A billiard table light interaction and training evaluation method based on a hitting event recognition

Abstract

The application discloses a billiard table light interaction and training evaluation method based on a hitting event recognition, relates to the technical field of intelligent sports equipment, and breaks the limitation that traditional leveling lacks real-time sensing by arranging an inclination and displacement sensor at a key node, acquiring a table surface posture in real time, and comparing the table surface posture with a horizontal safety threshold; a trend is extracted through long-period filtering, automatic non-inductive fine adjustment is triggered, an artificial calibration problem and a ball route deviation caused by a sub-horizontal table surface are eliminated, a real trajectory collected through vision is compared with a theoretical model to be more pure, and high confidence of a training evaluation report is ensured; meanwhile, actual posture changes are compared with expected values, and hardware faults are accurately distinguished; after diagnosis, a protection mode against over-adjustment interlocking is started immediately, error compensation instructions are cut off, and a locking protection mode is entered, red light is sent out through a light system to fundamentally avoid repeatedly error compensation caused by misjudgment of faults as drift, and slight faults are prevented from evolving into severe table surface oscillation or structural damage.

Description

Technical Field

[0001] This invention relates to the field of intelligent sports equipment technology, specifically to a method for interactive lighting and training evaluation of a billiard table based on ball-hitting event recognition. Background Technology

[0002] Initially, billiards training relied heavily on human experience and video playback, which had limitations such as delayed feedback and low quantification accuracy. With the advent of computer vision technology, researchers began to use top-mounted cameras for ball detection and trajectory tracking, enabling the digital recording of basic match data.

[0003] Traditional billiard tables often rely on one-time or periodic manual leveling (such as adjusting the threaded legs), lacking real-time sensing and closed-loop control of the table's posture. After leveling, due to slow-varying drift factors such as ground settlement, creep of the wooden structure, and material deformation caused by temperature and humidity changes, the table surface will gradually deviate from its initial level state. However, due to the lack of a multi-point tilt / displacement sensor network and drift compensation mechanism, this deviation cannot be automatically detected and corrected, relying solely on periodic manual calibration. The result is often that the table surface is in a "sub-level" state during matches or training, causing systematic and unexpected deviations in the ball's trajectory, severely impacting shot consistency.

[0004] The aforementioned static leveling failure problem was not automatically resolved by introducing a multi-point hydraulic leveling system. Instead, the increased system complexity exposed new vulnerabilities: when individual hydraulic actuators experience slow leakage or jamming, or when a tilt sensor exhibits zero-point drift, traditional methods struggle to distinguish between actual table tilt and "sensor / actuator malfunction." More critically, because the drift and attitude changes caused by actuator leakage, as described in Disadvantage 1, have similar time-domain characteristics, without residual analysis and consistency verification mechanisms, the system may misjudge the fault as normal drift and continue erroneous compensation, leading to over-adjustment or leveling oscillations. Ultimately, a minor leak may be repeatedly accumulated and corrected, gradually evolving into a complete collapse of the table's level, or even damage to the table structure. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for interactive and training evaluation of billiard table lighting based on ball-hitting event recognition, thereby resolving the aforementioned technical deficiencies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for interactive and training evaluation of billiard table lighting based on ball-hitting event recognition, specifically including the following steps: S1. A multi-point attitude sensing network is pre-deployed at the key force and support nodes of the billiard table to obtain the tilt angle and displacement status data of multiple points on the table in real time, so as to generate a set of real-time attitude data of the table. S2. Based on the established mapping relationship model between each support point and the overall platform posture, the real-time posture data set of the platform is analyzed to determine the abnormal area that deviates from the reference horizontal plane and the corresponding hydraulic actuator to be activated, and the target adjustment amount of each hydraulic actuator to be activated is initially calculated, and the closed-loop hydraulic leveling mode is started. S3. During the ongoing billiards training or competition, monitor and extract the characteristics of environmental and physical factors that cause the slow drift of the table surface in real time, and construct a set of slow drift states. Based on the set of slow drift states, predict the cumulative deformation assessment value of each support point, and determine whether the current table surface has deviated from the initial horizontal state due to ground settlement or material creep, so as to determine whether a fine-tuning compensation mechanism needs to be triggered. S4. Simultaneously, perform residual analysis and consistency verification on the operating status data of the multi-point attitude sensing network and each hydraulic actuator; then determine whether the currently detected table attitude change is caused by real slow drift or by hardware failure; when it is determined to be a hardware failure rather than real environmental drift, intercept the error compensation command, determine that the current billiard table enters the lock protection mode and triggers an abnormal alarm to avoid cumulative correction leading to the collapse of the entire table level.

