Pressure real-time online monitoring dental treatment machine air supply pressure stabilizing system

Abstract

The application discloses a pressure real-time online monitoring dental treatment machine gas supply pressure stabilizing system and belongs to the technical field of medical instrument automation. The application adopts a feedforward linkage framework of signal identification, disturbance prediction and buffer throttling, extracts pressure change rate and waveform characteristics through the combination of short-time window and long-time window, accurately identifies the disturbance type, matches the corresponding prediction sub-model to deduce the future pressure trajectory, makes the multi-stage buffer chamber adjust the pressure distribution and air resistance characteristics in advance, realizes active inhibition instead of passive compensation, significantly improves the pressure stability in the multi-terminal random gas use scene, determines the initial correction direction according to the disturbance source type, adjusts in the graded position, triggers the staged abnormal diagnosis when repeated failure occurs, quickly locates the fault root, outputs the operable treatment strategy, and guarantees the long-term reliable operation of the system. With the automatic optimization of equipment aging, the equipment life is prolonged, the operation and maintenance cost is reduced, and the intelligentization and high reliability management of the dental treatment machine gas supply system are improved.

Description

Technical Field

[0001] This invention relates to the field of medical device automation technology, specifically to a real-time online pressure monitoring system for the air supply and stabilization of a dental treatment machine. Background Technology

[0002] In clinical use, dental treatment units typically need to connect to multiple treatment terminals simultaneously, such as high-speed handpieces, scalers, and three-way guns. This often results in situations where multiple treatment terminals are randomly started and stopped, leading to drastic and frequent pressure fluctuations in the gas supply system. Traditional gas supply pressure stabilization solutions mostly employ passive accumulators or fixed-parameter pressure reducing valves, which are lag-compensation-type regulation mechanisms. These can only provide coarse compensation for slowly changing gas source pressure and cannot cope with the instantaneous pressure drop caused by sudden large-flow extractions, nor can they suppress periodic pressure oscillations caused by the resonance of the air column in the buffer chamber. Furthermore, they lack the ability to identify and anticipate disturbance types online, and the adjustment action lags significantly behind the moment the disturbance occurs. Clinically, this manifests as instantaneous handpiece rotation speed fluctuations and unstable atomization cooling effects during treatment. In severe cases, it can even cause saliva backflow or gas path contamination, directly affecting treatment safety and patient experience.

[0003] Meanwhile, the existing calibration strategies of gas supply systems generally rely on passive correction of a single pressure feedback deviation, without considering the differences in the types of disturbance sources. This results in a lack of physical basis for the calibration direction, a large number of iterations and a slow convergence speed. In particular, under complex operating conditions with multiple alternating or superimposed disturbances, repeated overshoot or regulation failure is likely to occur.

[0004] Furthermore, the system lacks real-time monitoring and early warning mechanisms for the performance degradation of actuators such as connecting valves and buffer chambers. Control parameters are usually set once based on factory calibration, and fixed thresholds are prone to frequent false alarms or control failures in the later stages of the entire life cycle. Equipment maintenance relies heavily on manual experience for troubleshooting, resulting in long fault location cycles and high maintenance costs, making it difficult to meet the urgent needs of modern dental treatment equipment for gas supply quality and intelligent operation and maintenance. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time online pressure monitoring system for the air supply and stabilization of dental treatment machines, which has disturbance prediction, adaptive calibration, graded diagnosis and performance self-memory functions, comprehensively improving the air supply quality and operation and maintenance management level of dental treatment machines, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring, comprising: The signal identification module is used to acquire pressure monitoring data in the gas supply source and main gas path in real time, perform trend fitting on the pressure monitoring data, extract the dynamic change characteristics of the gas path pressure, and identify the type of gas path disturbance by combining the preset waveform template. The disturbance prediction module is used to match the corresponding prediction sub-model to predict gas path pressure disturbances based on the gas path disturbance type and generate a prediction result dataset. The buffer throttling module is used to adjust the pressure distribution relationship between the chambers of the buffer unit in the main air path according to the prediction result dataset, and simultaneously adjust the air resistance characteristics on the output branch to obtain the actual pressure response data after adjustment. The adaptive calibration module is used to compare the adjusted actual pressure response data with the estimated values ​​of the expected pressure offset direction and amplitude in the prediction result dataset, calculate the deviation, and perform adaptive calibration based on the deviation. If the number of adaptive calibrations exceeds the preset threshold, the adaptive calibration is determined to have failed and a failure signal is generated. The anomaly diagnosis module is used to acquire failure signals and current gas path pressure monitoring data, and to perform anomaly diagnosis based on a graded anomaly diagnosis strategy. Based on the diagnosis results, it matches the corresponding fault type identifier, fault confidence level and fault handling strategy. The performance evaluation memory module is used to construct the regulation performance evaluation index for the current operating cycle, compare the regulation performance evaluation index with the historical cycle performance index, generate the performance degradation trend curve, and generate early warning signals based on the performance degradation trend curve.

