Intelligent flow sensor test analysis method and system

Abstract

The application discloses a kind of intelligent flow sensor test analysis method and system, it is related to sensor test technical field, comprising: by standard flow generating device to the measured sensor applies step flow excitation signal, the dynamic response data of sensor is synchronously collected;Feature parameter set is extracted from response data;Similarity matching calculation is carried out, and the matching degree evaluation value of each error type is obtained;Quantify the induced relationship strength between different error types;Combined with matching degree evaluation value and induced relationship strength generates composite weight coefficient;Formulate multistage compensation strategy based on composite weight coefficient.The application has the advantages that: combined with historical data driven error correlation network modeling, the induced mechanism and interaction of different error types can be analyzed, the pertinence of compensation strategy is greatly improved, a multistage error suppression mechanism is formed, stable control of comprehensive error is realized under complex working conditions, and system maintenance requirements are reduced.

Description

Technical Field

[0001] This invention relates to the field of sensor testing technology, specifically to a method and system for testing and analyzing intelligent flow sensors. Background Technology

[0002] In existing technologies, the testing and calibration of flow sensors largely rely on static calibration methods under single operating conditions, making it difficult to effectively identify complex errors in the dynamic response process. Traditional methods typically employ fixed flow point testing or manual judgment, which cannot accurately separate the coupled errors of multiple factors such as steady-state deviation, transient overshoot, and environmental interference. Especially under complex operating conditions such as temperature fluctuations and mechanical vibrations, errors are prone to misjudgment and compensation lag, resulting in low calibration efficiency and increased maintenance costs. Furthermore, existing error diagnosis systems lack correlation mining of historical fault data, making it difficult to quantify the induced relationships between different error types. This limits compensation strategies to local parameter adjustments, hindering the achievement of systemic optimization.

[0003] To address the aforementioned shortcomings, there is an urgent need for an intelligent testing and analysis method that integrates dynamic stimulus testing, multi-dimensional feature analysis, and error correlation modeling. Current technologies have not yet solved the following core problems: first, the lack of a tiered flow stimulus and a synchronous testing mechanism for multiple environmental parameters leads to insufficient completeness of the feature parameter set; second, the absence of an error correlation network model based on historical data makes it difficult to assess the weighted impact of composite errors; and third, insufficient synergy between hardware compensation and software algorithm correction hinders the accurate suppression of multi-level errors. Therefore, it is imperative to overcome these technological bottlenecks through innovations in dynamic response feature extraction, error correlation strength quantification, and collaborative calibration strategies. Summary of the Invention

[0004] To address the aforementioned technical problems, this paper provides a test and analysis method and system for intelligent flow sensors. This technical solution solves the problem that existing error diagnosis systems lack correlation mining of historical fault data, making it difficult to quantify the induced relationships between different error types, thus limiting compensation strategies to local parameter adjustments and making it difficult to achieve systematic optimization.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A test and analysis method for intelligent flow sensors, comprising:

[0007] A stepped flow excitation signal is applied to the sensor under test using a standard flow generator, and the dynamic response data of the sensor is collected simultaneously.

[0008] Extract a set of feature parameters from the response data, including steady-state deviation, transient overshoot, and noise interference level.

[0009] The similarity matching calculation is performed between the set of feature parameters and the typical fault modes in the pre-stored error knowledge base to obtain the matching degree evaluation value of each error type.

[0010] An error correlation network model is constructed based on historical maintenance records to quantify the strength of the induced relationship between different error types;

[0011] By combining the matching degree evaluation value and the strength of the induced relationship, a composite weight coefficient is generated to determine the dominant error type and related influencing factors;

[0012] A multi-level compensation strategy is formulated based on composite weight coefficients, and a coordinated calibration is implemented by adjusting hardware parameters and correcting software algorithms.

[0013] Preferably, the extraction of a feature parameter set from the response data, including steady-state deviation, transient overshoot, and noise interference level, specifically includes:

[0014] Record the full-scale response curve under constant pressure test conditions and calculate the percentage deviation of the rise time from the theoretical value.

[0015] The zero-point output drift was measured in a variable temperature environment, and a linear relationship model between temperature change and zero-point offset was established.

[0016] High-frequency noise components are extracted by wavelet decomposition, and the energy proportion of a specific frequency band is used as an indicator of interference level.

[0017] Preferably, the step of performing similarity matching calculations between the feature parameter set and typical fault modes in the pre-stored error knowledge base to obtain matching degree evaluation values ​​for each error type specifically includes:

[0018] The basic matching degree is obtained by comparing the number of common features between the current feature parameter and the standard features of each fault mode with the number of elements in the union of the current feature parameter and the standard features of each fault mode.

