Wearable vital signs simulator opto-plesio-pulse wave link failure detection method

Abstract

This invention discloses a method for detecting link faults in a wearable vital signs simulator (WVSS) based on photoplethysmography (PPG), belonging to the field of intelligent fault diagnosis. The invention acquires the WVSS's set parameters and calculates the expected ideal AC / DC ratio based on pre-stored calibration coefficients. Within a continuous diagnostic time window, it acquires a PPG digital sequence, calculates the measured AC / DC ratio, and differs it from the expected value to obtain the parameter domain residual. Through adaptive pulse template reconstruction, it obtains a microstructure residual sequence and calculates its energy in the key frequency band as the microstructure distortion intensity. It calculates the absolute value of the correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence within the continuous time window as a link integrity diagnostic indicator. After comparison with a preset threshold, it classifies and determines whether the link is intact, degraded, or faulty. This invention relies solely on the WVSS's own output signal, can distinguish between gain drift and waveform structural distortion, and possesses the ability to locate, classify, and interpret faults. It is suitable for production line calibration, on-site maintenance, and built-in self-test scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent fault diagnosis, and more specifically, to a method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator. Background Technology

[0002] With the increasing awareness of health management and the rapid growth of the wearable health device market (such as smartwatches and wristbands), the Wearable Vital Signs Simulator (WVSS) is becoming increasingly important as a standardized testing device. WVSS can preset and stably output various vital sign signals, establishing a unified benchmark for algorithm verification, performance testing, production line calibration, and quality certification of wearable devices.

[0003] However, existing WVSS-type PPG simulators typically support multiple parameter settings (such as R-curve type, pulse amplitude PI, pulse rate HR, transmission control parameters TLC, etc.), but lack closed-loop awareness of the physical integrity of the signal generation link. The following pain points exist in practice:

[0004] Hidden faults are difficult to detect: Even if user-defined parameters remain unchanged, hardware degradation or changes in assembly status can still introduce structural distortions unintended by the parameter settings, such as increased baseline drift, AC / DC ratio imbalance, rise time broadening, peak splitting, and abnormal harmonic aggregation. These distortions may not necessarily change the nominal parameter values ​​displayed on the UI, but they can cause the wearable device under test to misinterpret the simulator output as weak injection, motion artifacts, or poor signal quality, leading to invalid downstream test results.

[0005] Lack of quantifiable diagnostic methods: Current practices often rely on subjective judgment by visually inspecting waveforms or intensity bars, lacking quantifiable thresholds, traceable evidence, and adaptive capabilities, making it difficult to meet compliance requirements for verifiable output credibility.

[0006] Existing technologies have limitations:

[0007] The solution that focuses on PPG quality assessment at the wearable device end relies on accelerometers or edge-side algorithms to score the signal quality, and its detection object is the terminal device rather than the simulator output link.

[0008] The solution that focuses on the self-testing of the PPG sensor module relies on hardware parameters such as LED forward voltage drop, junction temperature, and photodiode dark current for health monitoring. This requires additional hardware, which increases the system complexity and cost.

[0009] Traditional spectral energy ratio thresholding methods only set a single index threshold in the signal domain, making it difficult to distinguish between morphological changes caused by parameter adjustment and structural distortions caused by link failures.

[0010] Therefore, existing technologies lack a specific integrity detection method for the PPG signal generation link of the WVSS wearable vital signs simulator that requires no additional hardware and can sensitively detect performance degradation and soft faults. This makes it impossible to meet the engineering requirements of production line calibration and equipment maintenance for efficiency, traceability, and interpretability. Summary of the Invention

[0011] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for detecting faults in the PPG signal generation link integrity of a wearable vital signs simulator. The aim is to achieve the following objectives: differentiate and locate faults at the gain / amplitude mapping layer and the waveform microstructure layer; output integrity, degradation, and fault levels; map diagnostic conclusions to understandable physical cause categories; complete the diagnosis based solely on the PPG digital sequence output by the WVSS itself, without relying on external reference sensors; and trace the diagnostic conclusions back to the factory calibration model, forming a verifiable chain of evidence for output credibility.

[0012] To achieve the above-mentioned objectives, this invention provides a method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator, the method comprising:

[0013] Step 1: Obtain the current setting parameters of the wearable vital signs simulator, including pulse curve type, pulse amplitude setting value, transmission control parameters, and pulse rate setting value;

[0014] Step 2: Based on the set parameters and the pre-stored calibration coefficients corresponding to the pulse curve type, calculate the expected value of the ratio of AC component to DC component;

[0015] Step 3: Acquire the PPG digital sequence output by the wearable vital signs simulator within multiple consecutive diagnostic time windows. For each diagnostic time window, perform parameter domain residual and microstructure distortion intensity calculation operations to obtain a corresponding parameter domain residual and a corresponding microstructure distortion intensity. The parameter domain residual and microstructure distortion intensity calculation operations include:

[0016] The measured value of the ratio of AC component to DC component is calculated based on the PPG digital sequence within the current diagnostic time window, and a parameter domain residual is generated based on the difference between the measured value and the expected value.

[0017] Based on the pulse rate setting value, the PPG digital sequence within the current diagnostic time window is reconstructed using a pulse waveform template to obtain a reconstructed sequence. A microstructure residual sequence is then generated based on the difference between the PPG digital sequence and the reconstructed sequence.

[0018] The microstructure residual sequence is subjected to frequency domain transformation, and the energy of the microstructure residual sequence in a preset key frequency band is calculated as the microstructure distortion intensity.

[0019] Step 4: Arrange the parameter domain residuals corresponding to the multiple consecutive diagnostic time windows in chronological order to form a parameter domain residual sequence; arrange the microstructure distortion intensity corresponding to the multiple consecutive diagnostic time windows in chronological order to form a microstructure distortion intensity sequence.

[0020] Step 5: Calculate the correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence, and use the absolute value of the correlation coefficient as the link integrity diagnostic index;

[0021] Step 6: Compare the link integrity diagnostic indicators with preset thresholds, and determine the integrity status of the link based on the comparison results.

