Embedded closed-loop calibration and fault-tolerant control method for animal-specific blood analyzers

Abstract

The present application relates to the embedded closed-loop calibration and fault-tolerant control method of the blood analyzer special for animals, and belongs to the technical field of blood medical instrument detection. The method comprises the following steps: obtaining a target value parameter of a species sample, inputting the target value parameter of the species sample to retrieve a complete target value vector of the corresponding species, and obtaining calibration self-checking control data; continuously and repeatedly measuring the same calibration control sample, and performing fault-tolerant comparison based on a species repeatability index threshold value pre-stored in a built-in multi-species target value library, and outputting a fault-tolerant control early warning scheme; extracting an out-of-group constituent feature higher than the repeatability index threshold value in a repeated measurement state, identifying an out-of-group measurement value, and performing window constraint judgment on the out-of-group measurement value; performing interpolation reconstruction on the identified out-of-group measurement value, outputting an adaptive species target value matching sequence, and combining a multi-level fault-tolerant triggering mechanism to write the output adaptive species target value matching sequence into a calibration fault-tolerant compensation period of the fault-tolerant control early warning scheme, set an internal inspection update time, and complete the closed-loop calibration fault-tolerant control.

Description

Technical Field

[0001] This invention belongs to the field of blood medical instrument testing technology, specifically relating to an embedded closed-loop calibration and fault-tolerant control method for animal-specific blood analyzers. Background Technology

[0002] Blood analyzers, as crucial in vitro diagnostic devices in clinical testing, are widely used in routine blood tests, inflammation screening, and disease-aided diagnosis. The accuracy and stability of their test results directly impact the reliability of clinical diagnostic decisions. To ensure testing accuracy, existing blood analyzers typically rely on periodic calibration mechanisms, using standard quality control samples or calibrators to correct parameters in the detection channels. However, in actual operation, factors such as optical system aging, detection channel sensitivity drift, flow path wear, environmental temperature fluctuations, and differences in samples from different species can cause systematic biases or random errors in the instruments, leading to increased fluctuations in test results and even the risk of misinterpretation.

[0003] In existing technologies, self-testing and fault-tolerant control are mostly achieved using fixed-period calibration or single-threshold judgment methods. This involves simply comparing the difference between the test result and a preset reference value to determine if calibration is needed. This approach has significant drawbacks: First, fixed-period calibration cannot be dynamically adjusted according to the actual operating status of the equipment, easily leading to over-calibration or calibration lag. Second, independent judgment of a single channel makes it difficult to identify coupling anomalies between multiple detection channels, and cannot effectively distinguish between transient disturbances and persistent drift. Third, the lack of structured analysis and anomaly index management for repeated measurement data makes it difficult to achieve true closed-loop fault-tolerant control.

[0004] Furthermore, in multi-species detection scenarios, the distribution characteristics and repeatability indicators of blood parameters differ among species. If a uniform threshold standard is still used for calibration, it can easily lead to false triggering or missed triggering. Existing systems typically do not perform refined modeling of species-specific target values ​​and lack long-term trend prediction and compensation mechanisms for instrument aging drift, resulting in static calibration baselines that cannot adapt to dynamic changes during equipment operation.

[0005] Therefore, it is necessary to propose an embedded closed-loop calibration and fault-tolerant control method based on multi-species target modeling, repeated measurement differential analysis, and multi-level fault-tolerant triggering mechanism. By constructing a dynamic calibration baseline, an adaptive tolerance adjustment mechanism, and a predictive internal test time scheduling strategy, the method can achieve real-time monitoring and self-optimization control of the blood analyzer's operating status, thereby improving the accuracy, stability, and reliability of the test results and the system's operation. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides an embedded closed-loop calibration and fault-tolerant control method for animal-specific blood analyzers. The objective of this invention can be achieved through the following technical solutions: S1: Obtain the target value parameters of species samples, construct an embedded integrated control architecture, input the target value parameters of the species samples to retrieve the complete target value vector of the corresponding species, and map the ideal calibration target value baseline to the blood analyzer calibration index to obtain calibration self-test control data; S2: Perform continuous repeated measurements on the same calibration control sample to generate a target value difference vector. Based on the pre-stored species repeatability index thresholds in the built-in multi-species target value library, perform fault-tolerant comparison, mark the target value difference vector, and output a fault-tolerant control early warning scheme. S3: Based on the target value difference vector, extract the outlier composition features of the repeated measurement state that are higher than the repeatability index threshold, identify the outlier measurement values, generate an anomaly index identifier set, and use the current valid measurement sequence as the elimination window to perform window constraint judgment on the outlier measurement values; S4: Construct a target value reconstruction matching model, interpolate and reconstruct the identified outlier measurement values, output an adaptive species target value matching sequence, and combine a multi-level fault-tolerant triggering mechanism to write the output adaptive species target value matching sequence into the calibration fault-tolerant compensation cycle of the fault-tolerant control early warning scheme, set the internal check update time, and complete the closed-loop calibration fault-tolerant control.

