsulation and PDMS lower encapsulation. The PDMS material has excellent flexibility, wear resistance, and waterproofness, which can effectively protect the internal sensors and circuits from interference and damage from the external environment. The core structure inside the sensor includes a PI (polyimide) substrate. The PI material has high mechanical strength, good flexibility, and high-temperature resistance, which can provide stable support for the sensor and adapt to the mechanical changes generated during hitting, ensuring that the sensor can still maintain high performance after multiple hits.
[0042] The core functional layer of the sensor consists of a static capacitive dielectric layer and a dynamic piezoelectric film. The static capacitive dielectric layer is used to detect the static pressure distribution during hitting, and can accurately measure the magnitude and distribution of the pressure when the cue contacts the billiard ball, thereby helping the user understand the uniformity and force control of the hit. The dynamic piezoelectric film is used to detect the dynamic impact force and vibration during hitting, and can convert the mechanical energy (such as instantaneous impact force and vibration) generated at the moment of hitting into electrical signals, thereby capturing the dynamic characteristics of the hitting action, such as the speed, acceleration of the hit, and the vibration frequency of the cue. The combination of these two functional layers enables the sensor to simultaneously detect static and dynamic mechanical changes and achieve multi-dimensional accurate detection of the hitting action.
[0043] When installing the sensor, ensure that its contact surface with the cue stick fits tightly to avoid errors in signal acquisition. The lower electrode and upper electrode of the sensor are connected to the data processing unit through highly conductive wires to ensure the stability and real-time performance of signal transmission. The connection part of the wires is shielded to prevent the influence of external electromagnetic interference on signal transmission. In addition, the installation position of the sensor is optimized to ensure that it does not affect the normal operation of the user during the hitting process, while being able to capture the mechanical characteristics of the hitting action to the greatest extent. Through this installation method, the system can collect high-quality static capacitance signals and dynamic piezoelectric signals in real time when the user hits the ball, providing a reliable basis for subsequent data processing and analysis.
[0044] Step Two:
[0045] Data Acquisition and Preprocessing: When the user hits the ball, the three-dimensional force sensor installed at the hitting end of the cue stick will collect the static capacitance signals and dynamic piezoelectric signals generated by the hitting action in real time. The static capacitance signals are generated by the static capacitance dielectric layer in the sensor and reflect the pressure distribution during hitting. By detecting the pressure change when the cue stick contacts the billiard ball, the static capacitance signals can accurately measure the magnitude of the hitting force and its distribution on the cue stick, thus helping the user understand the uniformity and force control of the hit. The dynamic piezoelectric signals are generated by the dynamic piezoelectric film in the sensor and reflect the impact force and vibration at the moment of hitting. The piezoelectric film can convert the mechanical energy (such as instantaneous impact force and vibration) generated during hitting into electrical signals, thereby capturing the dynamic characteristics of the hitting action, such as the speed, acceleration of the hit, and the vibration frequency of the cue stick. The combination of these two signals enables the system to comprehensively and multi-dimensionally detect the hitting action, including both the static force distribution and the dynamic instantaneous impact characteristics.
[0046] The collected static capacitance signals and dynamic piezoelectric signals are transmitted to the data processing unit through the lower electrode and upper electrode of the sensor. During the signal transmission process, the system ensures the stability and real-time performance of the signals through optimized circuit design, avoiding signal attenuation or interference. After receiving the signals, the data processing unit first performs preprocessing on them. The preprocessing steps include filtering, amplification, and digitization. The filtering operation is mainly used to remove high-frequency noise and low-frequency interference in the signals to ensure the purity of the signals; the amplification operation is used to enhance the intensity of the signals so that they can be effectively processed by the subsequent neural network model; the digitization operation converts the analog signals into digital signals for efficient processing and analysis by the computer. Through these preprocessing steps, the system can significantly improve the accuracy of the signals, providing a reliable data basis for subsequent hitting action analysis. The preprocessed signals will be input into the neural network model for further analysis to achieve accurate detection and real-time feedback of the hitting action.
[0047] Step Three:
[0048] Neural Network Model Training and Application: The preprocessed signal is input into the trained neural network model for further analysis. The neural network uses ELU (Exponential Linear Unit) as the activation function. The function formula is established as follows:
[0049]
[0050] When x is greater than or equal to 0, the function value is equal to x; when x is less than 0, the function value is α(e x -1), where α is a hyperparameter, usually taking a relatively small value, such as 0.2, etc. This function can effectively alleviate the problem of gradient disappearance and improve the non-linear fitting ability of the model.
[0051] ELU can effectively alleviate the problem of gradient disappearance and improve the non-linear fitting ability of the model, so as to better capture the complex mechanical characteristics in the hitting action. The loss function uses the Huber loss function. The Huber loss function combines the advantages of the mean square error (MSE) and the mean absolute error (MAE), and can maintain robustness when dealing with outliers, while ensuring the fitting accuracy of the model for most data. The algorithm is as follows:
[0052]
[0053] Here y is the true value, f(x) is the model prediction value, and δ is a hyperparameter.
[0054] The optimizer uses Stochastic Gradient Descent (SGD) combined with the Cosine Annealing strategy (CosineAnnealingLR). SGD can efficiently update the model parameters, while the cosine annealing strategy avoids the model falling into local optima by dynamically adjusting the learning rate, thereby improving the training efficiency and model accuracy. The algorithm used here is as follows:
[0055]
[0056] η t is the learning rate at the current iteration t, η min is the minimum learning rate, η max is the maximum learning rate, T cur is the current number of training steps, T max is the total number of training steps;
[0057] The cosine annealing strategy dynamically adjusts the learning rate. As the number of training steps increases, the learning rate gradually decreases from η max to η min , avoiding the model falling into local optima, thereby improving the training efficiency and the model.