[0007] Preferably, the multi-point attitude sensing network pre-deployed in step S1 specifically acquires multi-point attitude data in a matrix form on the platform by combining multiple tilt sensors and displacement sensors. After acquiring real-time status data from multiple points on the platform, the real-time attitude data of each acquisition point is compared with a preset absolute horizontal safety threshold. The specific judgment logic is as follows: When the real-time attitude data of any acquisition point exceeds the absolute horizontal safety threshold, it is marked as a non-horizontal anomaly point, and the closed-loop hydraulic leveling mode of step S2 is triggered. When the real-time attitude data of all collection points are within the absolute level safety threshold, the current platform is determined to be in a standard level state suitable for training and evaluation.

[0008] Preferably, in step S2, the target adjustment amount is calculated based on the established mapping relationship model, and the specific process is as follows: The impact of each hydraulic actuator's unit lifting motion on the local and overall horizontal posture of the platform was pre-tested and recorded. These impacts were then integrated to construct an impact coefficient mapping matrix between the leveling point and the platform posture. Obtain the spatial location and tilt deviation of the current non-horizontal anomaly point, and perform reverse derivation using the influence coefficient mapping matrix to calculate the target displacement or target pressure value required for each hydraulic actuator to be activated in order to offset the current tilt deviation.

[0009] Preferably, after activating the closed-loop hydraulic leveling mode, the corresponding hydraulic actuator is driven according to the calculated target displacement or target pressure value; During this driving process, the multi-point attitude perception network maintains high-frequency sampling and feeds back the new attitude data generated after adjustment to the control center in real time. The control center calculates the remaining deviation between the current attitude and the absolute horizontal plane. Using proportional-integral-differential logic, the output of the hydraulic actuator is continuously corrected based on the remaining deviation, and real-time iteration is performed until the remaining deviation is reduced to infinitely close to zero and is in a convergent state, thus completing a single closed-loop leveling.

[0010] Preferably, the slow-change drift state set in step S3 is derived from data sources including tabletop micro-tilt trend data extracted after smoothing and filtering through a long-period time window, and external temperature and humidity change data. When predicting the cumulative deformation assessment value based on the set of slow-drift states, the tilt change rate per unit time is correlated with the duration, and the deformation coefficient of the wood structure caused by temperature and humidity changes is added for comprehensive calculation. The calculated cumulative deformation assessment value is compared with the preset drift tolerance lower limit value: if the assessment value is greater than the drift tolerance lower limit value, the fine-tuning compensation mechanism is triggered.

[0011] Preferably, in step S4, residual analysis and consistency verification are used to distinguish between actual table tilt and sensor or actuator malfunctions. The specific processing procedure is as follows: After issuing the hydraulic adjustment command, calculate the theoretically expected value of the change in table posture; The actual attitude change values ​​collected by the multi-point attitude perception network are compared with the expected attitude change values ​​to obtain the adjustment residual values. If the adjustment residual value continues to be greater than the preset fault tolerance boundary value, it is determined that the current hydraulic actuator is not working normally according to the instruction or the sensor perception is distorted, which falls within the scope of hardware failure. If the adjusted residual value is less than or equal to the preset fault tolerance boundary value, then it is determined that the hardware response is normal and the current attitude change characteristics belong to normal slow drift.