[0022] The inventors discovered that existing technologies lack a method for diagnosing link integrity solely based on the WVSS's own output signal without relying on external sensors, making it difficult to distinguish between normal parameter changes and signal distortion caused by link faults. To address this issue, this method calculates the expected ideal AC / DC ratio using set parameters, and then differs it from the measured value to obtain the parameter domain residual. It then reconstructs the microstructure residual using an adaptive template, calculates its critical frequency band energy to determine the distortion intensity, and calculates the absolute value of the correlation coefficient between the parameter domain residual and the distortion intensity over a continuous time window as a diagnostic indicator, comparing it with a threshold to determine the integrity status. This method achieves link integrity diagnosis relying solely on the WVSS's own output signal, distinguishing between normal parameter changes and link faults, and possessing the capabilities of location, classification, and interpretation.

[0023] Preferably, the expected value of the ratio of AC component to DC component is calculated as follows:

[0024] ;

[0025] in, R represents the expected value of the ratio of the AC component to the DC component, and R is the pulse curve type. Set a value for the pulse amplitude. To transmit control parameters, The normalization constant is This represents the equivalent optical path efficiency coefficient of type R under ideal tooling conditions. This is the attenuation / saturation tendency coefficient caused by gain control and link nonlinearity. These are the coefficients of higher-order nonlinear terms.

[0026] The above calculation method integrates the effects of optical path efficiency, gain linearity, and higher-order nonlinearity, and can accurately reflect the ideal AC / DC ratio under different parameter combinations, providing a reliable benchmark for parameter domain residuals.

[0027] Preferably, the specific steps for reconstructing the pulse waveform template include:

[0028] Based on the pulse rate setpoint v and the sampling rate f s Determine the length T of the main period. ;

[0029] Based on the main period length T, a waveform segment of a main period is extracted from the PPG digital sequence within the current diagnostic time window, and the waveform segment is resampled to obtain a normalized period template.

[0030] The periodic template is smoothed.

[0031] The smoothed periodic template is periodically expanded to generate a reconstructed sequence with the same length as the current diagnostic time window.

[0032] This method determines the main period length based on the pulse rate, extracts a main period waveform segment, resamples and smooths it to obtain a period template, and then periodically unfolds and reconstructs the sequence. The difference between this reconstructed sequence and the original sequence yields the microstructure residual. The reconstructed sequence absorbs the ideal pulse morphology dominated by the R-curve type and pulse rate, while the residual sequence mainly contains non-ideal details introduced by the hardware link, such as overshoot, phase distortion, and saturation shearing. This achieves the extraction of the main period morphology from the PPG sequence and obtains the microstructure residual reflecting the non-ideal characteristics of the hardware link.

[0033] Preferably, the microstructure distortion intensity is calculated as follows:

[0034] The power spectrum of the microstructure residual sequence is obtained by performing a discrete Fourier transform on the microstructure residual sequence.

[0035] Based on the lower boundary frequency f1 and upper boundary frequency f2 of the preset key frequency band, determine the corresponding first frequency domain index k1 and second frequency domain index k2:

[0036] ;

[0037] ;

[0038] Where N is the length of the microstructure residual sequence, f s Sampling rate;

[0039] The sum of all power spectra within the range from the first frequency domain index k1 to the second frequency domain index k2 is calculated as the microstructure distortion intensity E. i :

[0040] ;

[0041] in, Let be the power spectrum of the microstructure residual sequence corresponding to the i-th diagnostic time window, k is the frequency domain index, and the subscript r indicates that the power spectrum originates from the microstructure residual sequence.

[0042] This method uses a discrete Fourier transform to obtain the power spectrum of the microstructure residual sequence, and calculates the sum of the power spectra within a preset key frequency band (covering the fundamental heart rate frequency and its harmonics) as the distortion intensity. Link faults often introduce distortion energy in the pulsation-related frequency band, and this index can serve as an interpretable measure of the intensity of non-parametric structural distortion. This method quantifies the degree of distortion in the microstructure residuals and maps it to a measurable index.

[0043] Preferably, the correlation coefficient is the Pearson correlation coefficient, and the link integrity diagnostic index is calculated as follows:

[0044] ;

[0045] Wherein, D is the link integrity diagnostic index, and C is the Pearson correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence, calculated as follows:

[0046] ;

[0047] Where K is the number of consecutive diagnostic time windows. The parameter domain residual corresponding to the i-th diagnostic time window. for The mean, E i The intensity of microstructural distortion corresponding to the i-th diagnostic time window. For E i The mean.

[0048] When the link is intact, the parameter domain residual and the distortion intensity are stably correlated; when there is a fault, the coupling relationship between the two is broken, the correlation decreases significantly, and the D value can sensitively reflect the link integrity status.

[0049] Preferably, the preset threshold includes a first threshold θ1 and a second threshold θ2, and satisfies θ1 > θ2;

[0050] When D≥θ1: the link is considered complete;

[0051] When θ2≤D<θ1: the link is determined to be degraded;

[0052] When D < θ2: a link failure is determined;

[0053] Where D represents the link integrity diagnostic metric.

[0054] This method can output three levels of diagnostic results, meeting the diagnostic precision requirements of different scenarios such as production line calibration, on-site operation and maintenance, and automated testing.

[0055] Preferably, the first threshold and the second threshold are determined in the following manner:

[0056] Multiple diagnostic indicator samples of the wearable vital signs simulator under the condition of a qualified link are collected to form a calibration sample set;

[0057] An empirical distribution function is constructed based on the calibration sample set to determine the false alarm control level. and Corresponding empirical quantiles and ,in, ;

[0058] The first threshold and the second threshold are obtained based on the empirical quantile:

[0059] .

[0060] This method collects diagnostic indicator samples under qualified link conditions, constructs an empirical distribution function, and determines the preset false alarm control level. and The corresponding empirical quantile is used as the threshold. This method scientifically and traceably determines the integrity threshold, avoiding the subjectivity and arbitrariness of empirical thresholds.

[0061] Preferably, the first threshold and the second threshold are determined based on a preset confidence risk δ:

[0062] The diagnostic indicators of the calibration sample set are sorted in ascending order to obtain the order statistics. ,in, The total number of windows for the calibration sample set;

[0063] For the first threshold θ1, the minimum rank r1 is selected to satisfy:

[0064] ;

[0065] The r1-th value D(r1) in the order statistics is taken as the first threshold θ1;

[0066] For the second threshold θ2, the minimum rank r2 is selected to satisfy:

[0067] ;

[0068] The r2th value D(r2) in the order statistic is used as the second threshold. .

