System for detecting hot plug state of light guide bundle of cold light source of endoscope

Abstract

The invention relates to the technical field of medical equipment monitoring and fault early warning, in particular to an endoscope cold light source light guide bundle hot plug state detection system, which comprises a data acquisition module used for acquiring a current noise time sequence signal, an interface core area temperature time sequence signal and a luminous flux time sequence signal of a connection interface in real time so as to generate a multi-physical field signal; the feature extraction module is used for generating a feature vector set; wherein the feature vector set comprises an electrical noise feature vector, a thermal instability feature vector and an optical coupling jitter feature vector; the risk assessment module generates a risk feature vector set; the health degree calculation module is used for calculating through a linear weighting model to generate a connection health degree index; the risk early warning module is used for generating a risk level; according to the invention, continuous tracking and quantitative evaluation of the connection state are realized, and the reliability and safety of the endoscope system are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of medical equipment monitoring and fault early warning technology, specifically to an endoscope cold light source beam hot-swap status detection system. Background Technology

[0002] The continuous development of modern network architecture has significantly increased the complexity of optical cable connections, especially in terms of manual management and monitoring. Currently, technicians typically maintain optical cables through regular on-site inspections, using tools such as microscopes or telescopes to examine their physical condition and record detailed inspection and maintenance information. However, this traditional method, which relies on manual physical inspection and testing, is inefficient and makes it difficult to detect problems in real time. Therefore, how to detect optical cable problems in a timely manner has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0003] The purpose of this invention is to provide an endoscope cold light source beam hot-plugging status detection system, aiming to overcome the limitations of existing technologies that can only perform post-fault diagnosis, and to provide a solution that can predict the trend of connection status deterioration and provide graded early warning. Specifically, the technical solution of this invention is as follows:

[0004] The data acquisition module is used to acquire the current noise timing signal, the temperature timing signal of the core area of ​​the interface, and the light flux timing signal in real time to generate multiphysics field signals.

[0005] The feature extraction module is used to generate a set of feature vectors based on the multi-physics field signals generated by the data acquisition module; the set of feature vectors includes electrical noise feature vectors, thermal instability feature vectors, and optical coupling jitter feature vectors.

[0006] The risk assessment module is used to normalize the feature vector set generated by the feature extraction module, combined with the preset ideal benchmark value and critical failure threshold, in order to generate a risk feature vector set.

[0007] The health score calculation module is used to calculate the connectivity health score index based on the risk feature vector set generated by the risk assessment module and combined with preset weight coefficients through a linear weighting model.

[0008] The risk warning module is used to compare the connection health index generated by the health calculation module with the preset attention warning threshold and danger warning threshold to generate a risk level.

[0009] Preferably, the feature extraction module generates an electrical noise feature vector, including:

[0010] Perform a Fourier transform on the current noise time-series signal to obtain the power spectral density function;

[0011] The power spectral density function is integrated over a preset frequency range to generate an electrical noise feature vector.

[0012] Preferably, the feature extraction module generates a thermal instability feature vector, including:

[0013] The first derivative of the temperature time series signal in the core area of ​​the interface with respect to time is calculated to generate the derivative value;

[0014] The derivative value is determined as the thermal instability eigenvector.

[0015] Preferably, the feature extraction module generates an optically coupled jitter feature vector, including:

[0016] Multiple luminous flux sample values ​​are collected within a preset sliding time window;

[0017] Calculate the standard deviation of all luminous flux sample values ​​within the sliding time window to generate the standard deviation value;

[0018] The standard deviation was determined as the optical coupling jitter feature vector.

[0019] Preferably, the risk assessment module generates a set of risk feature vectors, including:

[0020] The difference between any feature vector in the feature vector set and the ideal benchmark value corresponding to that vector is divided by the difference between the critical failure threshold corresponding to that vector and the ideal benchmark value to obtain a linear mapping result.

[0021] The linear mapping result is constrained within a preset closed interval to generate a normalized risk feature vector.

[0022] Preferably, the health calculation module generates a connection health index, including:

[0023] The comprehensive damage degree is obtained by multiplying each risk feature vector in the risk feature vector set with its corresponding preset weight coefficient and then summing the results.