[0007] Specifically, the embedded integrated control architecture includes: a data acquisition layer, a signal conditioning layer, and an execution feedback layer; The data acquisition layer is used to acquire the temperature control status simulation signal output by the blood analyzer detection channel, convert the temperature control status simulation signal into a digital measurement sequence, and call the multi-species target value database interface to obtain the standard target value parameters of the corresponding species samples. The signal conditioning layer is used to couple and condition the digital measurement sequence to construct a multi-dimensional feature fusion vector. The coupling conditioning methods include baseline correction, dynamic gain compensation, and temperature drift correction; The execution feedback layer outputs compensation control commands based on the fault-tolerant control early warning scheme, drives the calibration execution unit to adjust the gain, and feeds back the execution results to the data acquisition layer to form a closed-loop control path.

[0008] Specifically, the process for generating the calibration self-test control data is as follows: The ideal calibration target baseline is mapped to the blood analyzer calibration index, and a multidimensional calibration index matrix is ​​constructed. The multidimensional calibration index matrix establishes a calibration correspondence with the detection parameter dimension as the row vector and the channel feature parameter as the column vector. It performs a unified dimensional transformation on the calibration coefficients of different detection items to obtain a standardized calibration control vector. Threshold matching is performed through an embedded integrated control architecture, and calibration self-test control data is automatically output when the matching result exceeds the preset tolerance range. The threshold matching includes tolerance range determination, deviation identification, and abnormal gradient analysis.

[0009] Specifically, the steps for performing continuous repeated measurements on the same calibration control sample are as follows: Under constant detection conditions, the number of repeated measurements is set, and the sampling period and interval time are kept consistent to obtain a continuous set of sample measurements. Based on the continuous measurement set of the sample, the difference vector and standard deviation between adjacent measurement values ​​are calculated to complete the continuous repeated measurement of the same calibration control sample and generate the target value difference vector.

[0010] Specifically, the fault tolerance comparison method is as follows: Based on the corresponding species repeatability index thresholds in the built-in multi-species target value library, the correlation between multiple detection channels is jointly analyzed to identify single-channel anomalies and multi-channel coupling anomalies. The single-channel anomaly refers to an abnormal state in which the magnitude offset of the target value differential vector exceeds the repeatability index threshold of the corresponding species within a single detection channel, and the covariance relationship between this channel and other channels remains unchanged. The multi-channel coupling anomaly refers to an abnormal state in which the joint distribution of the target value differentiation vector in two or more channels deviates from the corresponding historical covariance structure. The tolerance of the target value difference vector is determined, and the tolerance range is adaptively adjusted according to the stability index of the current measurement sequence. When the difference measurement value exceeds the repeatability index threshold of the corresponding species, it is determined as an abnormal candidate sample and a graded fault tolerance label is generated.

[0011] Specifically, the output process of the fault-tolerant control early warning scheme is as follows: The hierarchical fault tolerance marker generated based on the fault tolerance comparison results is mapped to a fault tolerance decision vector, and the fault tolerance decision vector is input into the embedded fault tolerance decision engine to match the corresponding preset fault tolerance strategy template. The structured fault-tolerant decision vector includes anomaly level parameters and deviation magnitude coefficients; The embedded fault-tolerant decision engine performs weighted sorting of candidate strategies during the matching process, selects the optimal compensation path as the fault-tolerant control early warning scheme, packages the fault-tolerant control early warning scheme into a fault-tolerant control early warning data package, and writes it into the control buffer through the embedded integrated control architecture, triggering the execution feedback layer to prepare to execute the corresponding calibration compensation action.

[0012] Specifically, the method for generating the abnormal index identifier set is as follows: An anomaly persistence determination function is established to extract outlier composition features that exceed the repeatability index threshold under repeated measurement conditions, distinguish between transient anomalies and persistent anomalies, and establish an index mapping relationship for continuous measurement sequences to generate an anomaly index identifier set. The anomaly index identifier set includes an anomaly location index, an anomaly intensity parameter, and a duration marker.