[0069] Specifically, the calibration samples are sorted in ascending order to obtain the order statistics. Based on the binomial distribution probability condition, the minimum rank r is selected to satisfy the probability inequality, and the r-th order statistic is taken as the threshold. The threshold is then used to control the false alarm level. and With confidence risk δ, it has a clear statistical meaning, is auditable and recalculated, and ensures the statistical confidence of the threshold under limited calibration samples, avoiding the randomness caused by insufficient sample size.

[0070] Preferably, during the execution of step 3, the method further includes:

[0071] Calculate the mean value of the microstructure distortion intensity. Calculate the energy deviation for each diagnostic time window. , E i The intensity of microstructural distortion corresponding to the i-th diagnostic time window;

[0072] According to the parameter domain residual With the energy deviation Output the corresponding fault type, which includes:

[0073] when Greater than 0 and When the value is greater than 0, the output gain is nonlinear and faulty.

[0074] when Greater than 0 and When the value is less than 0, a saturation shear fault is output.

[0075] when Less than 0 and When the value is less than 0, there is an output coupling attenuation fault;

[0076] when Less than 0 and When the value is greater than 0, there is a stray interference fault in the output optical path;

[0077] When the parameter domain residual The absolute value is less than the preset first distortion threshold and the microstructure distortion intensity E i If the magnitude of the change within multiple consecutive diagnostic time windows exceeds the preset second distortion threshold, a sampling phase distortion fault is output.

[0078] Among them, the energy deviation is calculated. ,according to and The symbol combination output corresponds to the fault type: positive-positive output gain nonlinearity, positive-negative output saturation shearing, negative-negative output coupling attenuation, and output sampling phase distortion when the parameter domain residual is not significant and the distortion intensity fluctuates greatly. Mapping the diagnostic conclusions to understandable physical causes facilitates rapid troubleshooting and repair.

[0079] Preferably, the method further includes step 7 after step 6:

[0080] The integrity status and / or the fault type are output to a display device in a visual form, or a diagnostic report containing diagnostic time, diagnostic indicators, and judgment results is generated.

[0081] Specifically, it outputs the integrity status and / or fault type to a display device in a visual format, or generates a diagnostic report that includes diagnostic time, diagnostic indicators, and judgment results. The diagnostic results are presented to users in a usable format to meet the needs of production line recording, compliance auditing, and other requirements.

[0082] One or more technical solutions provided by this invention have at least the following technical effects or advantages:

[0083] Self-containment: This method can complete link integrity diagnosis based solely on the PPG digital sequence output by WVSS itself, without relying on external reference sensors or additional measurement points. It can be embedded in firmware to implement built-in self-test (BIST), without changing the hardware architecture, reducing deployment costs and improving maintenance convenience.

[0084] Interpretability: By decomposing anomalies into two types of information sources—parameter domain residuals and time domain microstructure residuals—it is possible to distinguish between gain control drift problems and waveform structural distortion problems. Diagnostic conclusions can be mapped to physical causes such as gain nonlinearity, saturation shearing, coupling attenuation, and sampling phase distortion, facilitating rapid rework on the production line and quick on-site troubleshooting.

[0085] Robustness: This method does not rely on absolute amplitude thresholds, but uses the cross-window consistency statistic (the absolute value of the Pearson correlation coefficient) as the core criterion. It has good robustness to user adjustments of parameters such as PI, TLC, and HR, and avoids misjudging normal parameter adjustments as faults.

[0086] Traceability: By explicitly introducing the parameter-signal deterministic prior of WVSS into the diagnostic process, the diagnostic conclusions can be traced back to the factory calibration model, forming a verifiable chain of evidence for the credibility of the output.

[0087] Quantifiable grading: By setting an integrity threshold, it outputs three levels of diagnostic results: complete, degraded, and faulty, to meet the application needs of different scenarios. Attached Figure Description

[0088] The accompanying drawings, which are provided to further illustrate embodiments of the invention and constitute a part of this invention, are not intended to limit the scope of the invention.

[0089] Figure 1 A flowchart illustrating the method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator. Detailed Implementation

[0090] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, where there is no conflict, the embodiments of the present invention and the features thereof can be combined with each other.

[0091] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0092] Those skilled in the art should understand that, in the disclosure of this invention, the terms "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the above terms should not be construed as limiting this invention.

[0093] It is understood that the term "a" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple, and the term "a" should not be understood as a limitation on the number.

[0094] Example 1;

[0095] Please refer to Figure 1 , Figure 1 This invention provides a method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator. The method includes:

[0096] Step 1: Obtain the current setting parameters of the wearable vital signs simulator, including pulse curve type, pulse amplitude setting value, transmission control parameters, and pulse rate setting value;

[0097] Step 2: Based on the set parameters and the pre-stored calibration coefficients corresponding to the pulse curve type, calculate the expected value of the ratio of AC component to DC component; that is, based on the set parameters and the pre-stored calibration coefficients corresponding to the pulse curve type, use the pulse amplitude setting value and the transmission control parameters to calculate the expected value of the ideal AC component to DC component ratio.

[0098] Step 3: Acquire the PPG digital sequence output by the wearable vital signs simulator within multiple consecutive diagnostic time windows. For each diagnostic time window, perform parameter domain residual and microstructure distortion intensity calculation operations to obtain a corresponding parameter domain residual and a corresponding microstructure distortion intensity. The parameter domain residual and microstructure distortion intensity calculation operations include:

[0099] The measured value of the ratio of AC component to DC component is calculated based on the PPG digital sequence within the current diagnostic time window, and a parameter domain residual is generated based on the difference between the measured value and the expected value.

[0100] Based on the pulse rate setting value, the PPG digital sequence within the current diagnostic time window is reconstructed using a pulse waveform template to obtain a reconstructed sequence. A microstructure residual sequence is then generated based on the difference between the PPG digital sequence and the reconstructed sequence.

[0101] The microstructure residual sequence is subjected to frequency domain transformation, and the energy of the microstructure residual sequence in a preset key frequency band is calculated as the microstructure distortion intensity.

[0102] Step 4: Arrange the parameter domain residuals corresponding to the multiple consecutive diagnostic time windows in chronological order to form a parameter domain residual sequence; arrange the microstructure distortion intensity corresponding to the multiple consecutive diagnostic time windows in chronological order to form a microstructure distortion intensity sequence.