[0024] The connection health index is generated by subtracting the overall damage level from the preset maximum health level.

[0025] Preferably, the risk levels generated by the risk warning module include:

[0026] When the connection health index is greater than the warning threshold, it is determined to be at a safe level;

[0027] When the connection health index is less than or equal to the attention warning threshold, but greater than the danger warning threshold, it is determined to be at the attention level.

[0028] When the connection health index is less than or equal to the danger warning threshold, it is determined to be at a danger level.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1. Achieve continuous tracking and quantitative evaluation of connection status: By deeply analyzing and integrating coupling signals from multiple dimensions such as electricity, heat, and light, the system can continuously track and quantitatively evaluate the entire process of connection status from health to failure. This method can identify early signs of connection deterioration and provide risk classification warnings with clear physical meaning and probability support, providing valuable decision-making time for equipment maintenance and operation, and significantly improving the reliability and safety of the endoscope system.

[0031] 2. Detecting thermal runaway trends earlier: By directly calculating the first derivative of the interface temperature with respect to time, this system can detect thermal runaway trends earlier than traditional fixed temperature threshold alarm methods. This method is not sensitive to absolute temperature, but is highly sensitive to the rate of acceleration of temperature change, and can provide more timely warnings before catastrophic overheating occurs.

[0032] 3. Precise quantification of optical signal jitter or instability: By calculating the standard deviation of the optical flux signal within a short time window, this system can accurately quantify the degree of jitter or instability of the output optical signal. Even small but high-frequency mechanical vibrations may cause the optical coupling jitter value detected by the system to increase sharply, thereby sensitively detecting early mechanical faults such as loose connections and micro-cracks in optical fibers.

[0033] 4. Providing data- and statistically based scientific decision-making: The system adopts a two-level early warning threshold and constructs a three-level risk response system. This hierarchical early warning mechanism avoids alarm fatigue caused by triggering high-level alarms due to slight deterioration of the situation, and also prevents the risk of not providing a sufficiently strong warning when the situation is extremely dangerous. By directly linking the threshold to the failure probability, the early warning behavior is no longer subjective, but a data- and statistically based scientific decision-making process, ensuring the timeliness, accuracy, and action guidance of the early warning, forming an efficient closed-loop risk management logic. Attached Figure Description

[0034] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0035] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0037] Example 1:

[0038] Please see Figure 1An endoscope cold light source beam hot-plugging status detection system includes:

[0039] The data acquisition module is used to acquire the current noise timing signal, the temperature timing signal of the core area of ​​the interface, and the light flux timing signal in real time to generate multiphysics field signals.

[0040] The feature extraction module is used to generate a set of feature vectors based on the multi-physics field signals generated by the data acquisition module; the set of feature vectors includes electrical noise feature vectors, thermal instability feature vectors, and optical coupling jitter feature vectors.

[0041] The risk assessment module is used to normalize the feature vector set generated by the feature extraction module, combined with the preset ideal benchmark value and critical failure threshold, in order to generate a risk feature vector set.

[0042] The health score calculation module is used to calculate the connectivity health score index based on the risk feature vector set generated by the risk assessment module and combined with preset weight coefficients through a linear weighting model.

[0043] The risk warning module is used to compare the connection health index generated by the health calculation module with the preset attention warning threshold and danger warning threshold to generate a risk level.

[0044] The hot-plugging status detection system for the endoscope cold light source beam disclosed in this embodiment aims to overcome the limitations of existing technologies that can only perform post-fault judgment, and provide a solution that can predict the trend of connection status deterioration and provide graded early warning. In a specific application scenario, the system is deployed on an endoscope cold light source device to monitor the connection reliability between the beam guide cable and the light source host interface in real time. The system completely includes a data acquisition module, a feature extraction module, a risk assessment module, a health calculation module, and a risk early warning module. The above modules work together to form a complete technical closed loop from the underlying physical signal perception to the top-level decision support.