[0012] Preferably, once the fault is determined to be within the scope of hardware failure, further refined fault diagnosis is performed, with the specific logic as follows: By comparing the output value curves of multiple sensors in adjacent areas, if the data of a single sensor undergoes a step change while the data of its surrounding adjacent sensors remain stable, it is diagnosed as a zero-point drift fault of that single sensor. Monitor the pressure maintenance status of a single hydraulic actuator under closed-loop control. If, without severe external impact, the pressure output value at a certain leveling point shows a continuous and slow decay, accompanied by uninstructed sinking of the local platform, then the hydraulic actuator is diagnosed as having a slow leakage or jamming fault.

[0013] Preferably, when a hardware fault such as hydraulic leakage or sensor zero drift is diagnosed, the over-adjustment interlock logic is immediately activated; The anti-over-adjustment interlock logic will forcibly cut off the hydraulic output commands of the fault point and related coordinated points, refuse to execute the drift compensation action generated based on error perception data, prevent minor faults from being repeatedly accumulated and corrected, which could cause severe oscillations of the entire tabletop posture and damage to the table structure. At the same time, the tabletop lighting system will flash a red warning beam to indicate maintenance.

[0014] Preferably, when steps S3 and S4 determine that the table surface is in a healthy slow-drift state and there is no hardware failure, the hydraulic system will be automatically driven to perform seamless fine-tuning and calibration during training or competition breaks, including between two games or when there is no ball hitting action on the table, based on the calculated cumulative deformation evaluation value, so that the table surface is always maintained on the ideal absolute level reference surface.

[0015] Preferably, after ensuring the pool table is in an absolutely level position, the ball-hitting event recognition and training assessment are initiated: The real ball rolling trajectory and impact event are captured by the top vision sensor. Since the systematic unexpected deviation caused by the sub-level state of the table has been eliminated by the bottom closed-loop leveling and drift compensation, the real trajectory data is accurately compared with the standard theoretical model. Based on the error data derived from the comparison, which is purely caused by the player's hitting technique, including hitting power, hitting point deviation, and spin control, a high-confidence training evaluation report is generated by combining the table lighting interactive system to project differentiated paths and correction guidance.

[0016] This invention provides a method for interactive and training evaluation of billiard table lighting based on ball-hitting event recognition. It has the following beneficial effects: This invention overcomes the limitations of traditional billiard tables in lacking real-time sensing and closed-loop control by deploying a multi-point attitude sensing network composed of tilt and displacement sensors at key stress and support nodes to acquire real-time state data and generate a set of real-time table attitude data. This data is then compared with a preset absolute level safety threshold. Based on the influence coefficient mapping matrix between the leveling point and the table attitude, the target displacement or target pressure value is derived inversely. Proportional-integral-differential logic is used to correct the output based on the remaining deviation to achieve precise closed-loop leveling. For slow-varying drift factors such as ground subsidence or material creep, a long-period time window smoothing filter is applied. By extracting minute tilt trend data and external temperature and humidity change data, a slow-drift state set is constructed to predict the cumulative deformation assessment value of each support point. When the assessment value is greater than the preset drift tolerance lower limit, a fine-tuning compensation mechanism is automatically triggered during the interval period to perform imperceptible fine-tuning calibration, overcoming the problem of relying on manual periodic calibration and eliminating the systematic unexpected offset caused by the sub-level state of the table. This makes the comparison between the real trajectory data obtained by the vision sensor and the standard theoretical model more pure, thereby ensuring the high confidence of the training evaluation report generated based on error data such as hitting force, hitting point deviation, and spin control.