[0103] Step 5: Calculate the correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence, and use the absolute value of the correlation coefficient as the link integrity diagnostic index;

[0104] Step 6: Compare the link integrity diagnostic indicators with preset thresholds, and determine the integrity status of the link based on the comparison results.

[0105] The principle of this invention is as follows:

[0106] Parameter domain residuals: Using the deterministic mapping between the R curve type, pulse amplitude setpoint, transmission control parameters and AC / DC ratio formed by WVSS factory calibration, an ideal AC / DC ratio expectation model is constructed to obtain the parameter domain residuals.

[0107] Temporal microstructure residuals: The output PPG sequence is reconstructed using an adaptive pulse template to remove the main periodic morphology, resulting in a microstructure residual sequence. The energy of this residual sequence within the key frequency band is then extracted as the waveform distortion intensity.

[0108] Over a continuous time window, the statistical correlation between the parameter domain residual sequence and the microstructure distortion intensity sequence is examined to see if it remains stable. High consistency indicates good link integrity; a collapse in consistency indicates degradation or failure.

[0109] The key to this invention lies in the fact that the fault is not random noise, but rather disrupts the physical coupling relationship between the parameter domain prediction error and the time domain distortion intensity. By verifying the stability of this coupling relationship, link degradation and soft faults can be detected sensitively.

[0110] The objective of this invention is achieved through the following technical solutions:

[0111] Step 1: Establishing an ideal AC / DC ratio expectation model: This invention is based on the mapping data of R curve type R-pulse amplitude setpoint-transmission control parameters and AC / DC ratio collected during the WVSS factory calibration stage, and constructs an analytical model;

[0112] Step 2: Generate parameter domain residuals: Collect the PPG digital sequence output by WVSS within multiple consecutive diagnostic time windows. For each diagnostic time window, calculate the measured AC / DC ratio and the difference between it and the expected value to obtain the parameter domain residuals. These residuals reflect the degree of deviation of the system-level mapping.

[0113] Step 3: Generating microstructure residuals and distortion intensity: For each diagnostic time window, the main period length is determined based on the pulse rate setting and sampling rate. A waveform segment of length T is extracted from the PPG digital sequence. After resampling and smoothing, a period template is obtained. The template is then periodically unfolded to reconstruct a reconstructed sequence with the same length as the original sequence. The difference is calculated to obtain the microstructure residual sequence. The power spectrum is obtained by performing a discrete Fourier transform on the microstructure residual sequence. The energy is calculated as the microstructure distortion intensity within the preset key frequency band (covering the heart rate fundamental frequency and its harmonics). This distortion intensity reflects the device-level distortion intensity.

[0114] Step 4: Dual-scale consistency test: Calculate the Pearson correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence corresponding to K consecutive diagnostic time windows, and take the absolute value as the link integrity diagnostic index. This index quantifies the stability of the coupling relationship between parameter domain prediction error and time domain distortion intensity.

[0115] Step 5: Classification and Fault Location: Set a first integrity threshold and a second integrity threshold. When D ≥ the first integrity threshold, the link is considered intact; when the second integrity threshold ≤ D < the first integrity threshold, the link is considered degraded; and when D < the second integrity threshold, the link is considered faulty. Simultaneously, during the calculation process, the corresponding fault type is output based on the sign combination of the parameter domain residual and the energy deviation of the microstructure distortion intensity.

[0116] Step 6: Output Results: Output the integrity status and / or fault type to a display device in a visual format, or generate a diagnostic report containing diagnostic time, diagnostic indicators, and judgment results. Through the synergistic effect of the above technical solutions, self-contained, hierarchical, and interpretable fault detection of the integrity of the WVSS PPG signal generation link is achieved.

[0117] The PPG signal is the absorption / scattering modulation response caused by changes in blood volume during the cardiac cycle after light irradiation of skin tissue. For the simulator, the temporal morphology of the PPG is constrained by both the deterministic parameters and waveform template, and the fidelity of the electronic and optical links.

[0118] The PPG generation chain of WVSS can be abstracted as follows:

[0119] ;

[0120] in: Let be a real number, representing the sample value of the PPG digital sequence output by WVSS at discrete time n. N is a natural number representing the number of sampling points for each diagnostic time window; f s A positive real number representing the sampling rate (fixed by the WVSS firmware or readable configuration); , representing the R curve type encoding (user-defined and firmware readable), R max π is a natural number representing the total number of types; π is a positive real number representing the pulse amplitude setting parameter PI (a numerical representation in percentage form, user-defined and firmware readable); τ is a natural number representing the transmission control parameter TLC (Transmission Level Control, firmware readable), used to compensate for differences in optical path attenuation caused by different wearable devices; v is a positive real number representing the pulse rate HR (a numerical representation in bpm form, user-defined and firmware readable); ... represents optional disturbance terms such as breathing modulation and ambient light interference.

[0121] Among them, the Wearable Vital Signs Simulator (WVSS) is a standardized testing device used to simulate human physiological signals in order to perform performance verification, algorithm calibration and quality testing on wearable health devices (such as smartwatches, bracelets, medical-grade monitoring devices, etc.).

[0122] In the R curve type encoding, R is used to distinguish different PPG waveform templates, such as standard pulse waveform, motion interference waveform, special population waveform, etc. Each waveform corresponds to a set of pre-stored template data.

[0123] Transmission control parameters (TLC) are used to adjust the optical emission intensity or receive gain of the WVSS to compensate for differences in optical path attenuation caused by different wearable fixtures or optical path coupling conditions. In practical applications, TLC is usually an integer range (e.g., 0-255), set by the user according to the fixture conditions. This parameter directly affects the AC / DC ratio of the signal, and therefore serves as a key input in the ideal AC / DC ratio expectation model.

[0124] Among them, PI (pulse amplitude setpoint) is the user-defined percentage of the desired pulse amplitude, and the ideal AC / DC ratio should change positively with PI.

[0125] The length N of the diagnostic time window should cover at least two complete pulse cycles or a fixed value such as 1024 or 2048, which can be adjusted according to actual hardware resources.

[0126] The value of K can be K≥10 to ensure the statistical stability of the correlation coefficient; a sliding window method can also be used for updating.

