Method and system for online monitoring of multi-instrument status in dental chair

Abstract

This application relates to the field of condition monitoring technology, and in particular to an online monitoring method and system for the status of multiple instruments in a dental chair. The method includes: clustering historical operating data of the monitoring equipment in its initial healthy state to construct a dynamic health model, which includes a health centroid and a health radius; calculating the deviation distance between the real-time operating data of the monitoring equipment and the health centroid; in response to a deviation distance not exceeding the health radius, determining the real-time operating data as a healthy point and adaptively updating the health centroid and health radius; otherwise, determining the real-time operating data as an abnormal point; statistically analyzing the frequency of abnormal points within a preset time window, and outputting a fault warning message when the frequency exceeds a warning threshold. The technical solution of this application can effectively reduce the false alarm rate and achieve accurate monitoring of the operating status of each instrument in the dental chair.

Description

Technical Field

[0001] This application relates to the field of condition monitoring technology, and in particular to an online monitoring method and system for the condition of multiple instruments in a dental chair. Background Technology

[0002] As a sophisticated and complex medical device, the dental chair integrates multiple instruments such as a lifting system and an ultrasonic scaler. The stability of its operation directly affects the safety and efficiency of treatment. Therefore, real-time status monitoring of the dental chair's instruments and providing accurate fault warnings are of great significance.

[0003] Currently, patent application CN111243722A discloses a method and system for real-time monitoring of various states during the operation of a dental chair. The method includes: setting several control nodes in the control function modules of the dental chair to acquire the working status of the control function modules and the working status of external accessories on each control function module; setting several detection nodes in the water and gas supply modules of the dental chair to determine whether the water and gas supply is normal; setting a coordinator motherboard on the dental chair, which is connected to each control function module and each detection node via a Zigbee network, and transmits the collected dental chair data to a cloud server via a WIFI module or GPRS module; and accessing the cloud server through a smart terminal to obtain the processed dental chair data, which includes MAC address, device number, device operating status record, device operating time record, device location, device configuration information, and fault description data.

[0004] The above method directly sets up several detection nodes in the water and air supply module of the dental chair and uses the detection nodes to determine whether the water and air supply module is supplying normally. However, the above method can only issue an early warning when the water and air supply module is not working properly, and does not take into account the impact of normal wear and tear and aging caused by long-term operation of the equipment on the monitoring results of the operating status, resulting in inaccurate monitoring results of the operating status of various instruments in the dental chair. Summary of the Invention

[0005] To address the technical problem of inaccurate monitoring results of the operating status of various instruments in a dental chair, this application provides an online monitoring method and system for the status of multiple instruments in a dental chair, which can effectively reduce the false alarm rate and achieve accurate monitoring of the operating status of various instruments in a dental chair.

[0006] In a first aspect, this application provides an online monitoring method for the status of multiple instruments in a dental chair. The monitoring method includes: clustering historical operating data of the monitoring device in its initial healthy state; constructing a dynamic health model in the cluster with the highest density; the dynamic health model includes a health centroid and a health radius; the monitoring device is any instrument in the dental chair; calculating the deviation distance between the real-time operating data of the monitoring device and the health centroid; in response to the deviation distance not being greater than the health radius, determining the real-time operating data as a healthy point, and adaptively updating the health centroid and health radius based on the healthy point; otherwise, determining the real-time operating data as an abnormal point, and marking the current moment as an abnormal point; statistically analyzing the frequency of occurrence of abnormal points within a preset time window, and outputting fault warning information when the frequency of occurrence exceeds a warning threshold.

[0007] By constructing a dynamic health model including a health centroid and a health radius in the densest cluster, and adaptively updating the model using real-time operating data identified as healthy points, the monitoring model can adapt to the health baseline drift caused by normal equipment aging. At the same time, by statistically analyzing the frequency of abnormal points within a preset time window to trigger fault warnings, instantaneous disturbances are effectively filtered out, achieving accurate monitoring of the operating status of the monitoring equipment.

