A method and system for simultaneous rejection of differential mode and common mode shifts in an array of acetylene sensors

Abstract

The present application relates to the technical field of gas sensing detection, and particularly relates to a differential mode common mode drift cooperative suppression method and system of an acetylene sensor array, which solves the technical problem of poor reliability of monitoring data in the prior art. The method comprises: obtaining detection data corresponding to multiple historical detections and a current detection of the acetylene sensor array; the detection data comprises a response signal sequence of each sensing unit in the acetylene sensor array and an environmental parameter sequence; performing drift analysis based on the detection data corresponding to the multiple historical detections and the current detection of the acetylene sensor array, to determine a differential mode drift weight of each sensing unit in the acetylene sensor array and a common mode drift amount of the acetylene sensor array; and performing cooperative filtering processing on the response signal sequence of each sensing unit in the detection data of the current detection of the acetylene sensor array according to the differential mode drift weight and the common mode drift amount, to obtain a target response signal sequence of each sensing unit after drift suppression.

Description

Technical Field

[0001] This invention relates to the field of gas sensing and detection technology, specifically to a method and system for collaborative suppression of differential-mode and common-mode drift in an acetylene sensor array. Background Technology

[0002] In fields such as industrial safety monitoring, power equipment fault early warning (e.g., dissolved gas analysis in transformer oil), and chemical process control, acetylene is a key characteristic gas, and its concentration changes are directly related to significant safety hazards. To improve detection reliability and anti-interference capabilities, acetylene sensor array technology is widely used, enhancing system robustness through multi-channel redundant sensing. The sensor array consists of multiple identical or similar sensing units, aiming to improve detection accuracy and stability through multi-unit collaborative operation.

[0003] However, due to limitations in sensitive material processing, packaging consistency, and long-term operational stability, sensor arrays face severe signal drift challenges in practical applications. Especially in complex and variable operating environments, environmental disturbances and device aging work together to cause continuous distortion of the output signal, severely restricting the practical development of high-precision, long-cycle online monitoring systems. Existing sensor drift suppression methods often address single types of drift, making it difficult to handle the complex situation of multiple drifts coexisting and coupling. This results in insufficient long-term stability of the sensor array, affecting the reliability of monitoring data and consequently reducing the accuracy of safety warnings. Summary of the Invention

[0004] To address the technical problem of poor reliability of monitoring data in existing technologies, the present invention aims to provide a method and system for collaborative suppression of differential-mode and common-mode drift in an acetylene sensor array. The specific technical solution adopted is as follows:

[0005] This application provides a method for coordinated suppression of differential and common-mode drift in an acetylene sensor array, including:

[0006] The detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array are obtained; the detection data includes the response signal sequence of each sensing unit in the acetylene sensor array and the environmental parameter sequence.

[0007] Drift analysis is performed based on the detection data corresponding to the multiple historical detections and the current detection to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array. The differential mode drift weight is used to characterize the degree of signal drift caused by individual differences of the corresponding sensing unit. The common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array caused by environmental factors.

[0008] Based on the differential mode drift weight and the common mode drift amount, the response signal sequences of each sensing unit in the current detection data of the acetylene sensor array are subjected to collaborative filtering to obtain the target response signal sequences of each sensing unit after drift suppression.

[0009] In one possible implementation, the method includes:

[0010] Based on the environmental parameter sequences in the detection data from multiple historical tests, the detection data from multiple historical tests are clustered and grouped to obtain multiple clusters;

[0011] Differential analysis was performed on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array;

[0012] Based on the environmental parameter sequence from multiple historical and current tests, a benchmark test is determined from the multiple historical and current tests.

[0013] The common-mode drift of the acetylene sensor array is determined by correlation analysis based on benchmark detection data, multiple historical detection data, and current detection data.

[0014] In one possible implementation, the method includes:

[0015] For each cluster and each sensing unit, the response characteristic difference value between the sensing unit in the current detection and the cluster is determined based on the response signal sequence of the sensing unit in each historical detection and the current detection in the cluster.

[0016] Based on the difference in response characteristics between the current detection and each cluster, the feature difference coefficient between the current detection and multiple historical detections of the sensing unit is determined.

[0017] For each sensing unit, the differential mode drift weight of the sensing unit is determined based on the characteristic difference coefficient between the current detection and multiple historical detections of each sensing unit in the acetylene sensor array.

[0018] In one possible implementation, the method includes:

[0019] For each cluster and each sensing unit, the characteristic coefficients between the current detection and each historical detection in the cluster are determined based on the response signal sequence of the sensing unit in each historical detection in the cluster. The characteristic coefficients include the response rise rate characteristic coefficient, the recovery rate characteristic coefficient, and the peak amplitude characteristic coefficient.