[0127] The key frequency band can be dynamically adjusted according to the real-time pulse rate ν, or a fixed frequency band can be used to cover the common heart rate range.

[0128] Among them, energy calculations can be performed without normalization, since the correlation coefficients used later are not sensitive to the magnitude; or normalization can be used.

[0129] The conditions for collecting qualified calibration samples are as follows: when the link has been confirmed to be qualified (e.g., using new equipment, or verified by an oscilloscope to be free of distortion), diagnostic index samples under multiple parameter combinations (R, π, τ) are collected to form a calibration sample set.

[0130] In this embodiment, the pulse waveform template reconstruction is achieved through the following steps:

[0131] (1) Determining the main cycle length: Calculate the main cycle length (number of sampling points) based on the pulse rate setting and sampling rate.

[0132] (2) Phase offset estimation: In order to ensure that the template has phase consistency, an ideal sine wave with a frequency matching the pulse rate setting is first generated. The PPG sequence in the current window is subjected to sliding cross-correlation with the coarse template, and the offset corresponding to the maximum cross-correlation value is taken as the phase offset.

[0133] (3) Periodic segment extraction and resampling: From the PPG sequence within the current window, extract a waveform segment of length T starting from the phase offset. Use linear interpolation to uniformly resample this segment into M points to obtain a periodic segment, where M is the number of template sampling points. In this embodiment, M=128.

[0134] (4) Template smoothing: The periodic segment is subjected to low-pass filtering and the cutoff frequency is set to 3 times the pulse fundamental frequency to obtain a smooth template.

[0135] (5) Periodic unfolding reconstruction: The smooth template is periodically unfolded according to the main period length T, and a weighted average is performed at the period boundary using a Hanning window to obtain a reconstructed sequence with the same length as the original sequence.

[0136] (6) Microstructure residual calculation: Calculate the difference between the original sequence and the reconstructed sequence.

[0137] The parameter symbols in the embodiments of the present invention are defined and explained below:

[0138] Diagnostic window and sequence:

[0139] Let be a real number, representing the PPG output sequence samples within the i-th diagnostic time window. , N is a natural number representing the number of sampling points per window. f s A positive real number represents the sampling rate. K is a natural number representing the number of windows used for correlation calculation.

[0140] Readable setting parameters:

[0141] R is the curve type code. π is a positive real number representing the PI setting value. This is the TLC setting value. v is a positive real number representing the HR setting value (in bpm).

[0142] Amplitude components and ratios (for the first) window):

[0143] Let be a real number, representing the mean of the measured DC components, calculated as follows:

[0144] ;

[0145] A positive real number, representing the measured effective value of the AC component, is calculated as follows:

[0146] ;

[0147] A positive real number, representing the measured AC / DC ratio, calculated as follows:

[0148] .

[0149] Expected model and parameter domain residuals:

[0150] is a real number representing the ideal AC / DC ratio expectation value given R, π, and τ, which is calculated from the factory calibration model or obtained by table lookup interpolation.

[0151] Let be a real number, representing the parameter domain residual of the i-th window, calculated as follows:

[0152] ;

[0153] Template reconstruction and temporal microstructure residuals:

[0154] Let be a real number, representing the reconstructed sequence obtained by the adaptive pulse template within the i-th window. The adaptive pulse template refers to a standard pulse waveform template that matches the current main pulse morphology, extracted, normalized, and smoothed from the real-time acquired PPG digital sequence based on the current pulse rate setting of the wearable vital signs simulator (WVSS). Its core is to adaptively match the current pulse frequency characteristics, stripping away the ideal main periodicity from the PPG sequence, allowing the remaining residual to reflect only the non-ideal microstructural distortions introduced by the hardware link. This is the core foundation for realizing microstructural residual calculation.

[0155] Let be a real number, representing the microstructure residual sequence within the i-th window, calculated as follows:

[0156] ;

[0157] Residual spectrum and key band energy:

[0158] For a complex number, it represents the pair of numbers. The discrete Fourier transform (DFT) coefficients, The calculation method is as follows:

[0159] ;

[0160] Where 𝑗 is the imaginary unit and e is the base of the natural logarithm.

[0161] A positive real number represents the residual power spectrum (unnormalized), calculated as follows:

[0162] ;

[0163] Let be a real number, representing the frequency corresponding to the k-th DFT frequency point (k is the frequency point index), calculated as follows:

[0164] ;

[0165] f1 and f2 are positive real numbers, representing the lower and upper boundaries of the key frequency band (defined by the implementation, covering the heart rate fundamental frequency and several harmonic frequencies), i.e., the preset lower boundary frequency f1 and upper boundary frequency f2 of the key frequency band. Here, f1 can be the fundamental frequency, and f2 can be the third harmonic or higher, or it can be set to a fixed range (e.g., 0.5Hz~10Hz) according to the actual signal bandwidth.

[0166] k1 and k2 are natural numbers, representing the frequency indexes obtained by mapping from f1 and f2, i.e., the first frequency domain index k1 and the second frequency domain index k2, calculated as follows:

[0167] ;

[0168] ;

[0169] E i Let be a positive real number, representing the residual energy of the key frequency band of the i-th window, calculated as follows:

[0170] ;

[0171] Two-scale consistency statistics and diagnostic indicators:

[0172] The Pearson correlation coefficient operator is used to calculate the residual sequence in the parameter domain. With microstructure residual energy sequence E i The linear correlation.

[0173] For the parameter domain residual sequence over k windows With microstructure residual energy sequence E i The correlation coefficient is obtained using the Pearson correlation coefficient operator.

[0174] The integrity diagnostic indicator is calculated as follows:

[0175] ;

[0176] The thresholds used for classification determination (configurable or obtained through factory statistical learning) are the first threshold θ1 and the second threshold θ2.

[0177] The specific steps of this method are described in detail below:

[0178] Step 1: Construct the ideal AC / DC ratio expected model ;

[0179] Factory calibration data and their physical meaning:

[0180] During the factory calibration phase, WVSS acquires output sequences for each R-curve type R under multiple combinations of PI settings π and TLC scan τ, calculates the corresponding ρ, and thus obtains a deterministic mapping. This mapping reflects the combined effects of optical path efficiency, gain link linearity, and higher-order nonlinearities.