[0007] Furthermore, the specific process by which the signal identification module identifies the type of airflow disturbance includes: Based on the continuous acquisition of raw pressure monitoring data by the pressure sensing unit, the raw pressure monitoring data is standardized and preprocessed to generate a time-domain regularized pressure dataset. At the same time, a circular buffer including short time windows and long time windows is constructed. Based on the short time window, the corresponding short time data segment in the time-domain regularized pressure dataset is retrieved, the pressure sequence within the short time data segment is extracted and the instantaneous rate of change is calculated to obtain the effective pressure rate of change; The effective pressure change rate is compared with the preset aerodynamic sensitivity threshold. If the effective pressure change rate exceeds the preset aerodynamic sensitivity threshold, it is determined that the monitoring node may be disturbed, and the determination results of adjacent detection nodes are obtained at the same time. If it is determined that there is a possibility of disturbance in both adjacent detection nodes, then a valid disturbance event is determined to exist. Based on the start trigger time of the valid disturbance event, pressure waveform segments of a preset duration before and after the start trigger time are extracted and integrated to generate the trigger time waveform segment. Based on the long time window, the corresponding long time data segment is retrieved from the time domain regularized pressure dataset. The low-frequency steady-state component and the high-frequency impact component are separated in the long time data segment, and the long-period trend feature parameters are extracted. The peak impact amplitude and transient response slope of the waveform segment at the trigger moment are extracted as instantaneous impact features and fused with trend feature parameters to construct a multidimensional disturbance feature vector. The multidimensional disturbance feature vector is correlated and matched with the preset waveform template, and the corresponding gas path disturbance type is determined based on the correlation similarity.

[0008] Furthermore, the prediction sub-models include: an impact response prediction sub-model, a gas supply attenuation prediction sub-model, a leakage compensation prediction sub-model, and a resonance suppression prediction sub-model, and each prediction sub-model corresponds one-to-one with the gas path disturbance type.

[0009] Furthermore, the specific process for predicting gas path pressure disturbances includes: Obtain the type label of the gas path disturbance and the effective pressure change rate at the current moment, and match the corresponding prediction sub-model according to the gas path disturbance type label; Extract the pressure sequence data and effective pressure change rate sequence data within a preset number of sampling periods before the current moment, and input them into the corresponding prediction sub-model to deduce the pressure change trajectory of each key node within a preset time window in the future, and draw the expected pressure response curve; Based on the expected pressure response curve, extract the curve feature parameters, and generate a prediction result dataset based on the curve feature parameters, including the disturbance prediction flag, expected pressure offset direction, amplitude estimate, expected disturbance arrival time, and disturbance duration period.

[0010] Furthermore, the disturbance prediction module also includes disturbance source type determination and feedback optimization: Obtain the disturbance prediction flag and the expected arrival time of the disturbance in the prediction result dataset. When the expected arrival time of the disturbance arrives, simultaneously collect the gas supply source pressure monitoring data, main gas path pressure monitoring data and output branch pressure monitoring data to generate a multi-node synchronous observation vector. The multi-node synchronous observation vector is correlated and compared with the estimated values ​​of the expected pressure offset direction and magnitude in the prediction result dataset to determine the actual occurrence node of the current disturbance event. Based on the pressure propagation time sequence of the actual occurrence node, the location of the disturbance source is traced back to generate the disturbance source type determination result. Based on the determination of the disturbance source type, a feedback optimization identifier is generated, and the length of the extrapolation time domain window of the prediction sub-model in the next sampling period and the initial pressure distribution weight of each chamber of the buffer unit are adjusted using the feedback optimization identifier.

[0011] Furthermore, the specific process of tracing the location of the disturbance source in reverse includes: Using the gas supply port as the first monitoring node and the output cavity port of the buffer unit as the second monitoring node, the pressure change time and pressure change rate sign of the first and second monitoring nodes at the expected disturbance arrival time are extracted respectively. Calculate the time difference between the pressure change time of the second monitoring node and the pressure change time of the first monitoring node, and use it as the disturbance propagation delay difference; If the difference in disturbance propagation delay is positive, the pressure change rate of the two monitoring nodes has the same sign, and it indicates that the pressure fluctuation on the gas supply side occurs earlier than that on the buffer unit side, then it is determined to be a disturbance on the gas supply side. If the difference in disturbance propagation delay is negative, or the sign of the rate of change of pressure at the gas supply port is opposite to the sign of the rate of change of pressure at the output chamber of the buffer unit, it indicates that the pressure fluctuation on the gas supply side occurs later than that on the buffer unit side or the fluctuation directions are different, and it is determined to be a disturbance on the gas supply side of the terminal. If the disturbance propagation delay difference is zero or the conditions for determining the sign of the corresponding pressure change rate are not met, then it is determined to be a pipeline structure resonance type. The determined disturbance source type is labeled as a disturbance source attribute tag, which serves as the disturbance source type determination result.

[0012] Furthermore, the specific process of adaptive calibration based on the deviation includes: The deviation is obtained, including pressure offset direction deviation, amplitude deviation and response time deviation. The deviation values ​​of each dimension in the deviation are normalized and weighted to generate a comprehensive deviation value. The overall deviation value is compared with the preset deviation calibration threshold to determine the current deviation level, and the corresponding gear calibration strategy is matched based on the current deviation level. Before executing the gear calibration strategy, the initial correction direction under the current deviation level is determined according to the type of disturbance source, and the corresponding gear calibration strategy is executed according to the determined initial correction direction. After calibration, collect steady-state pressure data of the calibrated gas supply port, buffer chamber connection node and output branch terminal, and compare them with the standard gas supply tolerance range; The difference between the steady-state pressure data of each monitoring node and the center value of the standard gas supply tolerance range is calculated as the residual pressure offset of each node, and the residual pressure offset of each node is compared with the half width of the standard gas supply tolerance range. If the residual pressure offset of all monitoring nodes falls back to the standard gas supply tolerance range, the calibration is deemed valid, and the initial correction direction, deviation level, and actual adjustment parameters are packaged to generate a calibration action label. If the residual pressure offset of any monitoring node does not converge to the standard gas supply tolerance range, a progressive calibration will be triggered while keeping the initial correction direction unchanged. The adjustment step size will be gradually increased until the residual pressure offset converges to the standard gas supply tolerance range, or the number of progressive calibrations reaches the preset number threshold. If, after the number of iterations reaches the preset threshold, there is still a residual pressure offset at any monitoring node that has not converged to the standard gas supply tolerance range, then the adaptive calibration is deemed to have failed.