[0013] Specifically, the execution process of the window constraint judgment is as follows: Set minimum effective sample size threshold, maximum allowable anomaly ratio threshold, and window span parameters to perform segmented statistical analysis on continuous measurement data; The segmented statistical analysis is used to count the number of valid measurement points, the number of abnormal candidate samples, and the mean of abnormal intensity within each sliding window, and to calculate the window stability index and the abnormality density coefficient. The window stability index is compared with a preset stability range. When the window stability index is higher than the stability threshold and the abnormality ratio is lower than the maximum allowable abnormality ratio threshold, it is determined to be an executable window to be removed, and the abnormal index identifier set is filtered by position constraints within the window.

[0014] Specifically, the method for constructing the target value reconstruction matching model is as follows: Based on the outlier matching sequence after window constraint judgment, a two-layer target value reconstruction matching model is constructed. The target value reconstruction and matching model includes a data reconstruction layer and a target value matching layer; The data reconstruction layer performs spectral energy recovery on the periodic fluctuation components for the locations of identified and removed outlier measurements, generating a reconstructed measurement vector. The target matching layer calculates the matching score index between the reconstructed measurement vector and the standard target vector based on the species-specific target baseline, and outputs the optimal matching result by adjusting the residual confidence between the target values.

[0015] Specifically, the multi-level fault-tolerant triggering mechanism includes: a deviation early warning judgment layer, a target value anomaly confirmation layer, and a calibration compensation execution layer; The deviation warning judgment layer is used to perform primary screening and rapid response judgment on the input fault tolerance decision vector. When the deviation amplitude is between the basic tolerance range and the warning tolerance range, a first-level warning signal is generated, and the abnormal timestamp and detection channel number are recorded. The target value anomaly confirmation layer is used to confirm and analyze anomalies that have triggered warnings. When an anomaly meets the persistence determination condition or the joint anomaly score index exceeds the set confirmation threshold, it is determined to be a valid anomaly event and an anomaly type code and an anomaly level confirmation identifier are generated. The calibration compensation execution layer is used to call the corresponding fault tolerance strategy template based on the anomaly confirmation result; The fault tolerance strategy template includes a gain compensation coefficient, a threshold correction amount, and a calibration cycle adjustment parameter.

[0016] Specifically, the method for setting the internal inspection update time is as follows: Using the abnormal trigger frequency and environmental drift coefficient as risk correction factors, the internal test trigger time parameter is output by superimposing the current running time and the self-test interval value. The range of the internal inspection trigger time parameter is limited to between the preset minimum internal inspection cycle and the maximum internal inspection cycle, and the internal inspection update time is set by constraint control through a nonlinear adjustment function. When the nonlinear adjustment function is in a stable operating state, it gradually converges the internal inspection update time towards the maximum internal inspection cycle; when an anomaly occurs, it compresses the internal inspection update time towards the minimum internal inspection cycle.

[0017] The beneficial effects of this invention are as follows: This invention achieves refined structural analysis of blood analyzer test results by constructing a multi-species target value modeling system and a repeated measurement differential analysis mechanism. Compared with existing fixed threshold judgment methods, it can significantly improve the accuracy and stability of anomaly identification. By introducing a multi-detection channel correlation joint analysis model, it effectively distinguishes between single-channel random disturbances and multi-channel coupling anomalies, reduces false triggering rate, and enhances the system's ability to identify systematic drift and structural mismatch.

[0018] A sliding window constraint judgment and anomaly index structured management mechanism are adopted to automatically remove outliers and dynamically reconstruct and match them while ensuring data continuity, thereby improving the statistical consistency and reliability of the measurement sequence. By constructing a two-layer target value reconstruction and matching model and a hierarchical fault-tolerant triggering mechanism, the early warning judgment, anomaly confirmation, and compensation execution form a layered decoupled structure, enhancing the accuracy and response efficiency of fault-tolerant control.

[0019] By introducing a system health assessment model and an anomaly risk prediction mechanism, the internal test update time is adaptively and dynamically adjusted, avoiding resource waste or response lag issues caused by fixed-period calibration. Through compensation effect feedback and a recursive update mechanism for model weights, a self-learning closed-loop calibration control process is formed, enabling the system to continuously optimize as the operating environment and equipment aging status change. Overall, this invention significantly improves the detection accuracy, operational stability, and intelligent fault tolerance of blood analyzers. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a schematic diagram of the embedded closed-loop calibration and fault-tolerant control method for the animal-specific blood analyzer of the present invention.