[0181] To facilitate real-time calculations on the firmware side, this invention employs an interpretable analytical model:

[0182] It is a positive real number, representing the TLC normalization constant (fixed by the firmware). is a positive real number, representing the equivalent optical path efficiency coefficient of type R under ideal tooling. It is a positive real number, representing the attenuation / saturation tendency coefficient caused by gain control and link nonlinearity. is a real number representing the coefficient of higher-order nonlinear terms (such as LED thermal effects and quadratic terms in amplification links).

[0183] Expected model formula and derivation logic:

[0184] For most WVSS PPG output links, the AC / DC ratio ρ increases approximately linearly with PI in the small signal region; however, when TLC upregulation causes the link to become nonlinear or saturate, the gain of the AC component relative to DC exhibits an exponential decay trend; simultaneously, a quadratic term shift exists under larger PI. Therefore, the following model is given:

[0185] (1);

[0186] This model satisfies:

[0187] As τ increases and the link tends to be nonlinear, the exponential term can describe the decreasing or non-monotonic trend of the effective AC / DC ratio.

[0188] As π increases, the quadratic term can absorb the thermal effect and the offset caused by nonlinear modulation;

[0189] coefficient Each R is uniquely identified and fixed in the firmware calibration table.

[0190] Online calculation method:

[0191] At runtime, WVSS reads the current settings R, π, τ, and loads the corresponding values ​​from non-volatile memory. Substituting into equation (1) yields If the firmware uses discrete lookup tables, then equation (1) is considered an equivalent approximation of the lookup result, and can be obtained online through bilinear or trilinear interpolation. The present invention does not limit the interpolation method and can be adjusted according to the specific implementation method.

[0192] Step 2: Adaptive pulse template reconstruction and generation of microstructure residuals :

[0193] The principal period is aligned with the phase:

[0194] T is a positive real number representing the main period length in terms of sampling points, and is calculated as follows:

[0195] (2);

[0196] Where v is the HR setting value. To ensure phase consistency of the template, in the first... First, estimate the pulse initiation phase shift within the window:

[0197] is a real number representing the phase offset of the i-th window (in terms of the number of samples). The value can be obtained by maximizing the autocorrelation or the cross-correlation with the coarse template. For example, an ideal sine wave can be generated using a pulse rate setpoint as a coarse template, the sliding cross-correlation can be calculated, and the offset corresponding to the maximum value can be taken. This invention does not limit its specific implementation, but only requires that a consistent periodic slice be obtained within each window.

[0198] Periodic normalized sampling and template generation:

[0199] M is a natural number representing the number of template sampling points (this parameter, the number of template sampling points, is preset as a constant in the firmware of the WVSS device and will not change during normal device operation; users cannot dynamically adjust it through the interface or external commands).

[0200] Let be a real number, representing a periodic segment extracted and resampled from one period within the i-th window. .

[0201] Take a sequence segment of length T The result was obtained through uniform resampling:

[0202] (3);

[0203] For real numbers, representing the... The standard template obtained by smoothing can be obtained by cubic spline interpolation or low-pass filtering.

[0204] Refactoring of Loop Templates:

[0205] Define the reconstruction sequence To periodically spread and smoothly transition the template:

[0206] (4);

[0207] in:

[0208] Let be a natural number, representing the number of periods covered in the i-th window, such that qT covers... interval; In implementation, a segmented window function can be used to smoothly connect the weights at the boundaries of each cycle.

[0209] The final microstructure residuals are obtained as follows:

[0210] (5);

[0211] Derivation Logic Explanation: Template Reconstruction It absorbs the ideal pulse morphology dominated by R and v, residual This mainly includes non-ideal details introduced by the hardware link, such as rising edge overshoot, phase distortion, modulation of quantization noise, and peaks / peaks caused by saturation shearing.

[0212] Among them, the ideal pulse morphology refers to the core characteristics of a standard PPG pulse waveform without hardware link distortion, determined by the R-curve type and pulse rate setting of the Wearable Vital Signs Simulator (WVSS). It is the basic pulse waveform that WVSS should output according to user-defined parameters, without any non-ideal deviations introduced by hardware faults, and is also the template reconstruction. To accurately fit and extract the core waveform features.

[0213] Step 3: Calculate the critical band residual energy E i ;

[0214] Will Transform to the frequency domain and calculate the energy of the key frequency band:

[0215] (6);

[0216] Select To cover the key frequency bands of the heart rate fundamental frequency and its harmonics, they are mapped to indices k1 and k2:

[0217] (7);

[0218] Calculate residual energy:

[0219] (8);

[0220] Derivation logic explanation: Link failures often introduce distortion energy in the pulsation-related frequency band (e.g., abnormal harmonic enhancement, phase distortion leading to energy diffusion), therefore E i It can serve as an interpretable measure of the intensity of non-parametric structural distortions.

[0221] The non-parametric structural distortion intensity refers to the degree of regular distortion at the microstructure level of the PPG waveform, caused solely by optical / electronic hardware link failures / degradation, independent of the influence of WVSS user-defined parameters (R curve type, PI pulse amplitude, TLC transmission control, v pulse rate, etc.). This is the core quantitative concept proposed in this invention to distinguish between normal parameter adjustments and hardware fault distortion, while (microstructure distortion intensity) is a specific numerical measure of this intensity. The core of this concept lies in its two main characteristics: non-parametric and structural. These two characteristics jointly define the detection boundary of the invention and are key to its superiority over traditional single-index threshold methods. Non-parametric means that the distortion is independent of user-defined parameters and strictly distinguished from parametric changes. In this invention, adaptive template reconstruction strips away the ideal principal form dominated by R and v, ensuring that the microstructure residual sequence contains only hardware distortion, thus achieving non-parametric measurement from the root. Structural means that the distortion is not random noise such as circuit thermal noise or electromagnetic interference, but rather a regular structural deviation at the waveform microstructure level caused by physical hardware link failures. Non-parametric structural distortion intensity means disregarding all user-defined simulator parameter settings and measuring only the severity of regular distortions in the pulse waveform details caused by hardware malfunctions; while the invention uses E... i To assign a specific numerical value to the severity of this deformation, therefore... This is an interpretable measure of the intensity, and the distortion energy of the link failure must be concentrated in the pulsation-related frequency band, which is a completely valid causal relationship.