[0013] Furthermore, the adaptive calibration module also includes: Acquire steady-state pressure data and corresponding historical deviation data, align the historical deviation data with the time axis based on time order, and extract the deviation convergence trajectory features; The deviation convergence trajectory features are associated with the perturbation source attribute labels to generate an association matrix between deviation and perturbation type. The deviation of the deviation convergence trajectory features from the preset standard convergence trajectory under each perturbation type is marked in the association matrix. A waveform template correction factor is generated based on the deviation, and the matching weight of the preset waveform template is dynamically corrected based on the waveform template correction factor.

[0014] Furthermore, the hierarchical abnormality diagnosis strategy includes primary diagnosis, secondary diagnosis and tertiary diagnosis. Each level of diagnosis is executed sequentially according to priority. When any primary diagnosis detects an abnormality, the subsequent diagnosis level is terminated, and the result of the current level diagnosis is output as the abnormality diagnosis conclusion.

[0015] Furthermore, the performance evaluation memory module also includes: taking the current operating cycle as the evaluation unit, at the end of each cycle, using all calibration action labels in the current cycle as the data source, calculating the cycle mean and cycle variance of deviation convergence speed, steady-state holding accuracy, calibration action amplitude and overshoot, and generating the performance evaluation index vector for the current cycle; storing the performance evaluation index vector for the current cycle sequentially into the historical performance database based on the corresponding timestamp, and updating the data points for the current cycle in the performance degradation trend curve.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention extracts the dynamic characteristics of gas path pressure in real time and identifies the type of disturbance. Combined with the corresponding prediction sub-model, it extrapolates the future pressure trajectory, enabling the buffer throttling module to adjust the pressure distribution of multi-stage chambers and the gas resistance characteristics of branches in advance. This achieves feedforward active intervention and blocks the propagation path of sudden pressure drop, overshoot, or resonance from the root.

[0017] 2. This invention performs progressive adaptive calibration based on three-dimensional deviation comparison of direction, amplitude, and time. When calibration fails, it triggers a hierarchical anomaly diagnosis that includes a first-level sensor self-test, a second-level mechanical fault identification, and a third-level model mismatch judgment. It peels away the fault source layer by layer and outputs a clear fault type, confidence level, and handling strategy, which greatly improves the system's fault tolerance and maintenance efficiency.

[0018] 3. This invention constructs periodic performance indicators and generates performance degradation trend curves to achieve early warning of performance degradation of buffer units and valves; at the same time, the warning information is fed back to the calibration module to dynamically adjust the calibration number threshold, so that the system can adaptively adjust the control tolerance as the equipment ages, avoid false alarms or loss of control caused by fixed thresholds, and provide data support for preventive maintenance. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the air supply and pressure stabilization system architecture for a dental treatment machine with real-time online pressure monitoring according to the present invention. Detailed Implementation

[0020] 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.

[0021] Please see Figure 1 A pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring, comprising: The signal identification module is used to acquire pressure monitoring data in the gas supply source and main gas path in real time, perform trend fitting on the pressure monitoring data, extract the dynamic change characteristics of the gas path pressure, and identify the type of gas path disturbance by combining the preset waveform template. The disturbance prediction module is used to predict gas pressure disturbances by matching the corresponding prediction sub-model based on the type of gas disturbance. For example, if the gas pressure drops suddenly when the next treatment terminal is started, the module generates a prediction result dataset, including disturbance prediction flags, expected pressure offset direction and amplitude estimates. The buffer throttling module is used to adjust the pressure distribution relationship between the chambers of the buffer unit in the main air path according to the prediction result dataset. The buffer unit includes multiple chambers arranged in series. Specifically, it adjusts the pressure distribution relationship of the high-pressure chamber, the pressure-stabilizing chamber, and the output chamber. The pressure distribution relationship is achieved by adjusting the opening of the connecting valve between adjacent chambers, and the air resistance characteristics on the output branch are adjusted simultaneously to achieve the pre-suppression and blocking of pressure fluctuations; and the actual pressure response data after adjustment is obtained. The adaptive calibration module is used to compare the adjusted actual pressure response data with the estimated values ​​of the expected pressure offset direction and amplitude in the prediction result dataset, calculate the deviation, including the pressure offset direction deviation, amplitude deviation and response time deviation, and perform adaptive calibration based on the deviation. If the number of adaptive calibrations exceeds the preset threshold, the adaptive calibration is determined to have failed and a failure signal is generated. The anomaly diagnosis module is used to acquire failure signals and current gas path pressure monitoring data, and to perform anomaly diagnosis based on a graded anomaly diagnosis strategy. Based on the diagnosis results, it matches the corresponding fault type identifier, fault confidence level and fault handling strategy. The performance evaluation memory module is used to construct the control performance evaluation indicators for the current operating cycle, including deviation convergence speed, steady-state holding accuracy, calibration action amplitude and overshoot. It compares the control performance evaluation indicators with the historical cycle performance indicators, generates a performance degradation trend curve, and generates an early warning signal when the performance indicators of multiple consecutive cycles drop beyond the preset warning value, prompting preventive maintenance of the buffer unit and connecting valve. The performance evaluation memory module further includes: taking the current operating cycle as the evaluation unit, at the end of each cycle, using all calibration action tags in the current cycle as the data source, calculating the cycle mean and cycle variance of deviation convergence speed, steady-state holding accuracy, calibration action amplitude, and overshoot, and generating the performance evaluation index vector for the current cycle; storing the performance evaluation index vector for the current cycle sequentially into the historical performance database based on the corresponding timestamp, updating the data points for the current cycle in the performance degradation trend curve, and when the warning signal is generated, packaging the warning signal and the performance evaluation index vector for the current cycle and pushing them to the adaptive calibration module as the basis for dynamic adjustment of the preset number threshold.