[0022] Figure 2 This is a schematic diagram illustrating the execution of the multi-level fault-tolerant triggering mechanism in the embedded closed-loop calibration and fault-tolerant control method of the animal-specific blood analyzer of the present invention. Detailed Implementation

[0023] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0024] Please see Figure 1 Embedded closed-loop calibration and fault-tolerant control method for animal-specific blood analyzers: S1: Obtain the target value parameters of species samples, construct an embedded integrated control architecture, input the target value parameters of the species samples to retrieve the complete target value vector of the corresponding species, and map the ideal calibration target value baseline to the blood analyzer calibration index to obtain calibration self-test control data; S2: Perform continuous repeated measurements on the same calibration control sample to generate a target value difference vector. Based on the pre-stored species repeatability index thresholds in the built-in multi-species target value library, perform fault-tolerant comparison, mark the target value difference vector, and output a fault-tolerant control early warning scheme. S3: Based on the target value difference vector, extract the outlier composition features of the repeated measurement state that are higher than the repeatability index threshold, identify the outlier measurement values, generate an anomaly index identifier set, and use the current valid measurement sequence as the elimination window to perform window constraint judgment on the outlier measurement values; S4: Construct a target value reconstruction matching model, interpolate and reconstruct the identified outlier measurement values, output an adaptive species target value matching sequence, and combine a multi-level fault-tolerant triggering mechanism to write the output adaptive species target value matching sequence into the calibration fault-tolerant compensation cycle of the fault-tolerant control early warning scheme, set the internal check update time, and complete the closed-loop calibration fault-tolerant control.

[0025] In this embodiment, the embedded integrated control architecture includes: a data acquisition layer, a signal conditioning layer, and an execution feedback layer; The data acquisition layer is used to acquire the temperature control status simulation signal output by the blood analyzer detection channel, convert the temperature control status simulation signal into a digital measurement sequence, and call the multi-species target value database interface to obtain the standard target value parameters of the corresponding species samples. The signal conditioning layer is used to couple and condition the digital measurement sequence to construct a multi-dimensional feature fusion vector. The coupling conditioning methods include baseline correction, dynamic gain compensation, and temperature drift correction; The execution feedback layer outputs compensation control commands based on the fault-tolerant control early warning scheme, drives the calibration execution unit to adjust the gain, and feeds back the execution results to the data acquisition layer to form a closed-loop control path.

[0026] In this embodiment, the process of generating the calibration self-test control data is as follows: The ideal calibration target baseline is mapped to the blood analyzer calibration index, and a multidimensional calibration index matrix is ​​constructed. The multidimensional calibration index matrix establishes a calibration correspondence with the detection parameter dimension as the row vector and the channel feature parameter as the column vector. It performs a unified dimensional transformation on the calibration coefficients of different detection items to obtain a standardized calibration control vector. Threshold matching is performed through an embedded integrated control architecture, and calibration self-test control data is automatically output when the matching result exceeds the preset tolerance range. The threshold matching includes tolerance range determination, deviation identification, and abnormal gradient analysis.

[0027] In this embodiment, the step of performing continuous repeated measurements on the same calibration control sample is as follows: Under constant detection conditions, the number of repeated measurements is set, and the sampling period and interval time are kept consistent to obtain a continuous set of sample measurements. Based on the continuous measurement set of the sample, the difference vector and standard deviation between adjacent measurement values ​​are calculated to complete the continuous repeated measurement of the same calibration control sample and generate the target value difference vector.

[0028] In this embodiment, an animal-specific blood analyzer from a pet hospital is used as an example. It is primarily used to analyze blood parameters in pets, including red blood cell count, white blood cell count, hemoglobin, and platelet count. The hospital processes approximately multiple samples daily, facing challenges such as target value differences across species, environmental interference, and instrument aging. The system aims to achieve closed-loop calibration and fault-tolerant control through an embedded microcontroller.

[0029] The system hardware includes: a multi-channel optical sensor, a direct-drive sample pump, a built-in multi-species target library, and an LCD display.

[0030] Specific technical solution: Obtain target parameters for species samples, construct an embedded integrated control architecture, and obtain calibration self-test control data; Input the species sample target parameter (e.g., species: cat, parameter: RBC) through the user interface (touchscreen). Construct an embedded integrated control architecture, including a data acquisition layer (acquiring analog signals of temperature control status, converting them into digital measurement sequences, and calling the multi-species target database interface); a signal conditioning layer (baseline correction: subtracting background noise; dynamic gain compensation); and an execution feedback layer (outputting compensation commands, driving gain adjustment, and forming a closed loop). Ideal calibration target value baseline acquisition: Perform trend fitting on multi-batch standard sample data of the same species, construct a species-specific target value baseline function, obtain the confidence interval [lower limit - upper limit] through standardized mapping, introduce an instrument aging drift compensation parameter, and perform covariance consistency correction (adjust the variance to the historical mean) on abnormal batches; Calibration self-check control data generation: Map the baseline to a calibration index (index = (measurement - baseline) / standard deviation), construct a multi-dimensional calibration index matrix (rows: detection parameters such as RBC; columns: channel characteristics such as gain), perform unified dimension conversion (normalize to [0,1]), execute threshold matching, and output data outside the range: {species: 'cat', index: 0.04, self-check: 'normal'}.