[0017] To address the potential for misjudgments in multi-point hydraulic leveling systems caused by actuator malfunctions or sensor zero-point drift, this invention introduces a residual analysis and consistency verification mechanism. It compares the actual attitude change values ​​collected by the multi-point attitude sensing network with the expected attitude change values ​​to obtain the adjustment residual value. This residual value is then compared with a preset tolerance boundary value to accurately distinguish between actual slow drift and hardware malfunctions. When the adjustment residual value consistently exceeds the tolerance boundary value, further analysis is conducted by comparing the output curves of sensors in adjacent areas and monitoring the pressure maintenance status of individual hydraulic actuators under closed-loop control. This enables the detection of single sensor zero-point drift faults or slow leakage of hydraulic actuators. The system provides refined diagnosis of stuck-out faults. Once a fault is diagnosed, the system immediately intercepts the error compensation command and activates the anti-over-adjustment interlock logic, forcibly cutting off the hydraulic output commands of the fault point and related coordinating points, and refusing to execute the drift compensation action generated based on the error perception data. Subsequently, the system determines that the billiard table has entered the locked protection mode and triggers an abnormal alarm by flashing a red warning beam through the tabletop lighting system. This measure fundamentally solves the defect of traditional methods that misjudge hardware faults as normal slow drift and continue to compensate incorrectly, effectively avoiding the serious consequences of minor faults being repeatedly accumulated and corrected, gradually evolving into severe oscillations of the entire tabletop posture and damage to the table structure. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the frame structure of S4 in this invention; Detailed Implementation The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figures 1 to 2 This invention provides a method for interactive and training evaluation of billiard table lighting based on ball-hitting event recognition, specifically including the following steps: S1. A multi-point attitude sensing network is pre-deployed at the key force and support nodes of the billiard table to obtain the tilt angle and displacement status data of multiple points on the table in real time, so as to generate a set of real-time attitude data of the table. S2. Based on the established mapping relationship model between each support point and the overall platform posture, the real-time posture data set of the platform is analyzed to determine the abnormal area that deviates from the reference horizontal plane and the corresponding hydraulic actuator to be activated, and the target adjustment amount of each hydraulic actuator to be activated is initially calculated, and the closed-loop hydraulic leveling mode is started. S3. During the ongoing billiards training or competition, monitor and extract the characteristics of environmental and physical factors that cause the slow drift of the table surface in real time, and construct a set of slow drift states. Based on the set of slow drift states, predict the cumulative deformation assessment value of each support point, and determine whether the current table surface has deviated from the initial horizontal state due to ground settlement or material creep, so as to determine whether a fine-tuning compensation mechanism needs to be triggered. S4. Simultaneously, perform residual analysis and consistency verification on the operating status data of the multi-point attitude sensing network and each hydraulic actuator; then determine whether the currently detected table attitude change is caused by real slow drift or by hardware failure; when it is determined to be a hardware failure rather than real environmental drift, intercept the error compensation command, determine that the current billiard table enters the lock protection mode and triggers an abnormal alarm to avoid cumulative correction leading to the collapse of the entire table level.

[0020] The multi-point attitude sensing network pre-deployed in step S1 specifically acquires multi-point attitude data in a matrix form on the platform by combining multiple tilt sensors and displacement sensors. After acquiring real-time status data from multiple points on the platform, the real-time attitude data of each acquisition point is compared with a preset absolute horizontal safety threshold. The specific judgment logic is as follows: When the real-time attitude data of any acquisition point exceeds the absolute horizontal safety threshold, it is marked as a non-horizontal anomaly point, and the closed-loop hydraulic leveling mode of step S2 is triggered. When the real-time attitude data of all collection points are within the absolute level safety threshold, the current platform is determined to be in a standard level state suitable for training and evaluation.

[0021] In step S2, the target adjustment amount is calculated based on the established mapping relationship model. The specific process is as follows: The impact of each hydraulic actuator's unit lifting motion on the local and overall horizontal posture of the platform was pre-tested and recorded. These impacts were then integrated to construct an impact coefficient mapping matrix between the leveling point and the platform posture. Obtain the spatial location and tilt deviation of the current non-horizontal anomaly point, and perform reverse derivation using the influence coefficient mapping matrix to calculate the target displacement or target pressure value required for each hydraulic actuator to be activated in order to offset the current tilt deviation.

[0022] After activating the closed-loop hydraulic leveling mode, the corresponding hydraulic actuator is driven according to the calculated target displacement or target pressure value. During this driving process, the multi-point attitude perception network maintains high-frequency sampling and feeds back the new attitude data generated after adjustment to the control center in real time. The control center calculates the remaining deviation between the current attitude and the absolute horizontal plane. Using proportional-integral-differential logic, the output of the hydraulic actuator is continuously corrected based on the remaining deviation, and real-time iteration is performed until the remaining deviation is reduced to infinitely close to zero and is in a convergent state, thus completing a single closed-loop leveling.