[0022] In this embodiment, by using signal identification and disturbance prediction, and feedforward active control based on waveform template recognition and sub-model deduction, pre-response time for physical actions is gained for the multi-level chamber buffer throttling module. This enables synchronous pre-matching of air resistance characteristics and precise intervention in pressure distribution, fundamentally suppressing the transmission path of pressure drop or overshoot. By introducing a three-dimensional deviation comparison mechanism, graded abnormality diagnosis is actively triggered when repeated calibration fails. Differentiated handling strategies are output based on fault confidence. The performance evaluation memory module, through the performance degradation trend curve, captures the performance degradation of the buffer unit and connecting valve and provides early warning, while feeding back to the adaptive calibration module as the basis for adaptive adjustment of the calibration number threshold. This allows the system to automatically adjust the control tolerance according to its actual aging state or changes in operating conditions, avoiding frequent false alarms or control failures caused by fixed thresholds in the later stages of the equipment's life cycle. This significantly improves the pressure stability and treatment safety of the dental treatment machine's air supply system in multi-terminal random air use scenarios, while providing traceable data support for preventive maintenance and intelligent operation and maintenance management of the equipment.

[0023] In this embodiment, the specific process by which the signal identification module identifies the type of airflow disturbance includes: Based on pressure sensing units deployed at the gas supply port, main gas path, multi-level buffer chamber connection nodes, and branch nodes of each treatment terminal, raw pressure monitoring data is continuously collected, including real-time gas supply pressure, main gas path steady-state pressure, pressure at each key node, unit millisecond-level pressure change rate, periodic pressure fluctuation frequency, and instantaneous pressure fluctuation amplitude. The raw pressure monitoring data is then standardized and preprocessed to generate a time-domain regularized pressure dataset. Simultaneously, a circular buffer consisting of a short time window and a long time window is constructed. The short time window points to the data segment of the most recent 0.1 seconds in the data pool and is used to calculate the instantaneous rate of change. The long time window points to the data segment of the most recent 1.5 seconds in the data pool and is used to identify low-frequency fluctuation trends. Based on the short time window, the corresponding short time data segment in the time-domain regularized pressure dataset is retrieved, the pressure sequence within the short time data segment is extracted and the instantaneous rate of change is calculated to obtain the effective pressure rate of change; The effective pressure change rate is compared with the preset aerodynamic sensitivity threshold. If the effective pressure change rate exceeds the preset aerodynamic sensitivity threshold, it is determined that the monitoring node may be disturbed, and the determination results of adjacent detection nodes are obtained at the same time. If it is determined that there is a possibility of disturbance in both adjacent detection nodes, then a valid disturbance event is determined to exist. Based on the start trigger time of the valid disturbance event, pressure waveform segments of a preset duration before and after the start trigger time are extracted and integrated to generate the trigger time waveform segment. Based on the long time window, the corresponding long time data segment in the time domain regularized pressure dataset is retrieved. The sliding Fourier transform method is used to separate the low-frequency steady-state component and the high-frequency impact component in the long time data segment, and three types of trend feature parameters are extracted: long-period pressure fluctuation amplitude, fluctuation period mean, and baseline pressure drift. The peak impact amplitude and transient response slope of the waveform segment at the trigger moment are extracted as instantaneous impact features and fused with trend feature parameters to construct a multidimensional disturbance feature vector. The multidimensional disturbance feature vector is correlated and matched with the preset waveform template. Based on the correlation similarity, the corresponding gas path disturbance type is determined, such as terminal instantaneous gas consumption impact disturbance, low frequency drift disturbance due to insufficient gas supply, pipeline leakage and slow pressure drop disturbance, and multi-level buffer cavity resonance periodic fluctuation disturbance. In this embodiment, the instantaneous rate of change is calculated using first-order difference operations, and the difference results are subjected to median filtering to remove high-frequency jitter noise. In this embodiment, the preset waveform template is a five-category hierarchical standard waveform template pre-stored in the module database.

[0024] In this embodiment, by constructing a multi-dimensional parameter collection system and a circular buffer through multi-node pressure sensing, the sensitivity of disturbance identification and the accuracy of classification are significantly improved, providing high-confidence prior information for feedforward active control and enhancing adaptability to complex working conditions.

[0025] In this embodiment, the prediction sub-model includes an impact response prediction sub-model, a gas supply attenuation prediction sub-model, a leakage compensation prediction sub-model, and a resonance suppression prediction sub-model: Among them, the impact response prediction sub-model corresponds to the type of instantaneous gas consumption impact disturbance at the terminal, and is used to deduce the pressure drop magnitude and recovery time of the main gas path and terminal branch nodes under the instantaneous high flow rate extraction condition. The sub-model for predicting gas supply attenuation corresponds to the low-frequency drift disturbance type of insufficient gas supply and is used to deduce the pressure drift trend and drift limit value of each key node under the condition of continuous attenuation of gas supply pressure. The leakage compensation prediction sub-model corresponds to the type of pipeline leakage with gradual pressure drop disturbance, and is used to deduce the pressure drop rate and cumulative pressure drop under small leakage conditions. The resonance suppression prediction sub-model corresponds to the type of periodic fluctuation disturbance of the multi-stage buffer cavity resonance and is used to deduce the pressure fluctuation frequency, fluctuation amplitude and attenuation envelope under the resonance condition of the air column in the buffer cavity.