[0045] The purpose of the data acquisition module is to capture all key physical information that can characterize the state of the connection interface in a comprehensive and lossless manner. In order to ensure data quality and system robustness, the data acquisition module performs outlier detection and filtering algorithms before sending the raw signal to the feature extraction module. This algorithm can identify and eliminate extreme values ​​that exceed physically acceptable ranges due to sensor malfunctions, such as temperatures below 0 degrees Celsius or negative luminous flux, thereby preventing abnormal data from contaminating subsequent evaluation processes. In this embodiment, the module integrates a composite sensor array at a preset key location on the light source interface. This array is a carefully designed collaborative monitoring unit, specifically including: a high-frequency current noise sensor for capturing high-frequency electrical signals generated by contact micro-movements or potential arc discharges; a miniature thermocouple for monitoring the Joule heating effect caused by changes in contact resistance; and a high-sensitivity photodiode for detecting luminous flux fluctuations caused by physical micro-displacements at the fiber bundle end face. During system operation, the data acquisition module continuously acquires data at a high sampling rate from the electrical, thermal, and optical physical dimensions of the connection interface, generating three core raw time-series signals in real time: current noise time-series signal... Temperature timing signal of the core area of ​​the interface and optical flux timing signal This set of multiphysics signal sets contains multiphysics coupling information. , , This data forms the foundation for all subsequent analysis and evaluation, fundamentally solving the problem that traditional detection methods cannot provide early warning of sudden failures due to their limited information dimensions.

[0046] The feature extraction module aims to extract and transform the state information contained in the complex, high-dimensional raw time-series signals into structured data that can directly quantify the degree of connection instability. In this embodiment, the module receives multi-physics field signals generated by the data acquisition module and processes them into a set of feature vectors through three parallel computation branches. This set of feature vectors explicitly includes three core vectors: an electrical noise feature vector aimed at quantifying electrical contact stability, a thermal instability feature vector aimed at capturing precursors of thermal runaway, and an optical coupling jitter feature vector aimed at measuring the precision of optical path alignment. These three vectors jointly depict the health profile of the connection interface from different physical dimensions.

[0047] The risk assessment module aims to transform raw feature vectors from different physical dimensions and with different dimensions into a unified and standardized risk measurement system. In this embodiment, the module receives a set of feature vectors generated by the feature extraction module and normalizes each feature vector in the set by combining a preset ideal benchmark value and a critical failure threshold. Here, the ideal benchmark value is the theoretically optimal feature value of the connection in a perfect and flawless state, and its value is obtained by static calibration testing on a brand-new, ideal connection component. The critical failure threshold is a critical point determined through a large number of destructive experiments. Once the feature value reaches this level, it indicates that the connection will experience functional failure in a very short time. Through this normalization process, a dimensionless set of risk feature vectors is generated, where the value of each element is constrained between 0 and 1, intuitively representing the evolution of this dimension from absolute safety to impending failure.

[0048] The purpose of the health assessment module is to integrate multi-dimensional risk assessment results into a single, intuitive, and easy-to-understand overall health status indicator. In this embodiment, the module receives a set of risk feature vectors generated by the risk assessment module and calculates the index using a linear weighted model, combined with preset weighting coefficients. These weighting coefficients are a set of values ​​corresponding to the three risk dimensions of electricity, heat, and light. These values ​​are not manually set but are derived through supervised machine learning algorithms, optimized after regression training on a dataset of thousands of labeled samples ranging from normal to various failure modes. The aim is to minimize the error between the final calculated result and the actual observed connection reliability. The module ultimately calculates and generates a connection health index ranging from 0 to 100 using this model.

[0049] The risk warning module aims to transform the quantified health index into clear instructions that operators can immediately understand and take action. In this embodiment, the module continuously compares the connection health index generated in real time by the health calculation module with two preset levels of attention warning thresholds and danger warning thresholds. The attention warning thresholds and danger warning thresholds are set based on statistical analysis of massive amounts of historical failure data. Through this comparison, the module ultimately generates a clear risk level and drives the corresponding user interface to issue a warning.