[0222] Step 4: Calculate the two-scale correlation coefficient C and the diagnostic index D:

[0223] Correlation coefficient formula:

[0224] The sequence ρ is obtained by applying K consecutive windows. i With E i Its Pearson correlation coefficient is defined as:

[0225] For real numbers, representing The mean is calculated as follows:

[0226] ;

[0227] For E i The mean is calculated as follows:

[0228] ;

[0229] The correlation coefficient is calculated as follows:

[0230] (9)

[0231] Define integrity diagnostic indicators:

[0232] (10);

[0233] Why consistency can characterize the integrity of a link:

[0234] When the link is intact and there is only a slight, interpretable gain drift, the parameter domain residuals Changes will cause microstructure residual energy E i The two undergo synergistic changes and exhibit a stable correlation.

[0235] When the fault enters the structural stages such as saturation shear, coupling offset, and phase distortion, With E i The physical coupling relationship is disrupted, the correlation decreases significantly or reverses, and thus D decreases.

[0236] Therefore, D can be used as a statistic of link integrity.

[0237] Step 5: Hierarchical Determination and Interpretable Location Rules:

[0238] Grading determination:

[0239] Given a first threshold θ1 and a second threshold θ2, and satisfying θ1 > θ2;

[0240] When D≥θ1: the link is considered complete;

[0241] When θ2≤D<θ1: the link is determined to be degraded;

[0242] When D < θ2: a link failure is determined;

[0243] Where D represents the link integrity diagnostic metric.

[0244] To avoid relying on empirical thresholds, this invention provides a traceable and reliable threshold determination method: the threshold is obtained statistically from qualified samples during the factory calibration stage or periodic review stage, and can be given in the form of quantiles and confidence levels.

[0245] Quantile thresholds based on qualified calibration samples:

[0246] Define the set of calibration samples under the qualified state (link integrity):

[0247] is a natural number, representing the total number of windows containing qualified samples;

[0248] The diagnostic indicators are calculated for the m-th qualified sample window according to equations (9) to (10). ;

[0249] For the set of qualified sample indicators, .

[0250] Constructing the empirical distribution function of qualified samples :

[0251] (12);

[0252] in This is an indicator function.

[0253] Define the empirical quantile function :

[0254] (13);

[0255] set up To meet the predetermined false alarm control level The threshold is then determined by the following formula:

[0256] (14);

[0257] From equation (14), we can obtain that, under the premise that the distribution of qualified states remains unchanged, the false alarm probability satisfies:

[0258] (15);

[0259] in This represents the probability under the condition that the link is qualified.

[0260] To cover normal fluctuations caused by different parameter combinations, this invention allows for... Layered by set parameters: for example, for each R curve type Establish separately And obtain according to equations (12)–(14) ; or to Thresholds are established for each gridded interval. This invention does not limit the layer dimension, but requires that the thresholds originate from traceable, qualified calibration samples.

[0261] Confidence correction for quantile thresholds (to avoid randomness caused by insufficient sample size):

[0262] To provide a threshold for the significance of confidence levels with a limited number of samples, this invention further defines:

[0263] Confidence risk determined for the threshold (set by the quality system);

[0264] To The order statistics after ascending sort.

[0265] is the rank used to correct the quantile threshold from empirical values ​​to confidence boundaries.

[0266] Let the random variable Y be distributed as follows: Select the smallest rank. satisfy:

[0267] (16);

[0268] Among them, parameters In the process of threshold confidence correction, the probability value of a random variable Y that follows a binomial distribution taking a value greater than or equal to rank r is the core probability judgment index used to screen the minimum rank r that satisfies the pre-set confidence risk δ. It only applies to the confidence correction steps of the first threshold θ1 and the second threshold θ2 in this invention.

[0269] The threshold with confidence significance is defined as follows:

[0270] (17);

[0271] And take this as a basis:

[0272] (18);

[0273] Equations (16)–(18) show that the threshold is given by the order statistic of the qualified samples and is explicitly tied to the false alarm control level. and With confidence risk δ, the threshold is made to have an auditable, recalculated, and traceable statistical meaning.

[0274] For the first threshold θ1, the minimum rank r1 is selected to satisfy:

[0275] ;

[0276] The r1-th value D(r1) in the order statistics is taken as the first threshold θ1;

[0277] For the second threshold θ2, the minimum rank r2 is selected to satisfy:

[0278] ;

[0279] The r2th value D(r2) in the order statistics is used as the second threshold θ2.

[0280] Interpretive localization (based on residual sign and trend):

[0281] To map the algorithm output to understandable physical causes, we define:

[0282] The value is a real number representing the energy deviation, and it is calculated as follows:

[0283] (11);

[0284] Combination and The sign and stability of [the symbol] can be used to give interpretive rules:

[0285] like Long-term positive and E i Synchronous rise: better suited to the harmonic enhancement-type degradation introduced by drive overshoot / insufficient bandwidth;

[0286] like Long-term negative and E i Synchronous descent: more consistent with degradation caused by coupling attenuation / decreased detection sensitivity leading to signal blunting;

[0287] like The absolute value increases while E i Instead, it decreases: This is more consistent with faults caused by saturation shearing / limiting that smooth out waveform details;

[0288] If E i Unstable transitions occur in the critical frequency band. Not significant: More consistent with sampling jitter / time base instability or phase distortion faults.

[0289] The fault types include:

[0290] when Greater than 0 and When the value is greater than 0, the output gain is nonlinear and faulty.

[0291] when Greater than 0 and When the value is less than 0, a saturation shear fault is output.

[0292] when Less than 0 and When the value is less than 0, there is an output coupling attenuation fault;

[0293] when Less than 0 and When the value is greater than 0, there is a stray interference fault in the output optical path;

[0294] When the parameter domain residual The absolute value is less than the preset first distortion threshold and the microstructure distortion intensity E i If the magnitude of the change within multiple consecutive diagnostic time windows exceeds the preset second distortion threshold, a sampling phase distortion fault is output.

[0295] The method also includes outputting the integrity status and / or the fault type to a display device in a visual form, or generating a diagnostic report that includes diagnostic time, diagnostic indicators and judgment results.