[0023] The slow-drift state set in step S3 is derived from data sources including tabletop micro-tilt trend data extracted after smoothing and filtering through a long-period time window, and external temperature and humidity change data. When predicting the cumulative deformation assessment value based on the set of slow-drift states, the tilt change rate per unit time is correlated with the duration, and the deformation coefficient of the wood structure caused by temperature and humidity changes is added for comprehensive calculation. The calculated cumulative deformation assessment value is compared with the preset drift tolerance lower limit value: if the assessment value is greater than the drift tolerance lower limit value, the fine-tuning compensation mechanism is triggered.

[0024] In step S4, residual analysis and consistency verification are used to distinguish between actual table tilt and sensor or actuator malfunctions. The specific processing procedure is as follows: After issuing the hydraulic adjustment command, calculate the theoretically expected value of the change in table posture; The actual attitude change values ​​collected by the multi-point attitude perception network are compared with the expected attitude change values ​​to obtain the adjustment residual values. If the adjustment residual value continues to be greater than the preset fault tolerance boundary value, it is determined that the current hydraulic actuator is not working normally according to the instruction or the sensor perception is distorted, which falls within the scope of hardware failure. If the adjusted residual value is less than or equal to the preset fault tolerance boundary value, then it is determined that the hardware response is normal and the current attitude change characteristics belong to normal slow drift.

[0025] Once the fault is determined to be a hardware failure, further refined fault diagnosis is performed. The specific logic is as follows: By comparing the output value curves of multiple sensors in adjacent areas, if the data of a single sensor undergoes a step change while the data of its surrounding adjacent sensors remain stable, it is diagnosed as a zero-point drift fault of that single sensor. Monitor the pressure maintenance status of a single hydraulic actuator under closed-loop control. If, without severe external impact, the pressure output value at a certain leveling point shows a continuous and slow decay, accompanied by uninstructed sinking of the local platform, then the hydraulic actuator is diagnosed as having a slow leakage or jamming fault.

[0026] When a hardware fault such as hydraulic leakage or sensor zero drift is diagnosed, the over-adjustment interlock logic is immediately activated. The anti-over-adjustment interlock logic will forcibly cut off the hydraulic output commands of the fault point and related coordinated points, refuse to execute the drift compensation action generated based on error perception data, prevent minor faults from being repeatedly accumulated and corrected, which could cause severe oscillations of the entire tabletop posture and damage to the table structure. At the same time, the tabletop lighting system will flash a red warning beam to indicate maintenance.

[0027] When steps S3 and S4 determine that the table surface is in a healthy slow-drift state and there is no hardware failure, the hydraulic system will be automatically driven to perform seamless fine-tuning and calibration based on the calculated cumulative deformation evaluation value during training or competition breaks, including between two games or when there is no ball hitting action on the table, so that the table surface is always kept on the ideal absolute level reference surface.

[0028] Once the pool table is ensured to be in an absolutely level position, initiate the ball-hitting event identification and training assessment: The real ball rolling trajectory and impact event are captured by the top vision sensor. Since the systematic unexpected deviation caused by the sub-level state of the table has been eliminated by the bottom closed-loop leveling and drift compensation, the real trajectory data is accurately compared with the standard theoretical model. Based on the error data derived from the comparison, which is purely caused by the player's hitting technique, including hitting power, hitting point deviation, and spin control, a high-confidence training evaluation report is generated by combining the table lighting interactive system to project differentiated paths and correction guidance.

[0029] Furthermore, the multi-point attitude sensing network pre-deployed in step S1 is specifically deployed using a matrix spatial topology. Specifically, at the junction of the wooden frame and supporting legs of a standard billiard table (including 6-leg or 8-leg structures), a high-precision laser displacement sensor and a dual-axis tilt sensor are coaxially installed at each support node. Thus, the real-time feedback data from each acquisition point constitutes an M×N dimensional real-time attitude data set for the table surface, capable of accurately capturing the pitch angles of the table surface along the X and Y axes and the vertical displacement of each support point. The preset "absolute horizontal safety threshold" is strictly quantified as follows: the relative height difference between any two diagonal support points ≤ 0.1 mm, and the absolute tilt angle deviation at any position on the entire table surface ≤ 0.05 degrees.