[0019] Based on the degree to which the test environment temperature deviates from the standard operating conditions, the matching degree of temperature-sensitive errors is enhanced by a preset proportional coefficient to obtain the temperature correction value;

[0020] The final matching score is obtained by adding the basic matching score to the temperature correction value.

[0021] Preferably, the step of constructing an error correlation network model based on historical maintenance records and quantifying the strength of the induced relationship between different error types specifically includes:

[0022] Establish a directed graph of error types. In the directed graph, nodes represent error types, the direction of the directed edges represents the induced relationship between error types, and the weight of the directed edges represents the induced probability between error types.

[0023] Specifically, if error type i induces error type j, then a directed edge is constructed from error type i to error type j.

[0024] Preferably, the method for determining the directed edge weights is as follows:

[0025] Determine the number of records in the historical maintenance log that contain error type i;

[0026] Determine the number of records in the historical maintenance log that contain both error type i and error type j;

[0027] The probability of error type i inducing error type j is obtained by comparing the number of records that simultaneously exhibit error type i with the number of records that exhibit error type j.

[0028] Preferably, the step of combining the matching degree evaluation value and the induced relationship strength to generate a composite weight coefficient and determine the dominant error type and associated influencing factors specifically includes:

[0029] Obtain all error types pointing to the error type, and form a set of cause error types for the error type;

[0030] Obtain all error types pointed to by the error type, and form a set of induced error types of the error type;

[0031] Based on the matching degree evaluation values ​​of all error types and the strength of the induced relationship between each error type, the composite weight coefficient of each error type is calculated through the correlation fitting formula.

[0032] Specifically, the correlation fitting formula is as follows:

[0033]

[0034] Among them, R i Let S be the composite weight coefficient for the i-th error type, β be the direct matching weight, and S be the weight coefficient for the i-th error type. i N represents the matching score for the i-th error type. i Let j be the set of induced error types for the i-th error type, and let j be N. i The j-th element in S j For N i The matching score of the j-th element in the dataset, w ij For the i-th error type, pair N i The induced probability of the j-th element in the matrix, U j For N i The set of causal error types for the j-th element in the N-th matrix, where K is N. i The j-th element in the set of causal error types is the k-th element, w. kj For N iThe j-th element in the set of causal error types is the k-th element of N. i The probability of triggering the j-th element in the equation, where γ is the correlation influence factor.

[0035] Preferably, the collaborative calibration, which involves formulating a multi-level compensation strategy based on composite weighting coefficients and implementing hardware parameter adjustments and software algorithm corrections, specifically includes:

[0036] The error types whose composite weight coefficients exceed the compensation threshold are selected as compensation errors;

[0037] When the compensation error is dominated by temperature drift, dual-channel compensation is activated: the hardware adjusts the bridge excitation voltage and the software corrects the flow conversion coefficient.

[0038] When the compensation error is the cumulative error of mechanical wear, the self-learning compensation algorithm is activated to predict the wear trend curve based on historical calibration data.

[0039] When the compensation error is caused by strong vibration interference, switch to high-frequency sampling mode and enable digital filtering, while triggering an installation stability alarm.

[0040] Furthermore, this solution proposes an intelligent flow sensor testing and analysis method system to implement the intelligent flow sensor testing and analysis method described above, including:

[0041] The signal excitation module is used to apply a stepped flow excitation signal to the sensor under test through a standard flow generator.

[0042] The dynamic data acquisition module, which is triggered synchronously with the signal excitation module, is used to acquire the dynamic response data of the sensor in real time and generate a time-domain waveform sequence.

[0043] The feature parameter extraction module includes a steady-state analysis unit, a transient analysis unit, and a noise analysis unit. The steady-state analysis unit calculates the rise time percentage deviation through the full-scale response curve. The transient analysis unit establishes a zero-point drift-temperature linear model based on variable temperature environment testing. The noise analysis unit uses a wavelet decomposition algorithm to extract the high-frequency noise energy ratio index.

[0044] The error matching engine connects to the feature parameter extraction module and has a built-in pre-stored error knowledge base and similarity calculator. The similarity calculator performs feature intersection ratio calculation and superimposes temperature deviation compensation, and outputs the matching degree evaluation value of each fault mode.

[0045] The network modeling module integrates a historical maintenance database and constructs a weighted directed graph model. The weights are calculated through error co-occurrence probability statistics, reflecting the strength of the induced relationship between error types.

[0046] The composite weight decision-maker receives the evaluation value of the error matching engine and the induced probability data of the correlation network model. It then fuses the direct matching weights and correlation influencing factors through the correlation fitting formula to generate composite weight coefficients for the error type.