[0031] Perform continuous repeated measurements on the same calibration control sample, generate a target value differentiation vector, perform fault-tolerant comparison, and output a fault-tolerant control warning scheme; Perform continuous repeated measurements on the same calibration control sample: Set the number of repetitions to be consistent with the sampling period interval, construct a time series matrix (rows: measurement values; columns: timestamps), calculate the difference vector (adjacent difference) and standard deviation; Generate a target value differentiation vector: vector = [measurement i - target value]; Fault-tolerant comparison: Based on the species repeatability index threshold, jointly analyze the channel correlation. For single-channel anomalies: amplitude deviation > threshold and covariance remains unchanged. For multi-channel coupling anomalies: the joint distribution deviates from the historical covariance. Adaptively adjust the tolerance interval, and generate a hierarchical fault-tolerant mark (low / medium / high) when exceeding the threshold; Output of the fault-tolerant control warning scheme: Map the mark to a decision vector, input it into an embedded fault-tolerant decision engine, and match a preset policy template (e.g.,'retest'). Package it into a data packet, write it into the cache, and trigger the execution feedback layer.

[0032] Based on the target value differentiation vector, extract the outlier composition features, generate an abnormal index identification set, and perform window constraint judgment; Extract outlier features: Establish an abnormal persistence judgment function (instantaneous: persistence < 2 times; persistent: ≥ 2 times), distinguish anomalies, and generate an abnormal index identification set (position index, intensity = offset, duration); Window constraint judgment: Set the minimum effective sample threshold, maximum anomaly ratio, and window span. Perform segmented statistics on continuous data: the number of valid points, the number of anomalies, and the intensity mean within the window. Calculate the stability index s = valid / total, and the anomaly density d = anomaly / W. If s > window constraint and d < P, it is determined as an executable exclusion window, and the identification set is screened.

[0033] A target value reconstruction matching model is constructed to interpolate and reconstruct outlier measurements, outputting an adaptive species target value matching sequence. Combined with a multi-level fault-tolerant triggering mechanism, closed-loop calibration fault-tolerant control is completed. Construct a target value reconstruction matching model: a two-layer structure, the data reconstruction layer: spectrum recovery (FFT to extract periodic components and reconstruct vectors), the target value matching layer: calculate the matching score and adjust the residual confidence; Output adaptive sequence. Multi-level fault-tolerant triggering mechanism: deviation early warning layer (initial screening, generating first-level signal); target value anomaly confirmation layer (confirming persistence or score > threshold); calibration compensation execution layer; Internal inspection update time settings: Risk correction factor f = anomaly frequency * drift coefficient, output time, non-linear adjustment function: t = min(max(t, min) cycle =5min), max cycle =30min); when stable, gradually increase the maximum value; when abnormal, compress the minimum value and write it into the early warning scheme compensation cycle to complete the closed loop (feedback to update the target value library).

[0034] In this embodiment, the fault tolerance comparison method is as follows: Based on the corresponding species repeatability index thresholds in the built-in multi-species target value library, the correlation between multiple detection channels is jointly analyzed to identify single-channel anomalies and multi-channel coupling anomalies. The single-channel anomaly refers to an abnormal state in which the magnitude offset of the target value differential vector exceeds the repeatability index threshold of the corresponding species within a single detection channel, and the covariance relationship between this channel and other channels remains unchanged. The multi-channel coupling anomaly refers to an abnormal state in which the joint distribution of the target value differentiation vector in two or more channels deviates from the corresponding historical covariance structure. The tolerance of the target value difference vector is determined, and the tolerance range is adaptively adjusted according to the stability index of the current measurement sequence. When the difference measurement value exceeds the repeatability index threshold of the corresponding species, it is determined as an abnormal candidate sample and a graded fault tolerance label is generated.

[0035] In this embodiment, the output process of the fault-tolerant control early warning scheme is as follows: The hierarchical fault tolerance marker generated based on the fault tolerance comparison results is mapped to a fault tolerance decision vector, and the fault tolerance decision vector is input into the embedded fault tolerance decision engine to match the corresponding preset fault tolerance strategy template. The structured fault-tolerant decision vector includes anomaly level parameters and deviation magnitude coefficients; The embedded fault-tolerant decision engine performs weighted sorting of candidate strategies during the matching process, selects the optimal compensation path as the fault-tolerant control early warning scheme, packages the fault-tolerant control early warning scheme into a fault-tolerant control early warning data package, and writes it into the control buffer through the embedded integrated control architecture, triggering the execution feedback layer to prepare to execute the corresponding calibration compensation action.