[0050] This embodiment constructs a complete closed-loop system from bottom-level multiphysics data acquisition to top-level risk warning through the collaborative work of the aforementioned modules. This system primarily focuses on three core physical dimensions: electricity, heat, and light. The model already provides high-precision predictions for conventional hot-plugging and long-term use scenarios. However, in extreme environments, the system's evaluation accuracy may be affected. Future development could consider adding monitoring of physical quantities such as environmental humidity and mechanical vibration to further improve the model's applicability and robustness. Through in-depth analysis and fusion evaluation of electrical, thermal, and optical coupling signals, continuous tracking and quantitative evaluation of the entire process of connection status from healthy to failed are achieved. This allows the system to identify early signs of connection deterioration and provide risk classification warnings with clear physical meaning and probabilistic support, thus providing a decision-making window for equipment maintenance and clinical operations, significantly improving the reliability and safety of the endoscope system.

[0051] Example 2:

[0052] The feature extraction module generates electrical noise feature vectors, including:

[0053] Perform a Fourier transform on the current noise time-series signal to obtain the power spectral density function;

[0054] The power spectral density function is integrated over a preset frequency range to generate an electrical noise feature vector.

[0055] This embodiment generates an electrical noise feature vector using the feature extraction module described in Embodiment 1. One specific implementation method aims to accurately quantify the total energy of electromagnetic noise that is highly correlated with connection failure, generated by contact micro-movements or weak arc discharges.

[0056] This process analyzes the current noise timing signal acquired by the data acquisition module. Perform a Fast Fourier Transform to obtain its power spectral density function in the frequency domain. This is used to analyze the energy distribution of different frequency components; based on this power spectral density function Within a preset frequency range Integral operations are performed within the range to generate the final electrical noise feature vector. Its calculation model is as follows: ;

[0057] in, Its source is current noise signal. Obtained through Fourier transform, its physical meaning is the distribution density of signal power at different frequency points; These are the lower and upper limits of a preset frequency range, respectively. The method for determining these two parameters is as follows: through continuous spectrum monitoring and data analysis of a large number of connection samples from a stable state to eventual failure, the characteristic frequency band with the strongest correlation to connection state deterioration and the highest signal-to-noise ratio is statistically identified. In this embodiment, the characteristic frequency band... Analysis determined that the frequency band from 1kHz to 100kHz is the strongest correlation between signals and contact fretting and arc discharge; this characteristic frequency band The determination method is as follows: perform spectral analysis on the current noise signal of a large number of labeled connection failure samples; use data analysis methods such as mutual information method or principal component analysis to identify the frequency range with the most significant change in power spectral density and the highest signal-to-noise ratio during the process of connection state evolving from healthy to failure.

[0058] Compared to simply calculating the total power or variance of the current signal, this embodiment, by integrating the power spectral density within a specific, experimentally calibrated frequency range, can more accurately capture the characteristic noise that indicates impending connection failure. This method effectively filters out background noise and power frequency interference unrelated to the connection status, resulting in a more accurate extracted electrical noise feature vector. It has higher sensitivity and specificity to early minor degradation phenomena at the interface, thus significantly improving the accuracy and timeliness of early warning.

[0059] Example 3:

[0060] The feature extraction module generates thermal instability feature vectors, including:

[0061] The first derivative of the temperature time series signal in the core area of ​​the interface with respect to time is calculated to generate the derivative value;

[0062] The derivative value is determined as the thermal instability eigenvector.

[0063] This embodiment generates thermal instability feature vectors using the feature extraction module described in Embodiment 1. One specific implementation method is to capture the abnormal temperature rise rate caused by the instantaneous increase in contact resistance and the resulting Joule heating positive feedback effect, which is a clear physical precursor to connection overheating failure.

[0064] This process acquires the time-series temperature signal of the core area of ​​the interface, which is collected in real time by the data acquisition module. Calculate the temperature signal versus time. The first derivative of , to generate a derivative value: ;

[0065] The system directly determines this derivative value as the thermal instability characteristic vector. In digital signal processing, this derivative is approximated by performing differential operations on continuous temperature sampling points. Although this embodiment uses simple differential operations to approximate the temperature rise rate, this method may lead to unstable calculation results in the case of high-frequency noise or insufficient sampling rate. To enhance physical fidelity, the least squares method can be used to linearly fit the temperature data within the sliding window, and the slope of the fitted line can be used as the temperature rise rate. This method can effectively filter out instantaneous noise and more stably reflect the long-term temperature change trend.