[0020] For each cluster, the feature matrix corresponding to the cluster is determined based on the feature coefficients between the current detection and each historical detection in the cluster by each sensing unit.

[0021] Local anomaly detection is performed on the feature matrix corresponding to each cluster to determine the differential mode drift weight of each sensing unit in the acetylene sensor array.

[0022] In one possible implementation, the method includes:

[0023] The environmental parameter sequences from multiple historical tests and the current test were compared with standard environmental parameters to determine the differences between the environmental parameter sequences from multiple historical tests and the current test and the standard environmental parameters.

[0024] Based on the differences between environmental parameter sequences from multiple historical and current tests and standard environmental parameters, a benchmark test is determined from multiple historical and current tests.

[0025] In one possible implementation, the method includes:

[0026] For each sensing unit, based on the response signal sequence and environmental parameter sequence of the sensing unit in the benchmark detection, multiple historical detections and the current detection, the environmental parameter difference sequence and response signal difference sequence of the sensing unit in each historical detection and the current detection are determined.

[0027] Based on the weighted multivariate function fitting of the environmental parameter difference sequence and response signal difference sequence of the sensing unit in each historical detection and the current detection, the common mode drift fitting function corresponding to the sensing unit is determined; the common mode drift fitting function is used to characterize the influence relationship of environmental parameter changes on the response signal;

[0028] Based on the differential sequence of environmental parameters detected by each sensing unit and the corresponding common-mode drift fitting function, the common-mode drift of the acetylene sensor array is determined.

[0029] In one possible implementation, the method includes:

[0030] For each historical test and each test in the current test, as well as for each sensing unit, determine the changes in environmental parameters and response signals of the sensing unit between the test and the benchmark test.

[0031] The common-mode drift fitting function corresponding to the sensing unit is determined by performing weighted multivariate function fitting on the changes in environmental parameters and response signals between the sensing unit and the benchmark detection in historical and current detections.

[0032] In one possible implementation, the method includes:

[0033] For each sensing unit, the change in environmental parameters between the current detection and the benchmark detection is input into the common-mode drift fitting function corresponding to the sensing unit to obtain the unit drift amount corresponding to the sensing unit.

[0034] The common-mode drift of the acetylene sensor array is obtained by weighted fusion of the unit drift values ​​corresponding to each sensing unit.

[0035] In one possible implementation, the method includes:

[0036] Based on the response signal sequence and common-mode drift of each sensing unit in the current detection data of the acetylene sensor array, the preliminary correction signal sequence of each sensing unit is determined.

[0037] For each sensing unit, based on the differential mode drift weight of the sensing unit and the preliminary correction signal sequence, the Kalman filter algorithm is used to filter the signal to obtain the target response signal sequence of the sensing unit after drift suppression.

[0038] This application provides a differential-mode and common-mode drift cooperative suppression system for an acetylene sensor array, comprising:

[0039] The data acquisition unit is used to acquire the detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array; the detection data includes the response signal sequence of each sensing unit in the acetylene sensor array and the environmental parameter sequence.

[0040] The drift analysis unit is used to perform drift analysis based on the detection data corresponding to the multiple historical detections and the current detection, to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array; the differential mode drift weight is used to characterize the degree of signal drift of the corresponding sensing unit due to individual differences; the common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array due to environmental factors.

[0041] The collaborative filtering unit is used to perform collaborative filtering on the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array according to the differential mode drift weight and the common mode drift amount, so as to obtain the target response signal sequence of each sensing unit after drift suppression.

[0042] The present invention has the following beneficial effects:

[0043] In view of the technical problem of poor data reliability in existing monitoring technologies, this application provides a method and system for collaborative suppression of differential-mode and common-mode drift in acetylene sensor arrays. This application acquires historical and current detection data from the acetylene sensor array, performs drift analysis to determine the differential-mode drift weight and common-mode drift amount, and then performs collaborative filtering on the current detection data based on the differential-mode drift weight and common-mode drift amount, effectively suppressing differential-mode and common-mode drift in the acetylene sensor array. The above technical solution fully utilizes drift factors in historical data, and through quantitative analysis of the different characteristics of differential-mode and common-mode drift, achieves targeted collaborative suppression, significantly improving the long-term stability and reliability of the acetylene sensor array, and providing more accurate and reliable detection data support for applications such as industrial safety monitoring and power equipment fault early warning. Attached Figure Description