[0008] Preferably, constructing the dynamic health model in the cluster with the highest density includes: performing DBSCAN clustering on historical operating condition data under the initial health state, and taking the cluster with the highest density as the core cluster of the initial health state; calculating the arithmetic mean of all historical operating condition data within the core cluster to obtain the health centroid in the dynamic health model; and calculating the average distance from all historical operating condition data within the core cluster to the health centroid to obtain the health radius of the dynamic health model.

[0009] Preferably, before clustering the historical operating condition data of the monitoring equipment in its initial health state, the monitoring method further includes: preprocessing the historical operating condition data, wherein the preprocessing includes standardization and smoothing filtering.

[0010] Preferably, the deviation between real-time operating data and the healthy centroid Satisfying Relationship:

[0011] ;in, The total number of dimensions of the operating condition data. For the first Preset weights for maintenance condition data and Real-time operating data and health quality The Middle The values ​​of maintenance condition data.

[0012] When calculating the deviation distance, preset weights are assigned to the operating condition data of different dimensions, so that the dimensions with a greater impact will contribute more to the deviation distance, thus realizing the accurate measurement of the differences in the operating status of the monitoring equipment.

[0013] Preferably, the first The method for obtaining the preset weights of the working condition data includes: taking the first weight from the historical working condition data... After deleting the maintenance condition data, the historical maintenance condition data under the initial healthy state are clustered again to obtain the simulated centroid. The deviation between the simulated centroid and the healthy centroid is used as the first centroid. The degree of impact of maintenance condition data; The preset weight of the maintenance condition data is the first The ratio of the degree of influence of the maintenance condition data to the sum of the degree of influence of all maintenance condition data.

[0014] The first in the historical working condition data After deleting the maintenance condition data, the deviation distance between the newly generated simulated centroid and the original healthy centroid is calculated, and this deviation is used as the first... The impact of each dimension's operating condition data is determined, and its weight is then established. This objectively and accurately assesses the importance of each dimension's operating condition data in defining the health status and enables the automated acquisition of preset weights for each dimension.

[0015] Preferably, the adaptive update of the healthy centroid includes: calculating the ratio of the deviation distance to the healthy radius, using the difference between 1 and the ratio as the confidence level of the real-time operating data; and using the product of the confidence level and the base learning rate as the update speed to update the healthy centroid.

[0016] A differentiated update rate was achieved, ensuring that the movement of the health centroid is mainly driven by high-confidence operating data located in the core region. This effectively prevents the contamination of the dynamic health model by noisy data with low confidence located at the boundary, thus guaranteeing the stability and accuracy of the adaptive update of the health centroid.

[0017] Preferably, the adaptive update of the health radius includes: dividing the cluster with the highest density into multiple radius intervals, and taking the average value of all historical operating condition data within the radius interval as the operating condition center of the radius interval; locating the radius interval to which the real-time operating condition data belongs based on the deviation distance between the real-time operating condition data and the health centroid, and calculating the similarity between the real-time operating condition data and the operating condition center of the corresponding radius interval; responding to a similarity greater than a similarity threshold, not updating the health radius; otherwise, taking the difference between the average radius of each radius interval and the corresponding radius interval as a sub-update quantity, taking the similarity between the real-time operating condition data and the operating condition center as the update weight of each radius interval, weighting and summing each sub-update quantity according to the update weight, taking the maximum value between the weighted summation result and 0 as the radius update quantity, and taking the sum of the health radius and the radius update quantity as the updated health radius.

[0018] By dividing the densest cluster into multiple radius intervals and setting a similarity threshold to trigger the update of the health radius, updates are only performed when there is a significant shift in the data distribution within the cluster, indicating that the fluctuation range has changed due to equipment aging. At the same time, the sub-update amounts are weighted and summed based on the similarity of the operating condition centers of each radius interval to determine the radius update amount, so that the updated health radius truly reflects the long-term trend of the normal fluctuation range of the monitoring equipment.

[0019] Preferably, the updated health radius Satisfying the relation:

[0020] ;

[0021] in, The health radius before the update. The number of radius intervals, For real-time operating condition data and radius interval Operating Condition Center similarity, It is the sum of the similarity between real-time operating condition data and the operating condition centers of all radius intervals; and Radius intervals The average radius of the radius interval to which the real-time operating data belongs.