[0047] The multi-level collaborative calibration device, based on the composite weighting coefficient triggering compensation strategy selection logic, includes a temperature drift dual-channel compensation unit, a mechanical wear self-learning unit, and a vibration interference suppression unit. The temperature drift dual-channel compensation unit synchronously adjusts the hardware bridge voltage and the software conversion coefficient. The mechanical wear self-learning unit predicts the wear curve based on historical data and dynamically corrects the gain parameters. The vibration interference suppression unit switches the high-frequency sampling mode and activates the digital filter and installation stability alarm function.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0049] This invention significantly improves the systematicness and reliability of flow sensor error diagnosis through stepped dynamic excitation testing and multi-source data fusion analysis. Based on multi-modal flow excitation modes and synchronous monitoring of environmental parameters, it achieves dynamic separation and precise source tracing of complex errors, effectively enhancing the identification capabilities of steady-state deviation and transient overshoot. Combined with historical data-driven error correlation network modeling, it can analyze the inducing mechanisms and interaction relationships of different error types, greatly improving the targeting of compensation strategies. Through the collaborative optimization of hardware compensation modules and software adaptive algorithms, a multi-level error suppression mechanism is formed, achieving stable control of comprehensive errors under complex operating conditions while reducing system maintenance requirements. This technical solution overcomes the limitations of traditional calibration methods, providing a more adaptable error management path for industrial measurement and control scenarios. Attached Figure Description

[0050] Figure 1 This is a flowchart of the intelligent flow sensor testing and analysis method proposed in this solution;

[0051] Figure 2 This is a flowchart of the method for extracting feature parameter sets proposed in this scheme;

[0052] Figure 3 This is a flowchart illustrating the method proposed in this scheme for obtaining matching degree evaluation values ​​for each error type.

[0053] Figure 4 The flowchart shows the method for determining the weight of directed edges proposed in this scheme.

[0054] Figure 5 This is a flowchart illustrating the method proposed in this scheme for determining the dominant error type and associated influencing factors.

[0055] Figure 6This is a flowchart of the collaborative calibration method proposed in this scheme, which involves adjusting hardware parameters and correcting software algorithms.

[0056] Figure 7 This is an architecture diagram of the electronic devices in this solution;

[0057] Figure 8 This is a schematic diagram of the computer-readable storage medium structure in this scheme.

[0058] The numbers on the map are:

[0059] 500 - Electronic device; 501 - Bus; 502 - CPU; 503 - ROM; 504 - RAM; 505 - Communication port; 506 - Input / output component; 507 - Hard disk; 508 - User interface; 600 - Computer-readable storage medium. Detailed Implementation

[0060] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0061] Reference Figure 1 As shown, a test and analysis method for an intelligent flow sensor includes:

[0062] A stepped flow excitation signal is applied to the sensor under test using a standard flow generator, and the dynamic response data of the sensor is collected simultaneously.

[0063] A stepped flow excitation signal simulates sudden or step-like flow changes that a sensor might encounter in actual operating conditions by gradually changing the flow value (e.g., from low to high or alternating between high and low). The synchronous acquisition system must use a high-precision clock source to ensure timing alignment between the excitation signal and the response data, avoiding distortion in dynamic characteristic analysis due to sampling delays. The standard flow generator must have programmable control capabilities to accurately reproduce stepped waveforms of different amplitudes and durations.

[0064] Extract a set of feature parameters from the response data, including steady-state deviation, transient overshoot, and noise interference level.

[0065] Steady-state deviation is obtained by calculating the difference between the mean value of the response curve during the stepped plateau period and the theoretical value, reflecting the sensitivity decay or zero drift of the sensor after long-term operation; transient overshoot is determined by capturing the ratio of the maximum deviation amplitude at the moment of step switching to the steady-state value, and is used to evaluate the damping characteristics and dynamic response speed of the sensor; noise interference level is quantified by analyzing the standard deviation of signal fluctuations or the proportion of high-frequency spectrum energy in the steady-state stage, which quantifies the sensor's anti-electromagnetic interference capability and the stability of the signal processing circuit.

[0066] The similarity matching calculation is performed between the feature parameter set and the typical fault modes in the pre-stored error knowledge base to obtain the matching degree evaluation value of each error type.

[0067] The error knowledge base is built based on historical fault cases and experimental data, and includes the threshold range of characteristic parameters for typical fault modes such as "zero drift", "sensitivity decrease", and "abnormal circuit noise". Similarity matching adopts a weighted Euclidean distance or cosine similarity algorithm. According to the importance of parameters, such as the steady-state deviation weight being higher than the noise level, the comprehensive matching degree is calculated, and the probabilistic evaluation value of each fault mode is output. The output matching degree evaluation value is a value in the range of 0-1.