[0044] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a system architecture diagram of a differential-mode and common-mode drift collaborative suppression system for an acetylene sensor array provided in one embodiment of the present invention;

[0046] Figure 2 This is a flowchart illustrating a method for collaborative suppression of differential and common-mode drift in an acetylene sensor array, provided in one embodiment of the present invention. Detailed Implementation

[0047] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a differential-mode and common-mode drift collaborative suppression method and system for an acetylene sensor array proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0049] In all division and logarithmic operations covered in this application, a smoothing mechanism is employed to prevent computer program crashes or invalid values ​​from being generated due to a zero denominator or a zero input. Specifically, a positive correction factor is superimposed on the denominator term of the division operation or the argument term of the logarithmic function. For example, the value is This ensures the robustness and feasibility of the algorithm under extreme conditions.

[0050] The normalization function mentioned in this application Unless otherwise specified, all values ​​are normalized using maximum and minimum values. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculated result exceeds the [0,1] interval, it is restricted to the [0,1] range by a truncation function (i.e., if the result is less than 0, it is taken as 0, and if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0051] The following description, in conjunction with the accompanying drawings, details the specific scheme of the differential-mode and common-mode drift collaborative suppression method and system for an acetylene sensor array provided by the present invention.

[0052] Please see Figure 1 This diagram illustrates a system architecture of a differential-mode and common-mode drift collaborative suppression system for an acetylene sensor array, according to an embodiment of the present invention. The acetylene sensor array differential-mode and common-mode drift collaborative suppression system 10 includes: a data acquisition unit 11, a drift analysis unit 12, and a collaborative filtering unit 13. The units communicate bidirectionally via a communication link, ensuring real-time interaction of acquired data and analysis results. The communication link can employ wired or wireless transmission methods to meet the communication needs of different monitoring scenarios.

[0053] The data acquisition unit 11 is used to acquire the detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array.

[0054] The detection data includes the response signal sequences of each sensing unit in the acetylene sensor array and the environmental parameter sequences.

[0055] For example, the data acquisition unit 11 may include a gas sensor module, an environmental parameter sensor module, and a data acquisition module. The gas sensor module contains multiple sensing units, each used to detect the concentration of acetylene gas in the environment and output a corresponding response signal. The environmental parameter sensor module includes temperature sensors, humidity sensors, etc., used to monitor changes in parameters of the working environment of the acetylene sensor array. The data acquisition module converts the analog signals output by each sensor into digital signals using an analog-to-digital converter, and acquires and stores the data according to a set sampling frequency. For example, the sampling frequency of the acetylene sensing unit can be set to 100Hz, and the sampling frequency of the environmental parameters can be set to 0.1Hz, collecting data for one minute each time to ensure data integrity and timeliness.

[0056] In some embodiments, this application can normalize the collected data using maximum and minimum value normalization. It should be noted that the maximum and minimum values ​​used for each type of data during normalization are based on the preset theoretical extreme values ​​of the sensor range. In this way, this application can construct a corresponding data sequence from the normalized response signal data, temperature data, and humidity data obtained from each detection according to the data acquisition order.

[0057] The drift analysis unit 12 is used to perform drift analysis based on the detection data corresponding to multiple historical detections and the current detection, to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array.

[0058] Among them, the differential mode drift weight is used to characterize the degree of signal drift caused by individual differences of the corresponding sensing unit, and the common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array caused by environmental factors.

[0059] The drift analysis unit 12 incorporates a data processing chip, enabling it to perform complex data analysis and model calculation tasks. In practical applications, differential mode drift primarily stems from manufacturing differences, localized contamination, or inconsistent aging rates among sensor units, manifesting as inconsistent responses between channels. Common mode drift, on the other hand, is mainly caused by environmental factors such as temperature and humidity fluctuations and background gas interference, resulting in synchronous shifts across all units. The drift analysis unit 12 can quantify the impact of these two types of drift by comprehensively analyzing the characteristic differences between historical and current data, providing fundamental parameters for subsequent collaborative suppression.

[0060] The collaborative filtering unit 13 is used to perform collaborative filtering on the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array according to the differential mode drift weight and the common mode drift amount, so as to obtain the target response signal sequence of each sensing unit after drift suppression.

[0061] The collaborative filtering unit 13 can be implemented using a digital signal processor, with built-in Kalman filtering and other algorithm modules, and dynamically adjusts the filtering strategy according to the parameters output by the drift analysis unit 12.