[0058] The neural network model analyzes the input signal and calculates multiple key parameters for hitting the ball, including hitting force, hitting angle, hitting spin, etc. The hitting force reflects the magnitude of the force when the user hits the ball, the hitting angle describes the angle when the cue contacts the billiard ball, and the hitting spin captures the spinning state of the billiard ball after being hit. These parameters can comprehensively reflect the user's hitting action and provide data support for subsequent feedback. Through the high-precision analysis of the neural network, the system can generate detailed hitting data in real time, helping the user understand the advantages and disadvantages of the hitting action and providing targeted improvement suggestions, thus significantly enhancing the training effect and competitive level.
[0059] Step Four
[0060] Real-time feedback and user adjustment: The data processing unit feeds back the analysis results to the user in real time through the display screen or voice prompt. The display screen can intuitively show key parameters such as the hitting force, angle, and spin of the hit, and at the same time provide a graphical hitting trajectory and mechanical analysis to help the user comprehensively understand the details of the hitting action. The voice prompt guides the user to adjust the hitting action through concise language, such as prompting the user to increase the hitting force, adjust the hitting angle, or reduce the spin, etc. The user can adjust the hitting strategy in real time according to this feedback information, thereby improving the accuracy and effect of hitting and gradually optimizing the hitting skills.
[0061] The system also supports the detection and feedback of five basic cueing techniques, including straight hitting, spin hitting, bank shot, etc. By analyzing the hitting parameters, the system can automatically identify the cueing technique used by the user and provide targeted feedback according to different cueing techniques. For example, for spin hitting, the system will prompt the user to adjust the hitting point to control the spin effect; for bank shot, the system will suggest that the user optimize the hitting angle to achieve the expected bank trajectory. This intelligent feedback mechanism not only helps the user master the skills of different cueing techniques, but also provides personalized training suggestions according to the user's hitting habits, thus significantly enhancing the training efficiency and competitive level.
[0062] Step Five
[0063] System optimization and maintenance: By continuously optimizing the neural network model and sensor configuration, the system can adapt to the hitting habits and training needs of different users. The optimization of the neural network model includes adjusting the network structure, improving the activation function and loss function, and introducing more efficient optimization algorithms, so as to improve the prediction accuracy and generalization ability of the model. At the same time, the sensor configuration has also gone through multiple iterations to ensure that it can accurately capture the static pressure and dynamic impact force of the hitting action, and adapt to the hitting styles and force changes of different users. In addition, the hardware structure of the system is designed simply, the sensors are easy to install, without complex debugging processes, and the maintenance cost is low, which is suitable for various billiards training scenarios such as homes, clubs, and professional training venues.
[0064] To ensure the long-term stable operation of the system, it is necessary to regularly inspect and calibrate the sensors. The calibration process includes performance tests on the static capacitive dielectric layer and the dynamic piezoelectric film of the sensors to ensure the accuracy and consistency of signal acquisition. The software part of the data processing unit supports online updates, and users can download the latest software version through the network to introduce new functions and optimize algorithms. For example, the update may include a more efficient neural network model, a new shot method recognition function, or a more intuitive user interface design. This flexible upgrade mechanism enables the system to always maintain technological leadership, meet the changing training needs of users, and extend the service life of the system.
[0065] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A dual-mode three-dimensional force sensor billiard hitting system, characterized in that, It includes the following structures: System structure: It has encapsulation above PDMS and encapsulation below PDMS; There is a PI substrate as the sensor support layer; it includes a static capacitive dielectric layer for detecting static force; it includes a dynamic piezoelectric thin film for detecting dynamic force; and lower and upper electrodes for collecting capacitive and piezoelectric signals; Neural network structure: The activation function is ELU, the loss function is the Huber loss function, the training method uses stochastic gradient descent combined with the cosine annealing strategy, and an early stopping strategy is set during the training process, with an early stopping patience value of 200; Hardware structure: It includes a billiard table, a cue, a three-dimensional force sensor installed at the hitting end of the cue, and a data processing unit.
2. The dual-mode three-dimensional force sensor billiard hitting system according to claim 1, wherein The signals of the static capacitive dielectric layer and the dynamic piezoelectric thin film are transmitted to the data processing unit through the lower and upper electrodes, and the data processing unit processes the signals through the trained neural network model to obtain the hitting force, angle, and rotation parameters.
3. A dual-mode three-dimensional force sensor billiard hitting system according to claim 1, characterized in that The system can detect various basic cueing methods such as straight hitting, spin hitting, and bank shot hitting.
4. A dual-mode three-dimensional force sensor billiard hitting system according to claim 1, characterized in that, The data processing unit performs real-time analysis on the hitting action and provides feedback information, and gives feedback suggestions on the hitting action.
5. A dual-mode three-dimensional force sensor billiard hitting system according to claim 1, characterized in that, The PI substrate has flexibility and mechanical strength, and can support the sensor and adapt to the mechanical changes during hitting.
6. A dual-mode three-dimensional force sensor billiard hitting system according to claim 1, characterized in that, The encapsulation above PDMS and the encapsulation below PDMS are used to protect the internal sensors and circuits, and buffer and protect the impact force through their own materials and structures.
7. A dual-mode three-dimensional force sensor billiard hitting system according to claim 1, characterized in that, The neural network structure adopts an early stopping strategy with an early stopping patience value of 200 during the training process.