[0068] An error correlation network model is constructed based on historical maintenance records to quantify the strength of the induced relationship between different error types;

[0069] The error correlation network is represented by a graph structure, where nodes represent error types and edge weights represent the strength of the induced relationship. The correlation strength is calculated by statistically analyzing the co-occurrence frequency of errors in historical data, combined with fault tree analysis or Bayesian network methods. The model needs to be dynamically updated; when new maintenance data is added, the network weight parameters are optimized through incremental learning.

[0070] By combining the matching degree evaluation value and the strength of the induced relationship, a composite weight coefficient is generated to determine the dominant error type and related influencing factors;

[0071] A multi-level compensation strategy is formulated based on composite weight coefficients, and a coordinated calibration is implemented by adjusting hardware parameters and correcting software algorithms.

[0072] Multi-level compensation strategies are executed according to weighted order. For example, hardware adjustments are prioritized, such as adjusting the sensor bridge resistance to compensate for zero-point drift, followed by software filtering to suppress noise. Hardware parameter adjustments employ PID closed-loop control to automatically adjust potentiometers or digital fine-tuning circuits; software corrections utilize adaptive Kalman filtering or compensation functions based on error models. Co-calibration requires verification that the compensated characteristic parameters fall within acceptable thresholds; otherwise, an iterative optimization process is triggered.

[0073] Specifically, refer to Figure 2 As shown, the feature parameter set extracted from the response data, which includes steady-state deviation, transient overshoot, and noise interference level, specifically includes:

[0074] Record the full-scale response curve under constant pressure test conditions and calculate the percentage deviation of the rise time from the theoretical value.

[0075] The constant pressure test applies a stable pressure step signal (e.g., a 0-100% FS step change) to ensure the sensor records a complete dynamic response process under conditions free from external interference. The rise time percentage deviation is calculated as (measured rise time - theoretical rise time) / theoretical value × 100%, where the theoretical value is derived from the sensor's design specifications or factory calibration data. This indicator effectively characterizes changes in the sensor's mechanical damping characteristics or the bandwidth attenuation of the signal conditioning circuit. During testing, it is essential to ensure that the rise time steepness of the excitation signal (e.g., <1ms) is higher than the sensor's response capability, and to capture detailed waveforms at a sampling rate of at least 1kHz.

[0076] The zero-point output drift was measured in a variable temperature environment, and a linear relationship model between temperature change and zero-point offset was established.

[0077] Variable temperature environment testing is typically conducted in a temperature-controlled chamber, cycling within the range of -20℃ to 80℃ at a rate of 0.5-5℃ / min, while simultaneously recording temperature values ​​and zero-point output. The linear model is fitted using the least squares method as ΔV = k·ΔT + b, where the slope k reflects the temperature drift coefficient (μV / ℃), and the intercept b corresponds to the zero-point deviation at room temperature. During modeling, lagging data from abrupt temperature changes (such as data from the first 30 seconds after a temperature change) must be removed, and the residual distribution must be verified to conform to Gaussian properties to confirm the rationality of the linear assumption. This model provides key parameters for temperature compensation algorithms, such as using real-time temperature sampling values ​​to reverse-correct the zero-point output in embedded systems.

[0078] High-frequency noise components are extracted by wavelet decomposition, and the energy proportion of a specific frequency band is used as an indicator of interference level.

[0079] Wavelet decomposition employs Daubechies basis functions to perform 5-8 levels of multi-resolution analysis on the signal, separating high-frequency detail components above 1kHz. The energy percentage of a specific frequency band is calculated as: (Sum of squares of wavelet coefficients in the target frequency band / Total energy in the entire frequency band) × 100%. The frequency band range is set according to the characteristics of the interference source. When the energy percentage exceeds a threshold, it indicates an electromagnetic compatibility problem or mechanical resonance risk. This method is better at capturing transient interference than Fourier transform, and the analysis sensitivity can be improved by selecting wavelet bases that match the noise characteristics.

[0080] Reference Figure 3 As shown, the similarity matching calculation between the feature parameter set and typical fault modes in the pre-stored error knowledge base is performed to obtain the matching degree evaluation value for each error type. Specifically, this includes:

[0081] The basic matching degree is obtained by comparing the number of common features between the current feature parameter and the standard features of each fault mode with the number of elements in the union of the current feature parameter and the standard features of each fault mode.

[0082] Based on the degree to which the test environment temperature deviates from the standard operating conditions, the matching degree of temperature-sensitive errors is enhanced by a preset proportional coefficient to obtain the temperature correction value;

[0083] The final matching score is obtained by adding the basic matching score to the temperature correction value.