[0030] Regarding the specific mathematical derivation model for calculating the target adjustment amount in step S2, this embodiment implements it by constructing an "influence coefficient mapping matrix" with the properties of the Jacobian matrix. During system initialization, the control center sequentially instructs each hydraulic actuator to output a unit lifting displacement (set to 1mm in this embodiment), while simultaneously recording the tilt angle and displacement changes at N acquisition points across the entire platform, fed back by the multi-point attitude sensing network. The partial derivatives of each hydraulic actuator (input variable) and the attitude changes at each acquisition point (output variable) are integrated to generate the influence coefficient mapping matrix H. When a non-horizontal anomaly is detected, the error vector EE containing the tilt deviation is obtained. The control center uses the pseudo-inverse matrix H+ of the influence coefficient mapping matrix, through a formula algorithm... By performing reverse derivation, the target adjustment vector ΔU (i.e., target displacement or target pressure value) of each hydraulic actuator required to offset the current error is accurately calculated.

[0031] During the activation of the closed-loop hydraulic leveling mode, the high-frequency sampling is configured with a sampling rate of no less than 100Hz. The control center inputs the residual deviation between the current feedback attitude and the absolute horizontal plane into the PID controller. The proportional (P) coefficient responds to the severity of the current deviation, the integral (I) coefficient eliminates the steady-state error caused by the static hydraulic friction dead zone, and the derivative (D) coefficient suppresses the overshoot oscillation in the initial stage of hydraulic drive. The system iterates in real time until the residual deviation remains within 0.02mm for 33 consecutive seconds. At this point, the system is determined to be in a convergent state, completing a single closed-loop leveling operation.

[0032] For the extraction of the slow-drift state set in step S3, this embodiment uses a sliding time window mean filtering algorithm (the window length is set to 30 minutes in this embodiment) to filter out instantaneous high-frequency disturbance noise caused by ball impact or players leaning against the table edge, and extract the table surface's minute tilt trend data that truly reflects ground subsidence or structural creep. When predicting the cumulative deformation assessment value, the control center uses the following quantification calculation logic: Cumulative deformation assessment value == (current moment filter tilt angle) Initial reference tilt angle) Where K is the pre-determined coefficient of thermal expansion and contraction of the wooden countertop structure, and ΔT is the difference between the temperature collected by the current temperature and humidity sensor and the reference temperature. When the evaluated value is greater than the preset drift tolerance lower limit (set to an equivalent settlement of 0.3 mm in this embodiment), the fine-tuning compensation mechanism is triggered.

[0033] In the refined fault diagnosis of step S4, the residual analysis that distinguishes between "real drift" and "hardware fault" has a strict numerical judgment standard: the control center calculates the difference between the theoretical attitude change expected value derived in reverse and the actual attitude change value transmitted back by the sensors. If this adjustment residual value is greater than the system fault tolerance boundary value (set to 15% of the expected value in this embodiment) for five consecutive control cycles, it is judged as a hardware fault. At this time, a refined investigation is performed, including: (1) Zero-point drift fault diagnosis: Calculate the derivative (i.e., rate of change) of the displacement values ​​of three or more sensors in adjacent areas with respect to time. If the rate of change of a single sensor shows a step change of more than 5 mm / s, while the rate of change of its adjacent sensors remains in a stable state below 0.01 mm / s, since the platform has rigid body continuity, it is impossible for a single point to tear instantaneously. Therefore, it is clearly diagnosed that the single sensor has a zero-point drift fault.

[0034] Actuator leakage or jamming diagnosis: After locking the servo valve of the hydraulic actuator, if the built-in pressure sensor shows that the pressure output value of a certain leveling point is continuously decreasing at a rate of more than 0.5 MPa per minute, and the displacement sensor in the corresponding area simultaneously detects a non-commanded subsidence of more than 0.1 mm, then the hydraulic actuator is diagnosed as having a slow leakage.