[0022] Preferably, dividing the cluster with the highest density into multiple radius intervals includes: dividing the healthy radius at equal intervals to obtain multiple dividing points; with the healthy centroid as the center, the area between adjacent dividing points is a radius interval.

[0023] In a second aspect, this application also provides an online monitoring system for the status of multiple instruments in a dental chair, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the online monitoring method for the status of multiple instruments in a dental chair according to the first aspect of this application.

[0024] The technical solution of this application has the following beneficial technical effects:

[0025] By clustering historical operating data of the initial health status of monitoring equipment, a dynamic health model including a health centroid and a health radius is constructed. Then, by calculating the deviation distance between real-time operating data and the health centroid, data points are determined to be healthy or abnormal. Furthermore, the health centroid and health radius are adaptively updated using real-time operating data identified as healthy points, enabling the dynamic health model to proactively learn and adapt to the slow changes in the health baseline caused by normal wear and tear or aging of the monitoring equipment, effectively avoiding false alarms caused by benign equipment drift. Simultaneously, by statistically analyzing the frequency of abnormal points within a preset time window, and only outputting fault warning information when the frequency exceeds a warning threshold, warnings are avoided from being triggered by a single abnormal point caused by instantaneous disturbances or random interference, thus significantly reducing the false alarm rate of fault warnings and achieving accurate monitoring of the operating status of various instruments in the dental chair. Attached Figure Description

[0026] Figure 1 This is a flowchart of an online monitoring method for the status of multiple instruments in a dental chair according to an embodiment of this application.

[0027] Figure 2 This is a comparison chart of the false alarm rates under the dynamic health model and the static health model according to the embodiments of this application.

[0028] Figure 3 This is a structural block diagram of an online monitoring system for the status of multiple instruments in a dental chair according to an embodiment of this application. Detailed Implementation

[0029] According to a first aspect of this application, this application provides an online monitoring method for the status of multiple instruments in a dental chair. The monitoring device can be any instrument in the dental chair, such as the chair's lifting system, ultrasonic scaler, water pump, etc., and can obtain the operating status of any instrument at any time, thereby achieving accurate fault prediction for each instrument.

[0030] Figure 1 This is a flowchart of an online monitoring method for the status of multiple instruments in a dental chair according to an embodiment of this application. Figure 1 As shown, the online monitoring method for the status of multiple instruments in a dental chair includes steps S101 to S104, which are described in detail below.

[0031] S101, cluster the historical operating data of the monitoring device in its initial health state, and construct a dynamic health model in the cluster with the highest density. The dynamic health model includes a health centroid and a health radius. The monitoring device is any instrument in a dental chair.

[0032] In one embodiment, historical operating condition data under the initial health state is collected. The initial health state can be a brand new dental chair that has not yet been put into use, or a dental chair that is known to be in a normal state. The historical operating condition data is multi-dimensional data closely related to the monitoring equipment. Taking the lifting system of the dental chair as an example, the historical operating condition data includes the current sequence of the drive motor and the vibration sequence of the vibration sensor during the lifting process.

[0033] Before clustering the historical operating condition data of the monitoring equipment in its initial health state, the monitoring method further includes: preprocessing the historical operating condition data, which includes standardization and smoothing filtering. Specifically, the purpose of standardization is to eliminate differences in different physical dimensions, and "maximum-minimum" standardization can be used to map the data to the [0,1] interval. For data of any dimension, the formula for standardization can be expressed as: ;

[0034] in, This is the original data. and These are the minimum and maximum values ​​for this dimension of data across all historical operating conditions, respectively. It should be noted that if... equal In this case, all data in that dimension All were assigned the value 0.5.

[0035] The purpose of smoothing filtering is to remove high-frequency noise from historical operating data. A moving average filter can be used to smooth the time series data points. For example, a sliding window with a width of 5 data points can be used, and the mean value within the window can be used as the center point after filtering to achieve the purpose of noise reduction.

[0036] After preprocessing, the historical operating data of the monitoring equipment in its initial health state can be clustered to construct a dynamic health model. Specifically, constructing the dynamic health model in the cluster with the highest density includes: performing DBSCAN clustering on the historical operating data in the initial health state, and taking the cluster with the highest density as the core cluster of the initial health state; calculating the arithmetic mean of all historical operating data within the core cluster to obtain the health centroid in the dynamic health model; and calculating the average distance from all historical operating data within the core cluster to the health centroid to obtain the health radius of the dynamic health model.