[0084] The specific calculation formula is as follows:

[0085]

[0086] Among them, S i Let F be the final matching evaluation value between the current feature parameters and the i-th fault mode, and G be the current feature parameter set. i Let be the standard feature set of the i-th fault mode, ΔT be the deviation of the current temperature from the standard operating condition, and α be the temperature sensitivity factor, with a value range of 0.05-0.2.

[0087] A matching degree calculation mechanism combining multi-dimensional feature comparison and dynamic environmental compensation significantly improves the accuracy and reliability of fault mode recognition and diagnosis. Based on the intersection and union operations of feature parameter sets and standard fault mode features, a quantitative evaluation framework for basic matching degree is established to ensure the objectivity of fault feature similarity judgment. A dynamic compensation algorithm for temperature deviation and sensitivity factors is introduced, which can adaptively adjust the matching weight of temperature-sensitive errors, effectively suppressing the risk of misjudgment caused by environmental interference. By superimposing and fusing basic matching degree and temperature correction values, a dual evaluation system that considers both static feature correlation and dynamic operating condition influences is formed, ensuring that fault diagnosis results conform to the matching rules of core parameters and reflect the transmission effect of environmental variables. This matching degree calculation model breaks through the limitations of traditional threshold criteria, providing a quantifiable decision-making basis for the parallel identification and hierarchical handling of multiple types of errors under complex operating conditions.

[0088] Based on historical maintenance records, an error correlation network model is constructed to quantify the strength of the induced relationship between different error types, specifically including:

[0089] Establish a directed graph of error types. In the directed graph, nodes represent error types, the direction of the directed edges represents the induced relationship between error types, and the weight of the directed edges represents the induced probability between error types.

[0090] Specifically, if error type i induces error type j, then a directed edge is constructed from error type i to error type j.

[0091] Specifically, refer to Figure 4 As shown, the method for determining the weight of a directed edge is as follows:

[0092] Determine the number of records in the historical maintenance log that contain error type i;

[0093] Determine the number of records in the historical maintenance log that contain both error type i and error type j;

[0094] The probability of error type i inducing error type j is obtained by comparing the number of records that simultaneously exhibit error type i with the number of records that exhibit error type j.

[0095] It is understandable that errors in flow sensing may exhibit correlations. For example, when temperature-sensitive errors, such as zero-point drift caused by thermal expansion, are not compensated for in a timely manner, they may further induce mechanical stress-type errors, such as range shifts caused by diaphragm deformation, through stress accumulation in the sensor's mechanical structure. Alternatively, signal distortion errors caused by electromagnetic interference, if not effectively suppressed, may lead to fluctuations in the reference voltage of the analog-to-digital conversion stage, thereby generating nonlinear response errors. Such error propagation chains are particularly significant under dynamic operating conditions. For instance, in high-temperature environments, temperature drift errors may appear before other errors and form compound faults through the sensor's internal coupling mechanism. This method constructs a causal error correlation network through statistical mining of historical maintenance data. Its core lies in revealing the potential inducing patterns between error types in a quantitative manner.

[0096] Reference Figure 5 As shown, by combining the matching degree evaluation value and the strength of the induced relationship to generate a composite weight coefficient, the dominant error type and associated influencing factors are determined, including:

[0097] Obtain all error types pointing to the error type, and form a set of cause error types for the error type;

[0098] Obtain all error types pointed to by the error type, and form a set of induced error types of the error type;

[0099] Based on the matching degree evaluation values ​​of all error types and the strength of the induced relationship between each error type, the composite weight coefficient of each error type is calculated through the correlation fitting formula.

[0100] Specifically, the correlation fitting formula is as follows:

[0101]

[0102] Among them, R i Let S be the composite weight coefficient for the i-th error type, β be the direct matching weight, and S be the weight coefficient for the i-th error type. i N represents the matching score for the i-th error type. i Let j be the set of induced error types for the i-th error type, and let j be N. i The j-th element in S j For N i The matching score of the j-th element in the dataset, w ij For the i-th error type, pair Ni The induced probability of the j-th element in the matrix, U j For N i The set of causal error types for the j-th element in the N-th matrix, where K is N. i The j-th element in the set of causal error types is the k-th element, w. kj For N i The j-th element in the set of causal error types is the k-th element of N. i The probability of triggering the j-th element in the equation, where γ is the correlation influence factor.