[0296] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0297] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator, characterized in that, The method includes: Step 1: Obtain the current setting parameters of the wearable vital signs simulator, including pulse curve type, pulse amplitude setting value, transmission control parameters, and pulse rate setting value; Step 2: Based on the set parameters and the pre-stored calibration coefficients corresponding to the pulse curve type, calculate the expected value of the ratio of AC component to DC component; Step 3: Acquire the PPG digital sequence output by the wearable vital signs simulator within multiple consecutive diagnostic time windows. For each diagnostic time window, perform parameter domain residual and microstructure distortion intensity calculation operations to obtain a corresponding parameter domain residual and a corresponding microstructure distortion intensity. The parameter domain residual and microstructure distortion intensity calculation operations include: The measured value of the ratio of AC component to DC component is calculated based on the PPG digital sequence within the current diagnostic time window, and a parameter domain residual is generated based on the difference between the measured value and the expected value. Based on the pulse rate setting value, the PPG digital sequence within the current diagnostic time window is reconstructed using a pulse waveform template to obtain a reconstructed sequence. A microstructure residual sequence is then generated based on the difference between the PPG digital sequence and the reconstructed sequence. The microstructure residual sequence is subjected to frequency domain transformation, and the energy of the microstructure residual sequence in a preset key frequency band is calculated as the microstructure distortion intensity. Step 4: Arrange the parameter domain residuals corresponding to the multiple consecutive diagnostic time windows in chronological order to form a parameter domain residual sequence; arrange the microstructure distortion intensity corresponding to the multiple consecutive diagnostic time windows in chronological order to form a microstructure distortion intensity sequence. Step 5: Calculate the correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence, and use the absolute value of the correlation coefficient as the link integrity diagnostic index; Step 6: Compare the link integrity diagnostic indicators with preset thresholds, and determine the integrity status of the link based on the comparison results.

2. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The expected value of the ratio of AC component to DC component is calculated as follows: ； in, R represents the expected value of the ratio of the AC component to the DC component, and R is the pulse curve type. Set a value for the pulse amplitude. To transmit control parameters, The normalization constant is This represents the equivalent optical path efficiency coefficient of type R under ideal tooling conditions. This is the attenuation / saturation tendency coefficient caused by gain control and link nonlinearity. These are the coefficients of higher-order nonlinear terms.

3. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The specific steps for reconstructing the pulse waveform template include: Based on the pulse rate setpoint v and the sampling rate f s Determine the length T of the main period. ; Based on the main period length T, a waveform segment of a main period is extracted from the PPG digital sequence within the current diagnostic time window, and the waveform segment is resampled to obtain a normalized period template. The periodic template is smoothed. The smoothed periodic template is periodically expanded to generate a reconstructed sequence with the same length as the current diagnostic time window.

4. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The microstructure distortion intensity is calculated as follows: The power spectrum of the microstructure residual sequence is obtained by performing a discrete Fourier transform on the microstructure residual sequence. Based on the lower boundary frequency f1 and upper boundary frequency f2 of the preset key frequency band, determine the corresponding first frequency domain index k1 and second frequency domain index k2: ； ； Where N is the length of the microstructure residual sequence, f s Sampling rate; The sum of all power spectra within the range from the first frequency domain index k1 to the second frequency domain index k2 is calculated as the microstructure distortion intensity E. i : ； in, Let be the power spectrum of the microstructure residual sequence corresponding to the i-th diagnostic time window, k is the frequency domain index, and the subscript r indicates that the power spectrum originates from the microstructure residual sequence.

5. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The correlation coefficient is the Pearson correlation coefficient, and the link integrity diagnostic index is calculated as follows: ； Wherein, D is the link integrity diagnostic index, and C is the Pearson correlation coefficient between the parameter domain residual sequence and the microstructure distortion intensity sequence, calculated as follows: ； Where K is the number of consecutive diagnostic time windows. The parameter domain residual corresponding to the i-th diagnostic time window. for The mean, E i The intensity of microstructural distortion corresponding to the i-th diagnostic time window. For E i The mean.

6. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The preset thresholds include a first threshold θ1 and a second threshold θ2, and satisfy θ1 > θ2; When D≥θ1: the link is considered complete; When θ2≤D<θ1: the link is determined to be degraded; When D < θ2: a link failure is determined; Where D is the link integrity diagnostic indicator.

7. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 6, characterized in that, The first threshold and the second threshold are determined in the following manner: Multiple diagnostic indicator samples of the wearable vital signs simulator under the condition of a qualified link are collected to form a calibration sample set; An empirical distribution function is constructed based on the calibration sample set to determine the false alarm control level. and Corresponding empirical quantiles and ,in, ; The first threshold and the second threshold are obtained based on the empirical quantile: 。 8. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 7, characterized in that, The first and second thresholds are determined based on a preset confidence risk δ: The diagnostic indicators of the calibration sample set are sorted in ascending order to obtain the order statistics. ,in, The total number of windows for the calibration sample set; For the first threshold θ1, the minimum rank r1 is selected to satisfy: ； The r1-th value D(r1) in the order statistics is taken as the first threshold θ1; For the second threshold θ2, the minimum rank r2 is selected to satisfy: ； The r2th value D(r2) in the order statistic is used as the second threshold. .

9. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 1, characterized in that, The process of performing step 3 also includes: Calculate the mean value of the microstructure distortion intensity. Calculate the energy deviation for each diagnostic time window. , E i The intensity of microstructural distortion corresponding to the i-th diagnostic time window; According to the parameter domain residual With the energy deviation Output the corresponding fault type, which includes: when Greater than 0 and When the value is greater than 0, the output gain is nonlinear and faulty. when Greater than 0 and When the value is less than 0, a saturation shear fault is output. when Less than 0 and When the value is less than 0, there is an output coupling attenuation fault; when Less than 0 and When the value is greater than 0, there is a stray interference fault in the output optical path; When the parameter domain residual The absolute value is less than the preset first distortion threshold and the microstructure distortion intensity E i If the magnitude of the change within multiple consecutive diagnostic time windows exceeds the preset second distortion threshold, a sampling phase distortion fault is output.

10. The method for detecting faults in the photoplethysmography (PPG) link of a wearable vital signs simulator according to claim 9, characterized in that, The method further includes step 7 after step 6: The integrity status and / or the fault type are output to a display device in a visual form, or a diagnostic report containing diagnostic time, diagnostic indicators, and judgment results is generated.

Images