[0037] The DBSCAN algorithm does not require a preset number of clusters and can determine the cluster size based on a set neighborhood radius. and minimum sample size The number of clusters is automatically determined based on the distribution of historical operating data; in this embodiment, the neighborhood radius can be set. The minimum sample size is 0.5. It is 10.

[0038] Healthy Mind It can be calculated using the following formula:

[0039] ;

[0040] in, As the core cluster, This represents the total number of historical operating condition data within the core cluster. For the first in the core cluster Historical operating condition data. Initial healthy center of mass. It characterizes the most stable operating conditions of the monitoring equipment under normal operating conditions.

[0041] Health Radius It can be calculated using the following formula:

[0042] ;

[0043] in, For the first in the core cluster Historical operating condition data To the initial health center The deviation distance. The range of normal fluctuations for monitoring equipment under normal operating conditions is defined. Understandably, in order to achieve accurate monitoring of operating conditions and avoid missed detections, after calculating the average distance, a correction factor less than 1 can be multiplied to appropriately reduce the healthy radius; for example, the correction factor can be 0.9.

[0044] Furthermore, since different operating conditions have varying degrees of impact on the monitoring equipment's operational status, in order to more accurately measure the deviation distance and ensure that it more precisely reflects the differences in operational status, it is also necessary to determine the preset weights for each dimension of operating condition data. Specifically, the first... The method for obtaining the preset weights of the working condition data includes: taking the first weight from the historical working condition data... After deleting the maintenance condition data, the historical maintenance condition data under the initial healthy state are clustered again to obtain the simulated centroid. The deviation between the simulated centroid and the healthy centroid is used as the first centroid. The degree of impact of maintenance condition data; The preset weight of the maintenance condition data is the first The ratio of the degree of influence of the maintenance condition data to the sum of the degree of influence of all maintenance condition data.

[0045] Assuming the healthy center of mass is When the first After obtaining the dimensional data, the remaining dimensions are re-clustered to obtain the simulated centroid. Calculating the health center of mass and simulated centroid Deviation distance This is the first The impact of maintenance condition data If the first one is removed... The working condition data caused a large drift distance of the centroid, i.e., the first... The impact of maintenance condition data It is very large, indicating that the first Maintenance condition data is crucial for defining health status; Preset weights for maintenance condition data The calculation is as follows:

[0046] ;in, This represents the sum of the impact of data under each operating condition.

[0047] In this way, each monitoring device builds a personalized dynamic health model, laying the data foundation for subsequent online adaptive monitoring.

[0048] S102, calculate the real-time operating data of the monitoring equipment and the deviation distance of the health centroid.

[0049] In one embodiment, after the monitoring equipment is put into use, it enters the online monitoring and adaptive learning phase. For newly collected real-time operating data... First, real-time operating data Perform the same standardization process, and then calculate the real-time operating data. Compared to a healthy mindset The deviation distance.

[0050] Understandably, the initial value of a healthy center of mass is... Because the health center is dynamically updated, therefore, the health center here... To collect real-time operating condition data The corresponding health mindset at that time.

[0051] Deviation between real-time operating data and the healthy center of mass Satisfying Relationship:

[0052] ;in, The total number of dimensions of the operating condition data. For the first Preset weights for maintenance condition data and Real-time operating data and health quality The Middle The values ​​of maintenance condition data.

[0053] S103, in response to the deviation distance not being greater than the healthy radius, if the real-time operating data is determined to be a healthy point, then the healthy centroid and healthy radius are adaptively updated based on the healthy point; otherwise, if the deviation distance is not greater than the healthy radius, the real-time operating data is determined to be an abnormal point, and the current time is marked as an abnormal point.

[0054] In one embodiment, after calculating the deviation distance... Then, compare it with the health radius. Comparisons are made. Understandably, the initial value of the health radius is... Since the health radius is also dynamically updated, the health radius here... To collect real-time operating condition data The corresponding health radius at that time.