[0062] In some embodiments, the collaborative filtering unit 13 can first perform preliminary correction by the common-mode drift amount to eliminate the overall offset caused by environmental factors, and then perform differentiated filtering on the pre-corrected signal according to the differential-mode drift weight of each sensing unit, focusing on suppressing signal fluctuations caused by individual differences, thereby achieving collaborative suppression of differential-mode and common-mode drift and improving the stability and reliability of the output signal.

[0063] It should be noted that the various embodiments of this application can be referenced or learned from each other. For example, the same or similar steps, method embodiments, system embodiments and device embodiments can be referenced from each other without limitation.

[0064] Please see Figure 2 The diagram illustrates a flowchart of a method for collaborative suppression of differential and common-mode drift in an acetylene sensor array according to an embodiment of the present invention. The method includes the following steps:

[0065] Step 201: Obtain the detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array.

[0066] The detection data includes response signal sequences from each sensing unit in the acetylene sensor array and environmental parameter sequences. Multiple historical detection data refers to the detection data collected by the sensor array during past normal operating cycles, used to provide a reference benchmark for drift analysis. Current detection data refers to the real-time detection data for which drift suppression is currently required. The response signal sequence is the sequence of signals output by the corresponding sensing unit in chronological order during a single detection process, reflecting the response characteristics of the sensing unit to acetylene gas. The environmental parameter sequences mainly include temperature and humidity sequences, reflecting changes in environmental conditions during the detection process.

[0067] The data acquisition process must ensure consistency. For example, each test collects 1 minute of data. The sampling frequency of the acetylene sensor array is set to 100Hz, and a single sensing unit can collect 6000 response signal data per test. The sampling frequency of the temperature sensor and humidity sensor is set to 0.1Hz, and each test can collect 6 environmental parameter data.

[0068] Meanwhile, to eliminate the impact of differences in the measurement ranges of different sensors, this application can normalize the collected raw data. For example, the normalization process can adopt the maximum-minimum normalization method, that is, using the preset theoretical extreme value of the sensor range as the reference benchmark for normalization, and mapping the raw data to the [0,1] interval.

[0069] Step 202: Perform drift analysis based on the detection data corresponding to multiple historical detections and the current detection to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array.

[0070] Among them, the differential mode drift weight is used to characterize the degree of signal drift caused by individual differences of the corresponding sensing unit, and the common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array caused by environmental factors.

[0071] Differential-mode drift weight reflects the individual performance differences of each sensing unit caused by manufacturing variations, local contamination, or inconsistent aging rates. These factors can lead to inconsistent response signals from different sensing units under the same environmental conditions. Common-mode drift, on the other hand, characterizes the synchronous impact of environmental factors such as temperature and humidity changes and background gas interference on the entire sensor array. These factors can cause synchronous shifts in the response signals of all sensing units. In practical applications, these two types of drift often coexist and are coupled with each other; suppressing only one type of drift is unlikely to achieve the desired stability improvement.

[0072] This application can quantify the impact of these two types of drift by collaboratively analyzing their characteristics, thus providing a basis for subsequent collaborative suppression. For example, drift analysis can be implemented using machine learning algorithms, such as cluster analysis, anomaly detection, and function fitting. By comparing the differences in response signals between current and historical detections, and combining environmental condition classification, differential drift caused by individual differences and common-mode drift caused by environmental factors can be separated.

[0073] Step 203: Based on the differential mode drift weight and the common mode drift amount, perform collaborative filtering on the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array to obtain the target response signal sequence of each sensing unit after drift suppression.

[0074] In one possible implementation, this application can determine the preliminary correction signal sequence of each sensing unit based on the response signal sequence and common-mode drift of each sensing unit in the current detection data of the acetylene sensor array. Then, for each sensing unit, based on the differential-mode drift weight of the sensing unit and the preliminary correction signal sequence, the target response signal sequence of the sensing unit after drift suppression is obtained by filtering with a Kalman filter algorithm.

[0075] The preliminary calibration signal sequence is an intermediate result after eliminating the influence of common-mode drift, providing a basis for subsequent differential-mode drift suppression. This application subtracts the common-mode drift from each element of the response signal sequence of each sensing unit in the currently detected data to obtain the preliminary calibration signal sequence of each sensing unit. This global calibration method effectively eliminates the overall signal offset caused by environmental factors, allowing the output signal of each sensing unit to return to a relatively stable reference level.

[0076] Kalman filtering is a recursive estimation algorithm that obtains the optimal state estimate by fusing system model predictions and actual observations. The Kalman filter algorithm is used to suppress random fluctuations caused by differential mode drift. By adjusting the observation noise parameters, the filter strength is matched to the degree of differential mode drift, achieving precise suppression.