[0036] In this embodiment, the method for generating the abnormal index identifier set is as follows: An anomaly persistence determination function is established to extract outlier composition features that exceed the repeatability index threshold under repeated measurement conditions, distinguish between transient anomalies and persistent anomalies, and establish an index mapping relationship for continuous measurement sequences to generate an anomaly index identifier set. The anomaly index identifier set includes an anomaly location index, an anomaly intensity parameter, and a duration marker.

[0037] In this embodiment, the execution process of the window constraint judgment is as follows: Set minimum effective sample size threshold, maximum allowable anomaly ratio threshold, and window span parameters to perform segmented statistical analysis on continuous measurement data; The segmented statistical analysis is used to count the number of valid measurement points, the number of abnormal candidate samples, and the mean of abnormal intensity within each sliding window, and to calculate the window stability index and the abnormality density coefficient. The window stability index is compared with a preset stability range. When the window stability index is higher than the stability threshold and the abnormality ratio is lower than the maximum allowable abnormality ratio threshold, it is determined to be an executable window to be removed, and the abnormal index identifier set is filtered by position constraints within the window.

[0038] In this embodiment, the anomaly persistence determination function is used to characterize the duration and stable trend characteristics of anomaly events over time. It is constructed based on the anomaly timestamp sequence, anomaly intensity score index, and anomaly trigger frequency recorded in the anomaly index identifier set. First, the number of consecutive anomalies N is counted within a sliding time window. c Total duration of the anomaly T d and the mean of the abnormal interval Δt avg And calculate the mean anomaly intensity S avg With the rate of change of intensity k s The specific calculation formula is as follows: , Where α, β, γ, and δ are weighting coefficients, and N th T th S th , Δt thThe corresponding preset judgment threshold parameters are: when Fp≥1, it is judged as a continuous anomaly; when Fp<1 and the peak value of the anomaly intensity exceeds the instantaneous disturbance threshold, it is judged as an instantaneous anomaly; when Fp is within the critical interval, it enters the delayed confirmation state and continues to collect data for recursive judgment.

[0039] In this embodiment, the method for constructing the target value reconstruction matching model is as follows: Based on the outlier matching sequence after window constraint judgment, a two-layer target value reconstruction matching model is constructed. The target value reconstruction and matching model includes a data reconstruction layer and a target value matching layer; The data reconstruction layer performs spectral energy recovery on the periodic fluctuation components for the locations of identified and removed outlier measurements, generating a reconstructed measurement vector. The target matching layer calculates the matching score index between the reconstructed measurement vector and the standard target vector based on the species-specific target baseline, and outputs the optimal matching result by adjusting the residual confidence between the target values.

[0040] In this embodiment, a calibration control sample of a certain species is used as the object. The same sample is subjected to N=6 consecutive repeated measurements. Assuming the blood analyzer contains m detection channels, the mean value of the j-th detection channel across N measurements is: , The corresponding sample standard deviation is: , Where σj is the sample standard deviation of the j-th detection channel in N repeated measurements. To sum the results of the first to Nth measurements, To average the cumulative results, N-1 represents the sample degrees of freedom, used to construct the sample variance of the unbiased estimate.

[0041] In this embodiment, as Figure 2 The multi-level fault-tolerant triggering mechanism shown includes: a deviation early warning judgment layer, a target value anomaly confirmation layer, and a calibration compensation execution layer; The deviation warning judgment layer is used to perform primary screening and rapid response judgment on the input fault tolerance decision vector. When the deviation amplitude is between the basic tolerance range and the warning tolerance range, a first-level warning signal is generated, and the abnormal timestamp and detection channel number are recorded. The target value anomaly confirmation layer is used to confirm and analyze anomalies that have triggered warnings. When an anomaly meets the persistence determination condition or the joint anomaly score index exceeds the set confirmation threshold, it is determined to be a valid anomaly event and an anomaly type code and an anomaly level confirmation identifier are generated. The calibration compensation execution layer is used to call the corresponding fault tolerance strategy template based on the anomaly confirmation result; The fault tolerance strategy template includes a gain compensation coefficient, a threshold correction amount, and a calibration cycle adjustment parameter.