[0026] In this embodiment, the specific process of predicting gas path pressure disturbances includes: Obtain the type label of the gas path disturbance and the effective pressure change rate at the current moment, and match the corresponding prediction sub-model according to the gas path disturbance type label; Extract the pressure sequence data and effective pressure change rate sequence data within a preset number of sampling periods before the current moment, and input them into the corresponding prediction sub-model. Use the Kalman filter estimation algorithm to deduce the pressure change trajectory of each key node within the preset time window in the future, and draw the expected pressure response curve. Based on the expected pressure response curve, extract the curve characteristic parameters, including the time of the first trough, the pressure value of the first trough, the slope of the pressure fluctuation decay envelope, and the adjustment time for the pressure to recover to a steady state. The prediction result dataset is generated based on the curve feature parameters, including: a disturbance prediction flag, which indicates whether a pressure overshoot event will occur within a preset time window; the expected pressure offset direction, indicating whether the pressure offset is in the upward or downward direction; the amplitude estimate, indicating the maximum expected offset relative to the current steady-state pressure and the offset deviation range; the expected arrival time of the disturbance and the number of disturbance duration periods.

[0027] In this embodiment, the disturbance prediction module further includes disturbance source type determination and feedback optimization: Obtain the disturbance prediction flag and the expected arrival time of the disturbance in the prediction result dataset. When the expected arrival time of the disturbance arrives, simultaneously collect the gas supply source pressure monitoring data, main gas path pressure monitoring data and output branch pressure monitoring data to generate a multi-node synchronous observation vector. The multi-node synchronous observation vector is correlated and compared with the estimated values ​​of the expected pressure offset direction and magnitude in the prediction result dataset to determine the actual occurrence node of the current disturbance event. Based on the pressure propagation time sequence of the actual occurrence node, the location of the disturbance source is traced back to generate the disturbance source type determination result. Based on the determination of the disturbance source type, a feedback optimization identifier is generated, and the feedback optimization identifier is used to adjust the inference time-domain window length of the prediction sub-model in the next sampling period and the initial pressure allocation weight of each chamber of the buffer unit. If the disturbance source type determination result is a gas supply source side fluctuation type, then the length of the inference time domain window is shortened and the initial pressure distribution weight of the gas supply source side chamber is increased. If the disturbance source type determination result is a terminal load mutation type, then the length of the inference time domain window is extended and the initial pressure distribution weight of the output branch side chamber is increased. If the disturbance source type is determined to be a pipeline structure resonance type, then the current simulation time window length remains unchanged and the initial pressure distribution weights of each chamber are balanced.

[0028] In this embodiment, the specific process of tracing the location of the disturbance source in reverse includes: The gas supply port is taken as the first monitoring node, and the output cavity port of the buffer unit is taken as the second monitoring node. The pressure change time and pressure change rate sign of the first and second monitoring nodes at the expected disturbance arrival time are extracted respectively. Calculate the time difference between the pressure change time of the second monitoring node and the pressure change time of the first monitoring node, and use it as the disturbance propagation delay difference; If the difference in disturbance propagation delay is positive, the pressure change rate of the two monitoring nodes has the same sign, and it indicates that the pressure fluctuation on the gas supply side occurs earlier than that on the buffer unit side, then it is determined to be a gas supply side disturbance, corresponding to the gas supply side fluctuation type. If the disturbance propagation delay difference is negative, or the sign of the pressure change rate at the gas supply port is opposite to the sign of the pressure change rate at the output chamber of the buffer unit, it indicates that the pressure fluctuation on the gas supply side occurs later than that on the buffer unit side or the fluctuation direction is different. This is determined to be a disturbance on the gas consumption side of the terminal, corresponding to a sudden change in the terminal load. If the disturbance propagation delay difference is zero or the conditions for determining the sign of the corresponding pressure change rate are not met, then it is determined to be a pipeline structure resonance type. The determined disturbance source type is labeled as a disturbance source attribute tag, which serves as the disturbance source type determination result; In this embodiment, the calculation of the disturbance propagation delay difference is based on the physical propagation characteristics of the gas path, without the need for additional flow sensors; In this embodiment, by comparing the correlation between the multi-node synchronous observation vector and the prediction result dataset, and combining the calculation of the disturbance propagation delay difference, the location of the disturbance source is traced in reverse, accurately distinguishing between gas source side fluctuation type, terminal load mutation type, and pipeline structure resonance type. Based on the disturbance source type determination result, a feedback optimization identifier is generated, and the extrapolation time domain window length of the prediction sub-model and the initial pressure allocation weight of each chamber of the buffer unit are dynamically adjusted in a differentiated manner, realizing the adaptive matching between the prediction strategy and the disturbance source attributes. This significantly improves the accuracy and adaptability of the gas supply pressure stabilization system in long-term operation and complex working conditions, and enhances the system's ability to distinguish and differentiate from multi-source disturbance scenarios such as gas source side fluctuation, terminal load mutation, and pipeline resonance.