[0066] A stable and healthy connection has an interface temperature It should remain stable or have only minor fluctuations, in which case its first derivative with respect to time is... The value is close to zero; once the connection deteriorates, the increase in contact resistance will lead to intensified local heat generation, forming a vicious cycle of increased resistance - increased temperature - further increased resistance. At this point, the rate of temperature rise is... The value will increase dramatically; this embodiment, by directly calculating the rate of temperature rise rather than the temperature itself, can detect this trend of thermal runaway earlier than the traditional fixed temperature threshold alarm method. Its advantage is that it is not sensitive to absolute temperature, but highly sensitive to the rate of acceleration of temperature change, thus providing a more timely warning before catastrophic overheating occurs.

[0067] Example 4:

[0068] The feature extraction module generates optically coupled jitter feature vectors, including:

[0069] Multiple luminous flux sample values ​​are collected within a preset sliding time window;

[0070] Calculate the standard deviation of all luminous flux sample values ​​within the sliding time window to generate the standard deviation value;

[0071] The standard deviation was determined as the optical coupling jitter feature vector.

[0072] This embodiment generates an optically coupled jitter feature vector using the feature extraction module described in Embodiment 1. One specific implementation method aims to quantify the instability of optical flux output caused by the physical micro-displacement between the end face of the fiber bundle and the focal point of the light source, which directly reflects the stability of the mechanical connection.

[0073] This process takes a preset length of Continuous acquisition within the sliding time window Each luminous flux sample value is denoted as . ( ), and calculate this The standard deviation of each luminous flux sample value is used to generate a standard deviation value; this standard deviation is directly determined as the optical coupling jitter feature vector. Its calculation model is as follows: ;

[0074] in, : for a length of The first time the data was collected within the sliding time window Each luminous flux sample value originates from the photodiode in the data acquisition module; : All within this window The arithmetic mean of the sample values; The sliding time window length is a key adjustable parameter. Its value is determined through experimental calibration, striking a balance between sensitivity and stability to find the window length that maximizes the signal-to-noise ratio. In this embodiment, the sliding time window length... After experimental calibration, 100 sample points were selected to achieve the best balance between sensitivity and stability;

[0075] Compared to simply monitoring whether the average luminous flux is below a certain threshold, this embodiment calculates the standard deviation of the luminous flux signal within a short time window, thus accurately quantifying the jitter or instability of the output optical signal. A tiny but high-frequency mechanical vibration may not significantly affect the average luminous flux, but it can cause... The value increases sharply; therefore, this method can sensitively detect early mechanical failures such as loose connections and microcracks in optical fibers, providing a direct and quantitative assessment of the mechanical stability of the connection.

[0076] Example 5:

[0077] The risk assessment module generates a set of risk feature vectors, including:

[0078] The difference between any feature vector in the feature vector set and the ideal benchmark value corresponding to that vector is divided by the difference between the critical failure threshold corresponding to that vector and the ideal benchmark value to obtain a linear mapping result.

[0079] The linear mapping result is constrained within a preset closed interval to generate a normalized risk feature vector.

[0080] This embodiment is a specific implementation of the risk assessment module described in Embodiment 1 that generates a set of risk feature vectors. It takes the original feature vectors generated in the preceding steps, which have different physical dimensions and numerical ranges, and... }, mapped to a unified, dimensionless Within the closed interval, this lays the foundation for subsequent multidimensional information fusion calculations;

[0081] This process applies to any eigenvector in the eigenvector set. Perform the following calculations: Calculate the real-time value of the vector and the corresponding ideal reference value. The difference And divide this difference by the critical failure threshold corresponding to the vector. Compared with the ideal benchmark value The difference To obtain a linear mapping result;

[0082] To enhance the robustness of the model, the linear mapping result is strictly constrained within a predefined closed interval. Within this process, the final normalized risk feature vector is generated. The complete calculation model is as follows: ;

[0083] in, : Represents any original feature vector calculated in real time ( Its source is the output of the feature extraction module; This is the ideal baseline value, representing the theoretical optimal value under perfect connection conditions. It originates from background values ​​obtained through static testing and calibration of a brand-new, high-performance connector. For example, in a specific application scenario... It can be a vector whose components are ideal values ​​of electrical noise, thermal, and optical characteristics, such as... ; : This is the critical failure threshold, representing the point at which a connection is about to experience functional failure; it originates from a physical critical point determined through extensive destructive testing, for example, Similarly, given a vector, its components can be set as follows: This represents the critical value for electrical noise, thermal, and optical characteristics when the connection is about to fail; This function ensures the final output value. Strictly limited to Within the closed interval, 0 represents the ideal state, and 1 represents the state that has reached or exceeded the critical failure state.