[0035] When the above fault is diagnosed, the system's "over-adjustment interlock logic" forcibly cuts off the PWM drive signal of the abnormal channel at the electrical and software control layers. At the same time, in order to avoid the "broken window effect," the "associated coordination points" on the same load-bearing beam as the faulty hydraulic cylinder will also be shielded, stopping all compensation actions.

[0036] The "seamless fine-tuning calibration" triggered when the system is in a healthy state is strictly limited to the following: the system must receive an idle signal from the top visual sensor indicating that "no ball rolls on the table for 6060 consecutive seconds" before it can be activated; and during the fine-tuning process, the lifting rate of the hydraulic actuator is forcibly limited to below 0.05mm / s, and the system operating noise is limited to below 30dB, ensuring that the table can silently return to absolute level while players are discussing tactics or during breaks. It is precisely based on this absolutely guaranteed level benchmark at the bottom layer that the deflection of the ball's rolling trajectory captured by the top visual sensor is 100% attributed by the system to the side spin friction or uneven force point generated when the player hits the ball, thus giving the final training evaluation report, which includes the force of the shot, the deviation of the point of contact, and the spin control, a competitive level of realism and confidence.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for interactive and training evaluation of billiard table lighting based on ball-hitting event recognition, characterized in that, Specifically, the steps include: A multi-point attitude sensing network is pre-deployed at key force and support nodes on the billiard table to acquire real-time tilt angle and displacement data of multiple points on the table, thereby generating a real-time attitude data set of the table. Based on the established mapping relationship model between each support point and the overall platform posture, the real-time posture data set of the platform is analyzed to determine the abnormal area that deviates from the reference horizontal plane and the corresponding hydraulic actuator to be activated, and the target adjustment amount of each hydraulic actuator to be activated is initially calculated, and the closed-loop hydraulic leveling mode is activated. During billiards training or competition, the characteristics of environmental and physical factors that cause slow drift of the table surface are monitored and extracted in real time to construct a set of slow drift states. Based on the set of slow drift states, the cumulative deformation assessment value of each support point is predicted to determine whether the current table surface has deviated from the initial horizontal state due to ground settlement or material creep, so as to determine whether a fine-tuning compensation mechanism needs to be triggered. Simultaneously, residual analysis and consistency verification are performed on the operating status data of the multi-point attitude perception network and each hydraulic actuator. Then determine whether the detected tabletop attitude change is caused by real slow drift or by hardware failure; when it is determined to be a hardware failure rather than real environmental drift, intercept the error compensation command, determine that the current billiard table enters the lock protection mode and triggers an abnormal alarm to avoid cumulative correction leading to the collapse of the entire tabletop level.

2. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition as described in claim 1, characterized in that: The pre-deployed multi-point attitude sensing network specifically acquires multi-point attitude data in a matrix format on the platform by combining multiple tilt sensors and displacement sensors. After acquiring real-time status data from multiple points on the platform, the real-time attitude data of each acquisition point is compared with a preset absolute horizontal safety threshold. The specific judgment logic is as follows: When the real-time attitude data of any acquisition point exceeds the absolute horizontal safety threshold, it is marked as a non-horizontal anomaly point, and the closed-loop hydraulic leveling mode is triggered. When the real-time attitude data of all collection points are within the absolute level safety threshold, the current platform is determined to be in a standard level state suitable for training and evaluation.

3. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition as described in claim 1, characterized in that: The specific process for calculating the target adjustment amount based on the established mapping relationship model is as follows: The impact of each hydraulic actuator's unit lifting motion on the local and overall horizontal posture of the platform was pre-tested and recorded. These impacts were then integrated to construct an impact coefficient mapping matrix between the leveling point and the platform posture. Obtain the spatial location and tilt deviation of the current non-horizontal anomaly point, and perform reverse derivation using the influence coefficient mapping matrix to calculate the target displacement or target pressure value required for each hydraulic actuator to be activated in order to offset the current tilt deviation.

4. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 3, characterized in that: After activating the closed-loop hydraulic leveling mode, the corresponding hydraulic actuator is driven according to the calculated target displacement or target pressure value. During this driving process, the multi-point attitude perception network maintains high-frequency sampling and feeds back the new attitude data generated after adjustment to the control center in real time. The control center calculates the remaining deviation between the current attitude and the absolute horizontal plane. Using proportional-integral-differential logic, the output of the hydraulic actuator is continuously corrected based on the remaining deviation, and real-time iteration is performed until the remaining deviation is reduced to infinitely close to zero and is in a convergent state, thus completing a single closed-loop leveling.

5. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 1, characterized in that: The data sources for the slowly changing drift state set include tabletop tilt trend data extracted after smoothing and filtering through a long-period time window, as well as external temperature and humidity change data. When predicting the cumulative deformation assessment value based on the set of slow-drift states, the tilt change rate per unit time is correlated with the duration, and the deformation coefficient of the wood structure caused by temperature and humidity changes is added for comprehensive calculation. The calculated cumulative deformation assessment value is compared with the preset drift tolerance lower limit value: if the assessment value is greater than the drift tolerance lower limit value, the fine-tuning compensation mechanism is triggered.

6. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 5, characterized in that: The actual table tilt is distinguished from sensor or actuator failure by residual analysis and consistency verification. The specific handling process is as follows: After issuing the hydraulic adjustment command, calculate the theoretically expected value of the change in table posture; The actual attitude change values ​​collected by the multi-point attitude perception network are compared with the expected attitude change values ​​to obtain the adjustment residual values. If the adjustment residual value continues to be greater than the preset fault tolerance boundary value, it is determined that the current hydraulic actuator is not working normally according to the instruction or the sensor perception is distorted, which falls within the scope of hardware failure. If the adjusted residual value is less than or equal to the preset fault tolerance boundary value, then it is determined that the hardware response is normal and the current attitude change characteristics belong to normal slow drift.

7. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 6, characterized in that: Once the fault is determined to be a hardware failure, further refined fault diagnosis is performed. The specific logic is as follows: By comparing the output value curves of multiple sensors in adjacent areas, if the data of a single sensor undergoes a step change while the data of its surrounding adjacent sensors remain stable, it is diagnosed as a zero-point drift fault of that single sensor. Monitor the pressure maintenance status of a single hydraulic actuator under closed-loop control. If, without severe external impact, the pressure output value at a certain leveling point shows a continuous and slow decay, accompanied by uninstructed sinking of the local platform, then the hydraulic actuator is diagnosed as having a slow leakage or jamming fault.

8. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 7, characterized in that: When a hardware fault such as hydraulic leakage or sensor zero drift is diagnosed, the over-adjustment interlock logic is immediately activated. The anti-over-adjustment interlock logic will forcibly cut off the hydraulic output commands of the fault point and related coordinated points, refuse to execute the drift compensation action generated based on error perception data, prevent minor faults from being repeatedly accumulated and corrected, which could cause severe oscillations of the entire tabletop posture and damage to the table structure. At the same time, the tabletop lighting system will flash a red warning beam to indicate maintenance.

9. The method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition according to claim 1, characterized in that: When the table surface is determined to be in a healthy slow-drift state and there is no hardware failure, the hydraulic system will be automatically driven to perform seamless fine-tuning and calibration during training or match breaks, including between games or when there is no ball hitting action on the table, based on the calculated cumulative deformation assessment value, so that the table surface is always kept on the ideal absolute level reference surface.

10. A method for interaction and training evaluation of billiard table lighting based on ball-hitting event recognition as described in claim 9, characterized in that: Once the pool table is ensured to be in an absolutely level position, initiate the ball-hitting event identification and training assessment: The real ball rolling trajectory and impact event are captured by the top vision sensor. Since the systematic unexpected deviation caused by the sub-level state of the table has been eliminated by the bottom closed-loop leveling and drift compensation, the real trajectory data is accurately compared with the standard theoretical model. Based on the error data derived from the comparison, which is purely caused by the player's hitting technique, including hitting power, hitting point deviation, and spin control, a high-confidence training evaluation report is generated by combining the table lighting interactive system to project differentiated paths and correction guidance.

Images