[0029] In this embodiment, the specific process of adaptive calibration based on deviation includes: The deviation is obtained, including pressure offset direction deviation, amplitude deviation and response time deviation. The deviation values ​​of each dimension in the deviation are normalized and weighted to generate a comprehensive deviation value. The comprehensive deviation value is compared with the preset deviation calibration threshold to determine the current deviation level, including micro deviation, medium deviation and over-limit deviation, and the corresponding gear calibration strategy is matched based on the current deviation level. Before implementing the gear calibration strategy, the initial correction direction at the current deviation level is determined based on the type of disturbance source: If the disturbance source is a fluctuating type on the gas supply side, the pressure ratio weight of the pressure stabilizing chamber should be increased and the pressure rise time constant should be extended during calibration. If the disturbance source is a terminal load abrupt change type, the output cavity response rate weight should be increased and the voltage regulation cycle should be shortened during calibration. If it is a resonant type of pipeline structure, then the pressure difference correction amount of the three-stage chambers of high pressure chamber, stabilizing chamber and output chamber is synchronously balanced; The corresponding gear calibration strategy is executed based on the determined initial correction direction, where: When the deviation level is micro-deviation, under the boundary constraints of the initial correction direction, the opening degree of the connecting valve is slightly adjusted and the output branch air resistance damping is flexibly corrected, while maintaining the basic pressure distribution ratio of the three-stage chambers unchanged. When the deviation level is medium deviation, under the boundary constraint of the initial correction direction, the pressure ratio fine adjustment of the three-level chambers and the branch air resistance closed-loop correction are synchronously linked, and the priority chamber indicated by the initial correction direction is taken as the primary execution target of this ratio adjustment, and the pressure adjustment amplitude of the other non-priority chambers is limited to not exceeding the preset passive adjustment threshold. When the deviation level is an out-of-limit deviation, under the boundary constraints of the initial correction direction, the initial pressure threshold of the high-pressure chamber, the stabilizing chamber, and the output chamber are reset in the whole domain. Based on the new pressure threshold of the priority chamber indicated by the initial correction direction, the pressure difference balance compensation of the remaining chambers is calculated in reverse to reconstruct the pressure difference balance system of the three chambers. After calibration, collect steady-state pressure data of the calibrated gas supply port, buffer chamber connection node and output branch terminal, and compare them with the standard gas supply tolerance range; The difference between the steady-state pressure data of each monitoring node and the center value of the standard gas supply tolerance range is calculated as the residual pressure offset of each node, and the residual pressure offset of each node is compared with the half width of the standard gas supply tolerance range. If the residual pressure offset of all monitoring nodes falls back to the standard gas supply tolerance range, the calibration is deemed valid. The initial correction direction, deviation level and actual adjustment parameters are packaged to generate a calibration action label, which is used to update the disturbance source confidence weight in the feedback optimization label. If the residual pressure offset of any monitoring node does not converge to the standard gas supply tolerance range, a progressive calibration will be triggered while keeping the initial correction direction unchanged. The adjustment step size will be gradually increased until the residual pressure offset converges to the standard gas supply tolerance range, or the number of progressive calibrations reaches the preset number threshold. If, after the number of iterations reaches the preset threshold, there is still a residual pressure offset at any monitoring node that has not converged to the standard gas supply tolerance range, then the adaptive calibration is deemed to have failed.

[0030] In this embodiment, the adaptive calibration module further includes: Acquire steady-state pressure data and corresponding historical deviation data. The historical deviation data includes the pressure offset direction deviation sequence, amplitude deviation sequence, and response time deviation sequence recorded at each calibration progress node within the current calibration cycle. Align the historical deviation data with the time axis based on the time sequence and extract the deviation convergence trajectory features, including the convergence speed coefficient, the residual deviation steady-state mean, and the overshoot peak ratio. The deviation convergence trajectory features are associated with the perturbation source attribute labels to generate an association matrix between deviation and perturbation type. The deviation of the deviation convergence trajectory features from the preset standard convergence trajectory under each perturbation type is marked in the association matrix. A waveform template correction factor is generated based on the deviation. The waveform template correction factor includes the adjustment amount of the matching weight of each preset waveform template and the offset of the similarity judgment threshold. The matching weight of the preset waveform template is dynamically corrected based on the waveform template correction factor: for disturbance types whose deviation exceeds the preset deviation threshold, the matching weight of the corresponding waveform template is reduced, the matching weight of other similar waveform templates in the same category is increased, and the matching similarity judgment threshold is narrowed or widened accordingly.

[0031] In this embodiment, by comprehensively analyzing deviation values ​​and classifying deviation levels, the core basis for determining the initial correction direction based on the type of disturbance source is to directly link the calibration strategy with the inherent characteristics of the disturbance. Under each deviation level, the initial correction direction serves as the boundary constraint, and differentiated graded adjustments are performed, significantly improving the pertinence, directionality, and physical rationality of the calibration actions. Furthermore, the calibration effectiveness is determined by the residual pressure offset, achieving efficient approximation within a finite number of steps. Combined with the backtracking of calibration results for adaptive adjustments, the system continuously optimizes its disturbance identification capability during long-term operation, significantly improving the calibration accuracy, convergence efficiency, and long-term adaptability of the gas supply pressure stabilization system in multi-source disturbance scenarios.