[0084] This normalization method transforms a raw physical measurement into a standardized index characterizing the degree of state evolution; an output value of 0.3 explicitly indicates that the current state has evolved 30% from the ideal state and is about to reach the critical failure point; furthermore, through the outer layer... and By constraining the function, computational overflow caused by abnormal sensor jumps or extreme noise can be effectively avoided, greatly enhancing the stability and reliability of the entire evaluation model.

[0085] Example 6:

[0086] The health calculation module generates a connection health index, including:

[0087] The comprehensive damage degree is obtained by multiplying each risk feature vector in the risk feature vector set with its corresponding preset weight coefficient and then summing the results.

[0088] The connection health index is generated by subtracting the overall damage level from the preset maximum health level.

[0089] This embodiment generates a connection health index using the health calculation module described in Embodiment 1. One specific implementation method aims to combine the risk feature vector set generated by the previous module, which describes each independent risk dimension, into a single set. This is achieved by integrating a physically interpretable model into a single, comprehensive health score ranging from 0 to 100.

[0090] This process will combine each risk feature vector in the risk feature vector set ( ) and their respective preset weight coefficients ( The products are multiplied and summed to obtain a comprehensive quantitative index, which is defined as the comprehensive damage degree in this embodiment. The calculation formula is as follows: ;

[0091] Weighting coefficients here satisfy The value of this weight is determined by training a supervised machine learning model on a complete dataset containing thousands of labeled samples covering various failure modes from unconnected to perfectly connected and critically contacted. In this embodiment, the weight coefficient is obtained by training a multiple linear regression model on a labeled dataset containing over 5000 samples ranging from healthy to failed. Each sample in the dataset contains a three-dimensional normalized risk feature vector. As input, and its corresponding true reliability score as output label;

[0092] Subtract the calculated total damage from a preset maximum health value of 100. The data is then scaled up proportionally to generate the final connection health index. The calculation formula is as follows: ;

[0093] Among them, the risk feature vector { The source of} is the output of the risk assessment module;

[0094] The core innovation of the linear weighted model used in this embodiment lies in the method of obtaining the weight coefficients. Weights are obtained through machine learning training from massive amounts of data, enabling the model to automatically learn and reflect the differences in importance of different physical dimensions under different failure scenarios. This data-driven approach, compared to simply averaging the risks of each dimension, generates a health index that more closely matches real-world failure patterns, thus significantly improving the accuracy and predictive ability of the assessment results. Although the model uses a linear weighting method, in practical applications, we have found that there may be complex nonlinear coupling relationships between risks of different physical dimensions. To balance computational efficiency and model interpretability, this embodiment chose a linear model. For applications with higher precision, future applications can use a neural network-based nonlinear regression model to replace the linear weighted model to better capture these complex interactions.

[0095] Example 7:

[0096] The risk levels generated by the risk warning module include:

[0097] When the connection health index is greater than the warning threshold, it is determined to be at a safe level;

[0098] When the connection health index is less than or equal to the attention warning threshold, but greater than the danger warning threshold, it is determined to be at the attention level.

[0099] When the connection health index is less than or equal to the danger warning threshold, it is determined to be at a danger level.

[0100] This embodiment generates risk levels using the risk warning module described in Embodiment 1. One specific implementation method aims to integrate continuously changing connectivity health indices. The risk levels are transformed into discrete levels with clear operational instructions, enabling effective human-computer interaction.