[0055] If the deviation distance is not greater than the health radius, and the real-time operating data is determined to be a healthy point, then the health centroid and health radius are adaptively updated based on the healthy point. This endows the dynamic health model with active learning capabilities, enabling it to adapt to slowly occurring health baseline changes caused by normal wear and tear. If the deviation distance is greater than the health radius, indicating that the data point has fallen outside the health range defined by the dynamic health model, it may be caused by abnormal equipment operating conditions, sudden failures, or strong environmental interference. The current moment is marked as an abnormal point, and these abnormal points will be used for fault frequency statistics in step S104.

[0056] As a preferred approach, the adaptive update of the healthy centroid includes: calculating the ratio of the deviation distance to the healthy radius, using the difference between 1 and the ratio as the confidence level of the real-time operating data; and using the product of the confidence level and the base learning rate as the update rate to update the healthy centroid.

[0057] Specifically, the confidence level of real-time operating condition data Satisfying the relation:

[0058] ;

[0059] in, The deviation between real-time operating data and the healthy center of mass. For the healthy radius; then the update speed. for: ;in, The base learning rate is a preset constant, which is set to 0.01 in this embodiment, and is used to control the base rate of centroid drift. It should be noted that, generally, the healthy radius... It is a non-zero value, to avoid the health radius A value of 0 results in a confidence level Unable to calculate, a health radius can be set in this embodiment. The minimum value, which is greater than 0.

[0060] The process of updating the aforementioned health center of mass is as follows:

[0061] ;

[0062] in, For the updated health center, For the previous health center, This is real-time operating condition data.

[0063] When real-time operating data When it happened to fall exactly on the health center before the update, , At this point, the update speed is at its maximum, and the health centroid moves towards this real-time operating condition data. The willingness to learn is strongest; when real-time operating data... When it falls on the edge of health, , At this point, the update speed is 0, and the real-time operating condition data... It does not participate in the updating of the health centroid at all; thus, it ensures that the movement of the health centroid is mainly driven by the operating condition data located in the core region (i.e., high confidence), while the operating condition data located at the boundary position (i.e., low confidence) will not pull on the health centroid, effectively preventing boundary noise data from contaminating the dynamic health model and ensuring the stability and accuracy of the adaptive update of the health centroid.

[0064] To enable the model to adapt to changes in the normal fluctuation range caused by equipment aging, the health radius also needs to be adaptively updated. As a preferred approach, the adaptive update of the health radius includes: dividing the highest-density cluster into multiple radius intervals, and using the average value of all historical operating condition data within each radius interval as the operating condition center of that interval; locating the radius interval to which the real-time operating condition data belongs based on the deviation distance between the real-time operating condition data and the health centroid, and calculating the similarity between the real-time operating condition data and the operating condition center of its respective radius interval; if the similarity is greater than a similarity threshold, the health radius is not updated; otherwise, the difference between the average radius of each radius interval and its respective radius interval is used as a sub-update quantity, and the similarity between the real-time operating condition data and the operating condition center is used as the update weight for each radius interval; the sub-update quantities are weighted and summed according to the update weights, and the maximum value between the weighted sum and 0 is used as the radius update quantity; the sum of the health radius and the radius update quantity is used as the updated health radius.

[0065] The process of dividing the cluster with the highest density into multiple radius intervals includes: dividing the healthy radius at equal intervals to obtain multiple dividing points; with the healthy centroid as the center, the area between adjacent dividing points is a radius interval.

[0066] For example, if the health radius is a value It can be divided into equal intervals. There are several radius intervals; radius interval 1 is centered on the health centroid, with a radius of... The average radius of the inner circular region, radius interval 1 is Radius interval 2 is centered on the health centroid, with a radius of... The average radius of the inner annular region, radius interval 2 is Thus, the average radii of radius interval 3 and radius interval 4 are respectively: and Then, within the cluster with the highest density, the average value of all historical operating data falling within each radius interval is calculated to obtain the operating center of each radius interval.