[0032] In this embodiment, the graded abnormality diagnosis strategy includes primary diagnosis, secondary diagnosis, and tertiary diagnosis, wherein: Level 1 diagnostics performs self-checks on the gas supply pressure sensor, main gas path pressure sensor and pressure sensors at each key node to determine whether there is sensor zero drift, signal loss or over-range fault, and generates sensor health status codes. Secondary diagnosis: When the sensor is in normal health condition, extract the current pressure fluctuation spectrum features and match them with a preset typical fault mode library to identify whether there are mechanical faults such as leakage in the buffer unit chamber, stuck connecting valve or blocked output branch, and generate a mechanical fault probability vector. If there are no abnormalities in the first and second level diagnoses, the third level diagnosis is determined to be a deviation in the identification of gas path disturbance type or a mismatch of the prediction sub-model, and a model adaptation anomaly code is generated. Diagnostic procedures at each level are executed sequentially according to priority. When any level of diagnosis detects an abnormality, the subsequent diagnostic levels are terminated, and the result of the current level diagnosis is output as an abnormal diagnosis conclusion. The abnormal diagnosis conclusion includes the fault type identifier, fault confidence level, and fault handling strategy.

[0033] In this embodiment, a progressive diagnostic architecture that stops upon anomaly enables layered identification and rapid localization of issues ranging from sensor hardware failures and mechanical actuator malfunctions to algorithm model adaptation anomalies. This avoids computational redundancy caused by indiscriminate full-scale diagnosis. Furthermore, by combining fault type identification, confidence level, and handling strategies, a clear and operable basis is provided for subsequent maintenance decisions, significantly improving the fault diagnosis efficiency, localization accuracy, and targeted handling of the gas supply pressure stabilization system under abnormal operating conditions.

[0034] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring, characterized in that, include: The signal identification module is used to acquire pressure monitoring data in the gas supply source and main gas path in real time, perform trend fitting on the pressure monitoring data, extract the dynamic change characteristics of the gas path pressure, and identify the type of gas path disturbance by combining the preset waveform template. The disturbance prediction module is used to match the corresponding prediction sub-model to predict gas path pressure disturbances based on the gas path disturbance type and generate a prediction result dataset. The buffer throttling module is used to adjust the pressure distribution relationship between the chambers of the buffer unit in the main air path according to the prediction result dataset, and simultaneously adjust the air resistance characteristics on the output branch to obtain the actual pressure response data after adjustment. The adaptive calibration module is used to compare the adjusted actual pressure response data with the estimated values ​​of the expected pressure offset direction and amplitude in the prediction result dataset, calculate the deviation, and perform adaptive calibration based on the deviation. If the number of adaptive calibrations exceeds the preset threshold, the adaptive calibration is determined to have failed and a failure signal is generated. The anomaly diagnosis module is used to acquire failure signals and current gas path pressure monitoring data, and to perform anomaly diagnosis based on a graded anomaly diagnosis strategy. Based on the diagnosis results, it matches the corresponding fault type identifier, fault confidence level and fault handling strategy. The performance evaluation memory module is used to construct the regulation performance evaluation index for the current operating cycle, compare the regulation performance evaluation index with the historical cycle performance index, generate the performance degradation trend curve, and generate early warning signals based on the performance degradation trend curve.

2. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 1, characterized in that, The specific process by which the signal identification module identifies the type of airway disturbance includes: Based on the continuous acquisition of raw pressure monitoring data by the pressure sensing unit, the raw pressure monitoring data is standardized and preprocessed to generate a time-domain regularized pressure dataset. At the same time, a circular buffer including short time windows and long time windows is constructed. Based on the short time window, the corresponding short time data segment in the time-domain regularized pressure dataset is retrieved, the pressure sequence within the short time data segment is extracted and the instantaneous rate of change is calculated to obtain the effective pressure rate of change; The effective pressure change rate is compared with the preset aerodynamic sensitivity threshold. If the effective pressure change rate exceeds the preset aerodynamic sensitivity threshold, it is determined that the monitoring node may be disturbed, and the determination results of adjacent detection nodes are obtained at the same time. If it is determined that there is a possibility of disturbance in both adjacent detection nodes, then a valid disturbance event is determined to exist. Based on the start trigger time of the valid disturbance event, pressure waveform segments of a preset duration before and after the start trigger time are extracted and integrated to generate the trigger time waveform segment. Based on the long time window, the corresponding long time data segment is retrieved from the time domain regularized pressure dataset. The low-frequency steady-state component and the high-frequency impact component are separated in the long time data segment, and the long-period trend feature parameters are extracted. The peak impact amplitude and transient response slope of the waveform segment at the trigger moment are extracted as instantaneous impact features and fused with trend feature parameters to construct a multidimensional disturbance feature vector. The multidimensional disturbance feature vector is correlated and matched with the preset waveform template, and the corresponding gas path disturbance type is determined based on the correlation similarity.

3. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 2, characterized in that, The prediction sub-models include: impact response prediction sub-model, gas supply attenuation prediction sub-model, leakage compensation prediction sub-model and resonance suppression prediction sub-model, and each prediction sub-model corresponds one-to-one with the gas path disturbance type.

4. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 3, characterized in that, The specific process for predicting gas path pressure disturbances includes: Obtain the type label of the gas path disturbance and the effective pressure change rate at the current moment, and match the corresponding prediction sub-model according to the gas path disturbance type label; Extract the pressure sequence data and effective pressure change rate sequence data within a preset number of sampling periods before the current moment, and input them into the corresponding prediction sub-model to deduce the pressure change trajectory of each key node within a preset time window in the future, and draw the expected pressure response curve; Based on the expected pressure response curve, extract the curve feature parameters, and generate a prediction result dataset based on the curve feature parameters, including the disturbance prediction flag, expected pressure offset direction, amplitude estimate, expected disturbance arrival time, and disturbance duration period.

5. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 4, characterized in that, The disturbance prediction module also includes disturbance source type determination and feedback optimization: Obtain the disturbance prediction flag and the expected arrival time of the disturbance in the prediction result dataset. When the expected arrival time of the disturbance arrives, simultaneously collect the gas supply source pressure monitoring data, main gas path pressure monitoring data and output branch pressure monitoring data to generate a multi-node synchronous observation vector. The multi-node synchronous observation vector is correlated and compared with the estimated values ​​of the expected pressure offset direction and magnitude in the prediction result dataset to determine the actual occurrence node of the current disturbance event. Based on the pressure propagation time sequence of the actual occurrence node, the location of the disturbance source is traced back to generate the disturbance source type determination result. Based on the determination of the disturbance source type, a feedback optimization identifier is generated, and the length of the extrapolation time domain window of the prediction sub-model in the next sampling period and the initial pressure distribution weight of each chamber of the buffer unit are adjusted using the feedback optimization identifier.

6. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 5, characterized in that, The specific process of tracing the location of the disturbance source in reverse includes: Using the gas supply port as the first monitoring node and the output cavity port of the buffer unit as the second monitoring node, the pressure change time and pressure change rate sign of the first and second monitoring nodes at the expected disturbance arrival time are extracted respectively. Calculate the time difference between the pressure change time of the second monitoring node and the pressure change time of the first monitoring node, and use it as the disturbance propagation delay difference; If the difference in disturbance propagation delay is positive, the pressure change rate of the two monitoring nodes has the same sign, and it indicates that the pressure fluctuation on the gas supply side occurs earlier than that on the buffer unit side, then it is determined to be a disturbance on the gas supply side. If the difference in disturbance propagation delay is negative, or the sign of the rate of change of pressure at the gas supply port is opposite to the sign of the rate of change of pressure at the output chamber of the buffer unit, it indicates that the pressure fluctuation on the gas supply side occurs later than that on the buffer unit side or the fluctuation directions are different, and it is determined to be a disturbance on the gas supply side of the terminal. If the disturbance propagation delay difference is zero or the conditions for determining the sign of the corresponding pressure change rate are not met, then it is determined to be a pipeline structure resonance type. The determined disturbance source type is labeled as a disturbance source attribute tag, which serves as the disturbance source type determination result.

7. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 6, characterized in that, The specific process of adaptive calibration based on deviation includes: The deviation is obtained, including pressure offset direction deviation, amplitude deviation and response time deviation. The deviation values ​​of each dimension in the deviation are normalized and weighted to generate a comprehensive deviation value. The overall deviation value is compared with the preset deviation calibration threshold to determine the current deviation level, and the corresponding gear calibration strategy is matched based on the current deviation level. Before executing the gear calibration strategy, the initial correction direction under the current deviation level is determined according to the type of disturbance source, and the corresponding gear calibration strategy is executed according to the determined initial correction direction. After calibration, collect steady-state pressure data of the calibrated gas supply port, buffer chamber connection node and output branch terminal, and compare them with the standard gas supply tolerance range; The difference between the steady-state pressure data of each monitoring node and the center value of the standard gas supply tolerance range is calculated as the residual pressure offset of each node, and the residual pressure offset of each node is compared with the half width of the standard gas supply tolerance range. If the residual pressure offset of all monitoring nodes falls back to the standard gas supply tolerance range, the calibration is deemed valid, and the initial correction direction, deviation level, and actual adjustment parameters are packaged to generate a calibration action label. If the residual pressure offset of any monitoring node does not converge to the standard gas supply tolerance range, a progressive calibration will be triggered while keeping the initial correction direction unchanged. The adjustment step size will be gradually increased until the residual pressure offset converges to the standard gas supply tolerance range, or the number of progressive calibrations reaches the preset number threshold. If, after the number of iterations reaches the preset threshold, there is still a residual pressure offset at any monitoring node that has not converged to the standard gas supply tolerance range, then the adaptive calibration is deemed to have failed.

8. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 7, characterized in that, The adaptive calibration module further includes: Acquire steady-state pressure data and corresponding historical deviation data, align the historical deviation data with the time axis based on time order, and extract the deviation convergence trajectory features; The deviation convergence trajectory features are associated with the perturbation source attribute labels to generate an association matrix between deviation and perturbation type. The deviation of the deviation convergence trajectory features from the preset standard convergence trajectory under each perturbation type is marked in the association matrix. A waveform template correction factor is generated based on the deviation, and the matching weight of the preset waveform template is dynamically corrected based on the waveform template correction factor.

9. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 1, characterized in that, The hierarchical anomaly diagnosis strategy includes primary diagnosis, secondary diagnosis and tertiary diagnosis. Each level of diagnosis is executed in order of priority. When any primary diagnosis detects an anomaly, the subsequent diagnosis level is terminated and the result of the current level diagnosis is output as the anomaly diagnosis conclusion.

10. The pressure-stabilizing system for a dental treatment machine with real-time online pressure monitoring as described in claim 1, characterized in that, The performance evaluation memory module further includes: taking the current operating cycle as the evaluation unit, at the end of each cycle, using all calibration action tags in the current cycle as the data source, calculating the cycle mean and cycle variance of deviation convergence speed, steady-state holding accuracy, calibration action amplitude, and overshoot, and generating the performance evaluation index vector for the current cycle; storing the performance evaluation index vector for the current cycle sequentially into the historical performance database based on the corresponding timestamp, updating the data points for the current cycle in the performance degradation trend curve, and when the warning signal is generated, packaging the warning signal and the performance evaluation index vector for the current cycle and pushing them to the adaptive calibration module as the basis for dynamic adjustment of the preset number threshold.

Images