[0103] This composite weighting coefficient calculation model innovatively applies the core ideas of the PageRank algorithm to error correlation network analysis, quantifying the impact of error propagation by simulating the weight transfer mechanism between web pages. Its design principle can be analogized as follows: each error type is treated as a network node, the induced probability corresponds to the link weight between nodes, and the composite weighting coefficient characterizes the global importance of a node in the system. The formula reconstruction mainly reflects the following PageRank characteristics:

[0104] (1) Random walk mechanism: induced error superposition term Simulate user browsing behavior along directed edges, where the influence of the current error i is partly determined by the weight S of the downstream errors j it may trigger. j and trigger probability Joint decision;

[0105] (2) Weight normalization: The denominator is summed using the inverse probability of the induced error set. Implement a "balanced outgoing chain weight" logic similar to PageRank to avoid weight calculation bias caused by high in-degree nodes;

[0106] (3) Damping Factor Introduction: Direct matching weight β and correlation influence factor γ, where β and γ satisfy β + γ = 1. In some preferred embodiments, the value of β ranges from 0.6 to 0.8, and the value of γ ranges from 0.2 to 0.4, corresponding to the random jump probability in PageRank. This is used to balance the contribution ratio of direct matching (local evidence) and network propagation (global correlation), ensuring the stability of the model between strong correlation chains and independent errors. This design makes the weight evaluation of error types not only dependent on their own matching degree S i Furthermore, it dynamically captures the potential risks of the network topology as an "error propagation hub," making it particularly suitable for root cause tracing and priority ranking in scenarios with multiple concurrent errors.

[0107] Reference Figure 6 As shown, the multi-level compensation strategy based on composite weighting coefficients, and the collaborative calibration of hardware parameter adjustment and software algorithm correction specifically include:

[0108] The error types whose composite weight coefficients exceed the compensation threshold are selected as compensation errors;

[0109] When the compensation error is dominated by temperature drift, dual-channel compensation is activated: the hardware adjusts the bridge excitation voltage and the software corrects the flow conversion coefficient.

[0110] When the compensation error is the cumulative error of mechanical wear, the self-learning compensation algorithm is activated to predict the wear trend curve based on historical calibration data.

[0111] When the compensation error is caused by strong vibration interference, switch to high-frequency sampling mode and enable digital filtering, while triggering an installation stability alarm.

[0112] A multi-level collaborative compensation mechanism driven by composite weights significantly improves the error suppression efficiency and system robustness of flow sensors under complex operating conditions. The weighted selection strategy based on error correlation networks accurately identifies core error types with high propagation risk, prioritizing targeted compensation for root cause errors such as temperature drift and mechanical wear, effectively blocking the error propagation chain. The dual-channel collaborative design of hardware parameters and software algorithms achieves deep coupling between physical layer signal conditioning and data layer model correction. For example, in temperature drift compensation, dynamic adjustment of the hardware excitation voltage can quickly eliminate thermal stress distortion, while synchronous correction of the software conversion coefficient further ensures output linearity. The self-learning compensation algorithm for mechanical wear, by integrating historical trend prediction and real-time data feedback, forms a progressive calibration capability, avoiding over- or under-adjustment problems associated with traditional fixed compensation amounts. The linkage mechanism of high-frequency sampling and digital filtering balances signal fidelity and noise suppression requirements under strong vibration interference scenarios. This strategy overcomes the limitations of a single compensation mode, enabling the system to possess intelligent control characteristics of parallel processing of multiple errors and hierarchical resolution of primary and secondary contradictions, significantly enhancing adaptability to all operating conditions and calibration timeliness.

[0113] Furthermore, based on the same inventive concept as the above method, this solution proposes an intelligent flow sensor testing and analysis system, comprising:

[0114] The signal excitation module is used to apply a stepped flow excitation signal to the sensor under test through a standard flow generator.

[0115] The dynamic data acquisition module, which is triggered synchronously with the signal excitation module, is used to acquire the dynamic response data of the sensor in real time and generate a time-domain waveform sequence.

[0116] The feature parameter extraction module includes a steady-state analysis unit, a transient analysis unit, and a noise analysis unit. The steady-state analysis unit calculates the percentage deviation of rise time through the full-scale response curve. The transient analysis unit establishes a zero-point drift-temperature linear model based on variable temperature environment testing. The noise analysis unit uses wavelet decomposition algorithm to extract the high-frequency noise energy ratio index.

[0117] The error matching engine connects to the feature parameter extraction module and has a built-in pre-stored error knowledge base and similarity calculator. The similarity calculator performs feature intersection ratio calculation and superimposes temperature deviation compensation, outputting the matching degree evaluation value of each fault mode.

[0118] The network modeling module integrates a historical maintenance database and constructs a weighted directed graph model. The weights are calculated through error co-occurrence probability statistics, reflecting the strength of the induced relationship between error types.

[0119] The composite weight decision-maker receives the evaluation value of the error matching engine and the induced probability data of the correlation network model. It then fuses the direct matching weights and correlation influencing factors through the correlation fitting formula to generate composite weight coefficients for the error type.