[0101] This module calculates the connection health index in real time. With two preset thresholds, namely attention warning thresholds and danger warning threshold The system compares the data; based on the comparison results, it determines the current connection status as one of three levels: security level. Note the level or hazard level ;

[0102] When connected to the health index Greater than the warning threshold At that time, the system determines the current risk level as a safe level. ;

[0103] When connected to the health index Less than or equal to the attention warning threshold And greater than the danger warning threshold At that time, the system determines the current risk level as the alert level. ;

[0104] When connected to the health index Less than or equal to the danger warning threshold At that time, the system determines the current risk level as dangerous. ;

[0105] The attention warning threshold here and danger warning threshold The setting is based on rigorous statistical evidence, derived from statistical analysis and risk assessment models of a large amount of historical failure data; in this embodiment, attention is paid to the warning threshold. and danger warning threshold After statistical analysis, it was set as follows: and These correspond to two critical points in the connection state that require attention and those that require immediate action, such as the danger warning threshold. The setting is based on the following: data analysis revealed that when the health index drops to a certain value, the conditional probability of a functional interruption of the connection within the next minute exceeds 95%.

[0106] This embodiment constructs a three-level risk response system by setting two levels of early warning thresholds. Compared with the traditional binary alarm system, this hierarchical early warning mechanism avoids alarm fatigue caused by triggering high-level alarms due to slight deterioration of the situation, and also prevents the risk of not providing a sufficiently strong warning when the situation is extremely dangerous. By directly linking the threshold to the failure probability, the early warning behavior is no longer subjective, but a scientific decision based on data and statistics, ensuring the timeliness, accuracy and action guidance of the early warning, and ultimately forming an efficient closed-loop risk management logic.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An endoscope cold light source beam guide hot-plugging status detection system, characterized in that, include: The data acquisition module is used to acquire the current noise timing signal, the temperature timing signal of the core area of ​​the interface, and the light flux timing signal in real time to generate multiphysics field signals. The feature extraction module is used to generate a set of feature vectors based on the multi-physics field signals generated by the data acquisition module; the set of feature vectors includes electrical noise feature vectors, thermal instability feature vectors, and optical coupling jitter feature vectors. The risk assessment module is used to normalize the feature vector set generated by the feature extraction module, combined with the preset ideal benchmark value and critical failure threshold, in order to generate a risk feature vector set. The health score calculation module is used to calculate the connectivity health score index based on the risk feature vector set generated by the risk assessment module and combined with preset weight coefficients through a linear weighting model. The risk warning module is used to compare the connection health index generated by the health calculation module with the preset attention warning threshold and danger warning threshold to generate a risk level.

2. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The feature extraction module generates electrical noise feature vectors, including: Perform a Fourier transform on the current noise time-series signal to obtain the power spectral density function; The power spectral density function is integrated over a preset frequency range to generate an electrical noise feature vector.

3. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The feature extraction module generates thermal instability feature vectors, including: The first derivative of the temperature time series signal in the core area of ​​the interface with respect to time is calculated to generate the derivative value; The derivative value is determined as the thermal instability eigenvector.

4. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The feature extraction module generates optically coupled jitter feature vectors, including: Multiple luminous flux sample values ​​are collected within a preset sliding time window; Calculate the standard deviation of all luminous flux sample values ​​within the sliding time window to generate the standard deviation value; The standard deviation was determined as the optical coupling jitter feature vector.

5. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The risk assessment module generates a set of risk feature vectors, including: The difference between any feature vector in the feature vector set and the ideal benchmark value corresponding to that vector is divided by the difference between the critical failure threshold corresponding to that vector and the ideal benchmark value to obtain a linear mapping result. The linear mapping result is constrained within a preset closed interval to generate a normalized risk feature vector.

6. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The health calculation module generates a connection health index, including: The comprehensive damage degree is obtained by multiplying each risk feature vector in the risk feature vector set with its corresponding preset weight coefficient and then summing the results. The connection health index is generated by subtracting the overall damage level from the preset maximum health level.

7. The endoscope cold light source beam guide hot-plugging status detection system according to claim 1, characterized in that, The risk levels generated by the risk warning module include: When the connection health index is greater than the warning threshold, it is determined to be at a safe level; When the connection health index is less than or equal to the attention warning threshold, but greater than the danger warning threshold, it is determined to be at the attention level. When the connection health index is less than or equal to the danger warning threshold, it is determined to be at a danger level.

Images