[0067] The similarity between real-time operating condition data and the operating condition center within its corresponding radius is negatively correlated with the deviation distance; the greater the deviation distance, the smaller the similarity between the real-time operating condition data and the operating condition center within its corresponding radius. Real-time operating condition data and radius interval Operating Condition Center similarity Satisfying the relation:

[0068] ;

[0069] in, For real-time operating condition data and radius interval Operating Condition Center The similarity threshold is set at 0.9. If the similarity is greater than the threshold, the real-time operating data is considered a typical healthy point within its radius range, indicating that the boundary has no expansion trend, and therefore the healthy radius is not updated. Conversely, if the similarity is low, it indicates that the data distribution has shifted within the densest cluster, and the aging of the monitoring equipment has caused changes in the normal fluctuation range, thus triggering an update of the healthy radius.

[0070] Specifically, the updated health radius Satisfying the relation:

[0071] ;

[0072] in, The health radius before the update. The number of radius intervals, For real-time operating condition data and radius interval Operating Condition Center similarity, It is the sum of the similarity between real-time operating condition data and the operating condition centers of all radius intervals; and Radius intervals The average radius of the radius interval to which the real-time operating data belongs.

[0073] in, This is the sub-update quantity, representing real-time operating condition data. Average radius level of the radius interval and The difference in radius between them; It updates the weights, representing real-time operating condition data. and The degree of similarity; the maximum value function This ensures that the health radius does not shrink, conforms to the characteristics of equipment aging, and avoids false detections caused by unreasonable shrinkage of the health radius; it ensures that the update of the health radius is based on the density distribution characteristics of data points within the core cluster, and can truly reflect the long-term trend of the normal fluctuation range of the monitoring equipment.

[0074] In this way, by distinguishing between healthy and abnormal points in real-time operating data, and using healthy points to update the health centroid and health radius in the dynamic health model, continuous tracking of the health status of the monitoring equipment is achieved. This not only adapts to the benign drift caused by the natural aging of the equipment, but also effectively prevents abnormal data from contaminating the model.

[0075] It should be noted that the real-time monitoring of the operating status of the subsequent monitoring equipment is carried out on the updated dynamic health model.

[0076] S104: Calculate the frequency of occurrence of abnormal points within a preset time window, and output fault warning information when the frequency of occurrence exceeds the warning threshold.

[0077] In one embodiment, a single anomaly may be caused by random disturbances, which is insufficient to determine a fault. Therefore, a preset time window is used for statistical analysis to count the frequency of anomalies within the preset time window. Only when the frequency of anomalies exceeds a warning threshold is the monitoring equipment determined to be faulty, and a fault warning message is output.

[0078] The warning threshold is used to distinguish between random disturbances of anomalies and actual faults of monitoring equipment. Therefore, the warning threshold can be set according to the tolerance of the monitoring equipment for false alarm rate. In this embodiment, the warning threshold is set to 0.1. When the frequency of anomalies in the preset time window is greater than the warning threshold, it means that the anomalies generated by the monitoring equipment in the preset time window have exceeded the tolerance for false alarm rate, eliminating the possibility of random disturbances and determining that the monitoring equipment has an operational fault. The preset time window is a time period of 30 moments with the current moment as the end point.

[0079] In this way, by statistically analyzing the frequency of anomalies within a preset time window, accurate fault diagnosis is achieved, effectively reducing the false alarm rate caused by transient disturbances, and enabling real-time monitoring of the operating status of various instruments in the dental chair. Please refer to [link / reference needed]. Figure 2 The figure shows a comparison of the false alarm rates under the dynamic health model and the static health model according to the embodiments of this application. As can be seen from the figure, the dynamic health model can keep the fault warning at a stable and low false alarm rate, ensuring the accuracy of the monitoring results of the operating status of the monitoring equipment.

[0080] According to a second aspect of this application, this application also provides an online monitoring system for the status of multiple instruments in a dental chair. Figure 3 This is a structural block diagram of an online monitoring system for the status of multiple instruments in a dental chair, according to an embodiment of this application. Figure 3 As shown, the system 50 includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an online monitoring method for the status of multiple instruments in a dental chair according to the first aspect of this application. The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their configuration and functions are known in the art and will not be described further here.

[0081] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of this application, and these all fall within the scope of protection of this application.