[0042] In this embodiment, the method for setting the internal check update time is as follows: Using the abnormal trigger frequency and environmental drift coefficient as risk correction factors, the internal test trigger time parameter is output by superimposing the current running time and the self-test interval value. The range of the internal inspection trigger time parameter is limited to between the preset minimum internal inspection cycle and the maximum internal inspection cycle, and the internal inspection update time is set by constraint control through a nonlinear adjustment function. When the nonlinear adjustment function is in a stable operating state, it gradually converges the internal inspection update time towards the maximum internal inspection cycle; when an anomaly occurs, it compresses the internal inspection update time towards the minimum internal inspection cycle.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An embedded closed-loop calibration and fault-tolerant control method for an animal-specific blood analyzer, characterized in that, include: S1: Obtain the target value parameters of species samples, construct an embedded integrated control architecture, input the target value parameters of the species samples to retrieve the complete target value vector of the corresponding species, and map the ideal calibration target value baseline to the blood analyzer calibration index to obtain calibration self-test control data; S2: Perform continuous repeated measurements on the same calibration control sample to generate a target value difference vector. Based on the pre-stored species repeatability index thresholds in the built-in multi-species target value library, perform fault-tolerant comparison, mark the target value difference vector, and output a fault-tolerant control early warning scheme. S3: Based on the target value difference vector, extract the outlier composition features of the repeated measurement state that are higher than the repeatability index threshold, identify the outlier measurement values, generate an anomaly index identifier set, and use the current valid measurement sequence as the elimination window to perform window constraint judgment on the outlier measurement values; S4: Construct a target value reconstruction matching model, interpolate and reconstruct the identified outlier measurement values, output an adaptive species target value matching sequence, and combine a multi-level fault-tolerant triggering mechanism to write the output adaptive species target value matching sequence into the calibration fault-tolerant compensation cycle of the fault-tolerant control early warning scheme, set the internal check update time, and complete the closed-loop calibration fault-tolerant control.

2. The method according to claim 1, characterized in that, The embedded integrated control architecture includes: a data acquisition layer, a signal conditioning layer, and an execution feedback layer; The data acquisition layer is used to acquire the temperature control status simulation signal output by the blood analyzer detection channel, convert the temperature control status simulation signal into a digital measurement sequence, and call the multi-species target value database interface to obtain the standard target value parameters of the corresponding species samples. The signal conditioning layer is used to couple and condition the digital measurement sequence to construct a multi-dimensional feature fusion vector. The coupling conditioning methods include baseline correction, dynamic gain compensation, and temperature drift correction; The execution feedback layer outputs compensation control commands based on the fault-tolerant control early warning scheme, drives the calibration execution unit to adjust the gain, and feeds back the execution results to the data acquisition layer to form a closed-loop control path.

3. The method according to claim 2, characterized in that, The method for obtaining the ideal calibration target baseline is as follows: Trend fitting was performed on multiple batches of standard sample data of the same species to construct a species-specific target baseline function; Based on the species-specific target baseline function, the parameters of each detection channel are standardized and mapped to obtain the confidence interval range of the species-specific target value. Then, the instrument aging drift compensation parameter is introduced to perform covariance consistency correction on abnormal batch samples and generate an ideal calibration target baseline.

4. The method according to claim 1, characterized in that, The process for generating the calibration self-test control data is as follows: The ideal calibration target baseline is mapped to the blood analyzer calibration index, and a multidimensional calibration index matrix is ​​constructed. The multidimensional calibration index matrix establishes a calibration correspondence with the detection parameter dimension as the row vector and the channel feature parameter as the column vector. It performs a unified dimensional transformation on the calibration coefficients of different detection items to obtain a standardized calibration control vector. Threshold matching is performed through an embedded integrated control architecture, and calibration self-test control data is automatically output when the matching result exceeds the preset tolerance range. The threshold matching includes tolerance range determination, deviation identification, and abnormal gradient analysis.

5. The method according to claim 1, characterized in that, The steps for performing continuous repeated measurements on the same calibration control sample are as follows: Under constant detection conditions, the number of repeated measurements is set, and the sampling period and interval time are kept consistent to obtain a continuous set of sample measurements. Based on the continuous measurement set of the sample, the difference vector and standard deviation between adjacent measurement values ​​are calculated to complete the continuous repeated measurement of the same calibration control sample and generate the target value difference vector.

6. The method according to claim 1, characterized in that, The fault tolerance comparison method is as follows: Based on the corresponding species repeatability index thresholds in the built-in multi-species target value library, the correlation between multiple detection channels is jointly analyzed to identify single-channel anomalies and multi-channel coupling anomalies. The single-channel anomaly refers to an abnormal state in which the magnitude offset of the target value differential vector exceeds the repeatability index threshold of the corresponding species within a single detection channel, and the covariance relationship between this channel and other channels remains unchanged. The multi-channel coupling anomaly refers to an abnormal state in which the joint distribution of the target value differentiation vector in two or more channels deviates from the corresponding historical covariance structure. The tolerance of the target value difference vector is determined, and the tolerance range is adaptively adjusted according to the stability index of the current measurement sequence. When the difference measurement value exceeds the repeatability index threshold of the corresponding species, it is determined as an abnormal candidate sample and a graded fault tolerance label is generated.