[0120] The multi-level collaborative calibration device selects the compensation strategy based on the composite weight coefficient. It includes a temperature drift dual-channel compensation unit, a mechanical wear self-learning unit, and a vibration interference suppression unit. The temperature drift dual-channel compensation unit synchronously adjusts the hardware bridge voltage and the software conversion coefficient. The mechanical wear self-learning unit predicts the wear curve based on historical data and dynamically corrects the gain parameters. The vibration interference suppression unit switches the high-frequency sampling mode and activates the digital filter and installation stability alarm function.

[0121] Furthermore, the method according to the embodiments of this application can also be achieved by means of... Figure 7 The architecture of the electronic device shown is used to implement this. For example... Figure 7 As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, ROM 503, RAM 504, a communication port 505 connected to a network, an input / output component 506, a hard disk 507, etc. The storage device in the electronic device 500, such as ROM 503 or hard disk 507, may store a smart flow sensor testing and analysis method provided in this application. The electronic device 500 may also include a user interface 508. Of course, Figure 7 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 7 One or more components in the illustrated electronic device.

[0122] Figure 8 This is a schematic diagram of a computer-readable storage medium structure provided in one embodiment of this application. Figure 8The diagram illustrates a computer-readable storage medium 600 according to one embodiment of this application. The computer-readable storage medium 600 stores computer-readable instructions. When executed by a processor, the computer-readable instructions can perform a smart flow sensor test and analysis method according to an embodiment of this application, as described above with reference to the accompanying drawings. The storage medium 600 includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0123] In summary, the advantages of this invention are as follows: Through stepped dynamic excitation testing and multi-source data fusion analysis, the systematic nature and reliability of flow sensor error diagnosis are significantly improved. Based on multi-modal flow excitation modes and synchronous monitoring of environmental parameters, dynamic separation and precise source tracing of composite errors are achieved, effectively enhancing the identification capabilities of steady-state deviation and transient overshoot. Combined with historical data-driven error correlation network modeling, the inducing mechanisms and interaction relationships of different error types can be analyzed, significantly improving the targeting of compensation strategies. Through the collaborative optimization of hardware compensation modules and software adaptive algorithms, a multi-level error suppression mechanism is formed, achieving stable control of comprehensive errors under complex operating conditions while reducing system maintenance requirements. This technical solution overcomes the limitations of traditional calibration methods, providing a more adaptable error management path for industrial measurement and control scenarios.

[0124] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for testing and analyzing intelligent flow sensors, characterized in that, include: A stepped flow excitation signal is applied to the sensor under test using a standard flow generator, and the dynamic response data of the sensor is collected simultaneously. Extract a set of feature parameters from the response data, including steady-state deviation, transient overshoot, and noise interference level. The similarity matching calculation is performed between the set of feature parameters and the typical fault modes in the pre-stored error knowledge base to obtain the matching degree evaluation value of each error type. An error correlation network model is constructed based on historical maintenance records to quantify the strength of the induced relationship between different error types; By combining the matching degree evaluation value and the strength of the induced relationship, a composite weight coefficient is generated to determine the dominant error type and related influencing factors; A multi-level compensation strategy is formulated based on composite weight coefficients, and a coordinated calibration is implemented by adjusting hardware parameters and correcting software algorithms. Specifically, the step of constructing an error correlation network model based on historical maintenance records and quantifying the strength of the induced relationship between different error types involves: Establish a directed graph of error types. In the directed graph, nodes represent error types, the direction of the directed edges represents the induced relationship between error types, and the weight of the directed edges represents the induced probability between error types. Specifically, if error type i will induce error type j, then a directed edge from error type i to error type j is constructed; The process of combining the matching degree evaluation value and the strength of the induced relationship to generate a composite weight coefficient, and determining the dominant error type and associated influencing factors, specifically includes: Obtain all error types pointing to the error type, and form a set of cause error types for the error type; Obtain all error types pointed to by the error type, and form a set of induced error types of the error type; Based on the matching degree evaluation values ​​of all error types and the strength of the induced relationship between each error type, the composite weight coefficient of each error type is calculated through the correlation fitting formula. Specifically, the correlation fitting formula is as follows: Among them, R i Let S be the composite weight coefficient for the i-th error type, β be the direct matching weight, and S be the weight coefficient for the i-th error type. i N represents the matching score for the i-th error type. i Let j be the set of induced error types for the i-th error type, and let j be N. i The j-th element in S j For N i The matching score of the j-th element in the dataset, w ij For the i-th error type, pair N i The induced probability of the j-th element in the matrix, U j For N i The set of causal error types for the j-th element in the N-th matrix, where K is N. i The j-th element in the set of causal error types is the k-th element, w. kj For N i The j-th element in the set of causal error types is the k-th element of N. i The probability of triggering the j-th element in the equation, where γ is the correlation influence factor.