Claims

1. A method for online monitoring of multi-instrument status in a dental chair, characterized by, The monitoring method comprises: clustering historical working condition data of the monitoring device in an initial healthy state, constructing a dynamic health model in the cluster with the highest density, the dynamic health model comprising a health centroid and a health radius, the monitoring device being any instrument in a dental chair; calculating a deviation distance of real-time working condition data of the monitoring device from the health centroid; in response to the deviation distance being not greater than the health radius, determining that the real-time working condition data is a healthy point, and then performing adaptive updating of the health centroid and the health radius according to the healthy point, otherwise, determining that the real-time working condition data is an abnormal point, and marking the current time as an abnormal point; the adaptive updating of the health radius comprises: dividing the cluster with the highest density into a plurality of radius intervals, and taking the average of all historical working condition data in the radius interval as a working condition center of the radius interval; locating the radius interval to which the real-time working condition data belongs according to the deviation distance of the real-time working condition data from the health centroid, and calculating the similarity between the real-time working condition data and the working condition center of the radius interval; in response to the similarity being greater than a similarity threshold, the health radius is not updated, otherwise, the difference between the average radius of each radius interval and the radius interval to which it belongs is taken as a sub-update amount, the similarity between the real-time working condition data and the working condition center is taken as an update weight of each radius interval, each sub-update amount is weighted and summed according to the update weight, the maximum value between the weighted sum result and 0 is taken as a radius update amount, and the sum of the health radius and the radius update amount is taken as an updated health radius; statistically counting the frequency of the abnormal point in a preset time window, and outputting a fault warning information when the frequency is greater than a warning threshold.

2. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, The constructing of the dynamic health model in the cluster with the highest density comprises: DBSCAN clustering of the historical working condition data in the initial healthy state, and taking the cluster with the highest density as a core cluster of the initial healthy state; calculating the arithmetic mean of all historical working condition data in the core cluster to obtain the health centroid in the dynamic health model; and calculating the average distance from all historical working condition data in the core cluster to the health centroid to obtain the health radius of the dynamic health model.

3. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, Before clustering the historical working condition data of the monitoring device in the initial healthy state, the monitoring method further comprises: preprocessing the historical working condition data, the preprocessing comprising standardization processing and smoothing filtering.

4. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, Real-time operating data and deviation distance from health centroid Satisfies relationship: ; wherein, is the total number of dimensions of the working condition data, is the preset weight of the i-th dimensional working condition data, and are the values of the i-th dimensional working condition data in the real-time working condition data and the health centroid, respectively.

5. A method of on-line monitoring of status of multiple instruments in a dental chair as claimed in claim 4 wherein, No. The method for obtaining the preset weights of the working condition data includes: taking the first weight from the historical working condition data... After deleting the maintenance condition data, the historical maintenance condition data under the initial healthy state are clustered again to obtain the simulated centroid. The deviation between the simulated centroid and the healthy centroid is used as the first centroid. The degree of impact of maintenance condition data; The preset weight of the maintenance condition data is the first The ratio of the degree of influence of the maintenance condition data to the sum of the degree of influence of all maintenance condition data.

6. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, The adaptive updating of the health centroid comprises: calculating the ratio of the deviation distance to the health radius, taking the difference between 1 and the ratio as the confidence of the real-time working condition data, and taking the product of the confidence and a basic learning rate as an update speed to update the health centroid.

7. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, Updated health radius satisfies the relationship ； wherein, is the health radius before the update, is the number of radius intervals, is the real-time operating condition data and radius interval operating condition center similarity, is the sum of the similarity of the real-time operating condition data and all radius interval operating condition centers; and is the average radius of radius interval and the radius interval to which the real-time operating condition data belongs.

8. The method of online monitoring of status of multiple instruments in a dental chair as claimed in claim 1 wherein, The dividing of the cluster with the highest density into a plurality of radius intervals comprises: equally dividing the health radius to obtain a plurality of division points; and taking the health centroid as the center and the region between adjacent division points as a radius interval.

9. An online monitoring system of multi-instrument status in dental chair, characterized in that, The monitoring method comprises a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement an online monitoring method of a plurality of instrument states in a dental chair according to any one of claims 1 to 8.

Images