7. The method according to claim 1, characterized in that, The output process of the fault-tolerant control early warning scheme is as follows: The hierarchical fault tolerance marker generated based on the fault tolerance comparison results is mapped to a fault tolerance decision vector, and the fault tolerance decision vector is input into the embedded fault tolerance decision engine to match the corresponding preset fault tolerance strategy template. The structured fault-tolerant decision vector includes anomaly level parameters and deviation magnitude coefficients; The embedded fault-tolerant decision engine performs weighted sorting of candidate strategies during the matching process, selects the optimal compensation path as the fault-tolerant control early warning scheme, packages the fault-tolerant control early warning scheme into a fault-tolerant control early warning data package, and writes it into the control buffer through the embedded integrated control architecture, triggering the execution feedback layer to prepare to execute the corresponding calibration compensation action.

8. The method according to claim 1, characterized in that, The method for generating the abnormal index identifier set is as follows: An anomaly persistence determination function is established to extract outlier composition features that exceed the repeatability index threshold under repeated measurement conditions, distinguish between transient anomalies and persistent anomalies, and establish an index mapping relationship for continuous measurement sequences to generate an anomaly index identifier set. The anomaly index identifier set includes an anomaly location index, an anomaly intensity parameter, and a duration marker.

9. The method according to claim 1, characterized in that, The execution process of the window constraint judgment is as follows: Set minimum effective sample size threshold, maximum allowable anomaly ratio threshold, and window span parameters to perform segmented statistical analysis on continuous measurement data; The segmented statistical analysis is used to count the number of valid measurement points, the number of abnormal candidate samples, and the mean of abnormal intensity within each sliding window, and to calculate the window stability index and the abnormality density coefficient. The window stability index is compared with a preset stability range. When the window stability index is higher than the stability threshold and the abnormality ratio is lower than the maximum allowable abnormality ratio threshold, it is determined to be an executable window to be removed, and the abnormal index identifier set is filtered by position constraints within the window.

10. The method according to claim 9, characterized in that, The method for constructing the target value reconstruction matching model is as follows: Based on the outlier matching sequence after window constraint judgment, a two-layer target value reconstruction matching model is constructed. The target value reconstruction and matching model includes a data reconstruction layer and a target value matching layer; The data reconstruction layer performs spectral energy recovery on the periodic fluctuation components for the locations of identified and removed outlier measurements, generating a reconstructed measurement vector. The target matching layer calculates the matching score index between the reconstructed measurement vector and the standard target vector based on the species-specific target baseline, and outputs the optimal matching result by adjusting the residual confidence between the target values.

11. The method according to claim 1, characterized in that, The multi-level fault-tolerant triggering mechanism includes: a deviation early warning judgment layer, a target value anomaly confirmation layer, and a calibration compensation execution layer; The deviation warning judgment layer is used to perform primary screening and rapid response judgment on the input fault tolerance decision vector. When the deviation amplitude is between the basic tolerance range and the warning tolerance range, a first-level warning signal is generated, and the abnormal timestamp and detection channel number are recorded. The target value anomaly confirmation layer is used to confirm and analyze anomalies that have triggered warnings. When an anomaly meets the persistence determination condition or the joint anomaly score index exceeds the set confirmation threshold, it is determined to be a valid anomaly event and an anomaly type code and an anomaly level confirmation identifier are generated. The calibration compensation execution layer is used to call the corresponding fault tolerance strategy template based on the anomaly confirmation result; The fault tolerance strategy template includes a gain compensation coefficient, a threshold correction amount, and a calibration cycle adjustment parameter.

12. The method according to claim 1, characterized in that, The method for setting the internal inspection update time is as follows: Using the abnormal trigger frequency and environmental drift coefficient as risk correction factors, the internal test trigger time parameter is output by superimposing the current running time and the self-test interval value. The range of the internal inspection trigger time parameter is limited to between the preset minimum internal inspection cycle and the maximum internal inspection cycle, and the internal inspection update time is set by constraint control through a nonlinear adjustment function. When the nonlinear adjustment function is in a stable operating state, it will gradually converge the internal inspection update time towards the maximum internal inspection cycle. When an anomaly occurs, the internal inspection update time will be compressed towards the minimum internal inspection cycle.

Images