2. The intelligent flow sensor testing and analysis method according to claim 1, characterized in that, The extraction of the feature parameter set from the response data, which includes steady-state deviation, transient overshoot, and noise interference level, specifically includes: Record the full-scale response curve under constant pressure test conditions and calculate the percentage deviation of the rise time from the theoretical value. The zero-point output drift was measured in a variable temperature environment, and a linear relationship model between temperature change and zero-point offset was established. High-frequency noise components are extracted by wavelet decomposition, and the energy proportion of a specific frequency band is used as an indicator of interference level.

3. The intelligent flow sensor testing and analysis method according to claim 2, characterized in that, The step of performing similarity matching calculations between the feature parameter set and typical fault modes in the pre-stored error knowledge base to obtain matching degree evaluation values ​​for each error type specifically includes: The basic matching degree is obtained by comparing the number of common features between the current feature parameter and the standard features of each fault mode with the number of elements in the union of the current feature parameter and the standard features of each fault mode. Based on the degree to which the test environment temperature deviates from the standard operating conditions, the matching degree of temperature-sensitive errors is enhanced by a preset proportional coefficient to obtain the temperature correction value; The final matching score is obtained by adding the basic matching score to the temperature correction value.

4. The intelligent flow sensor testing and analysis method according to claim 3, characterized in that, The method for determining the weight of the directed edge is as follows: Determine the number of records in the historical maintenance log that contain error type i; Determine the number of records in the historical maintenance log that contain both error type i and error type j; The probability of error type i inducing error type j is obtained by comparing the number of records that simultaneously exhibit error type i with the number of records that exhibit error type j.

5. The intelligent flow sensor testing and analysis method according to claim 4, characterized in that, The aforementioned multi-level compensation strategy based on composite weighting coefficients, and the collaborative calibration involving hardware parameter adjustment and software algorithm correction, specifically includes: The error types whose composite weight coefficients exceed the compensation threshold are selected as compensation errors; When the compensation error is dominated by temperature drift, dual-channel compensation is activated: the hardware adjusts the bridge excitation voltage and the software corrects the flow conversion coefficient. When the compensation error is the cumulative error of mechanical wear, the self-learning compensation algorithm is activated to predict the wear trend curve based on historical calibration data. When the compensation error is caused by strong vibration interference, switch to high-frequency sampling mode and enable digital filtering, while triggering an installation stability alarm.

6. A smart flow sensor testing and analysis system, characterized in that, The method for testing and analyzing intelligent flow sensors as described in any one of claims 1-5 includes: The signal excitation module is used to apply a stepped flow excitation signal to the sensor under test through a standard flow generator. The dynamic data acquisition module, which is triggered synchronously with the signal excitation module, is used to acquire the dynamic response data of the sensor in real time and generate a time-domain waveform sequence. The feature parameter extraction module includes a steady-state analysis unit, a transient analysis unit, and a noise analysis unit. The steady-state analysis unit calculates the rise time percentage deviation through the full-scale response curve. The transient analysis unit establishes a zero-point drift-temperature linear model based on variable temperature environment testing. The noise analysis unit uses a wavelet decomposition algorithm to extract the high-frequency noise energy ratio index. The error matching engine connects to the feature parameter extraction module and has a built-in pre-stored error knowledge base and similarity calculator. The similarity calculator performs feature intersection ratio calculation and superimposes temperature deviation compensation, and outputs the matching degree evaluation value of each fault mode. The network modeling module integrates a historical maintenance database and constructs a weighted directed graph model. The weights are calculated through error co-occurrence probability statistics, reflecting the strength of the induced relationship between error types. The composite weight decision-maker receives the evaluation value of the error matching engine and the induced probability data of the correlation network model. It then fuses the direct matching weights and correlation influencing factors through the correlation fitting formula to generate composite weight coefficients for the error type. The multi-level collaborative calibration device, based on the composite weighting coefficient triggering compensation strategy selection logic, includes a temperature drift dual-channel compensation unit, a mechanical wear self-learning unit, and a vibration interference suppression unit. The temperature drift dual-channel compensation unit synchronously adjusts the hardware bridge voltage and the software conversion coefficient. The mechanical wear self-learning unit predicts the wear curve based on historical data and dynamically corrects the gain parameters. The vibration interference suppression unit switches the high-frequency sampling mode and activates the digital filter and installation stability alarm function.

7. An electronic device, characterized in that, include: At least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed, enables the at least one processor to perform the intelligent flow sensor test and analysis method as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent flow sensor test and analysis method according to any one of claims 1-5.

Images