[0077] In some embodiments, the Kalman filter algorithm can dynamically adjust the observation noise parameters according to the differential mode drift weight to achieve differentiated processing for units with different drift degrees. In this application, the response signal of the sensing unit can be regarded as the system state, and the degree of confidence of the filter in the observation value can be controlled by adjusting the observation noise covariance matrix. For units with large differential mode drift weights (i.e., units with severe drift), the observation noise should be increased to reduce the confidence in the current observation value and rely more on the system model for smoothing. For units with small differential mode drift weights (i.e., units with light drift), the observation noise should be reduced to maintain high signal fidelity.

[0078] For example, this application can correct the observation noise in the Kalman filter algorithm based on the differential mode drift weight, and the corrected observation noise satisfies the following formula:

[0079]

[0080] in, The first in the acetylene sensor array Corrected observation noise for each sensing unit The first acetylene sensor in the array obtained according to the original Kalman filter algorithm. The observation noise of each sensing unit The differential mode sensitivity coefficient can be determined experimentally, with a range of 3-10. In this embodiment, it is set to 5. The larger the differential mode sensitivity coefficient, the more sensitive it is to fluctuations in the response signal caused by differential mode. The first in the acetylene sensor array Differential mode drift weights of each sensing unit.

[0081] Differential mode drift weight The larger the value, the better the correction for observation noise. The larger the value, the less confidence the Kalman filter has in the current measurement, and the stronger the filtering smoothing effect, thus effectively suppressing signal fluctuations caused by differential mode drift; conversely, the smaller the differential mode drift weight, the less confidence the Kalman filter has in the current measurement, and the stronger the filtering smoothing effect, thus effectively suppressing signal fluctuations caused by differential mode drift. The smaller the value, the better the correction for observation noise. The closer The weaker the filtering and smoothing effect, the more dynamic details of the signal are preserved.

[0082] Thus, this application can input the preliminary correction signal sequence into the Kalman filter algorithm after correcting the observation noise, and after steps such as prediction and updating, output the target response signal sequence, which is the accurate detection signal after eliminating differential mode drift and common mode drift.

[0083] Based on the above technical solution, this application acquires historical and current detection data from an acetylene sensor array, performs drift analysis to determine the differential-mode drift weight and common-mode drift amount, and then performs collaborative filtering on the current detection data according to the differential-mode drift weight and common-mode drift amount, effectively suppressing differential-mode drift and common-mode drift in the acetylene sensor array. This technical solution fully utilizes drift factors in historical data, and through quantitative analysis of the different characteristics of differential-mode drift and common-mode drift, achieves targeted collaborative suppression, significantly improving the long-term stability and reliability of the acetylene sensor array, and providing more accurate and reliable detection data support for applications such as industrial safety monitoring and power equipment fault early warning.

[0084] As a possible embodiment of this application, step 202 above can be implemented through the following steps:

[0085] Step 301: Based on the environmental parameter sequence in the detection data from multiple historical detections, cluster the detection data from multiple historical detections to obtain multiple clusters.

[0086] This application can group historical detection data with similar environmental conditions into one category, thereby analyzing individual differences of sensing units under similar environmental backgrounds and avoiding interference from environmental factors on differential mode drift analysis.

[0087] In some embodiments, this application may employ the K-means clustering algorithm, using the environmental parameter sequences (such as temperature sequences and humidity sequences) from each historical detection as feature vectors to calculate the distance between samples. For example, the distance between two samples can be defined as the sum of the Euclidean distance of the temperature sequence and the Euclidean distance of the humidity sequence. The clustering results divide the historical detection data into multiple clusters, where the detection data from each historical detection within each cluster share similar environmental backgrounds, providing a foundation for subsequent differential drift analysis.

[0088] Step 302: Perform difference analysis on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array.

[0089] In some embodiments, this application may quantify the differential drift weight based on the difference in response signals between the current detection and historical detections within each cluster.

[0090] For example, this application can calculate the difference in response signal characteristics between the current detection and the historical detection within the cluster for each cluster and each sensing unit, such as response rise rate, recovery rate, peak amplitude and baseline value.

[0091] The aforementioned characteristic parameters can effectively reflect the performance status of the sensing unit. For example, as the sensitive material ages, the amplitude of the response signal decreases. As the performance of the microheater deteriorates, the response rise and recovery rates slow down, and baseline drift occurs. By analyzing the differences between current and historical detections in these characteristics, the degree of performance change of each sensing unit can be quantified, thereby determining the differential mode drift weight.

[0092] It should be noted that the differential mode drift weight should reflect the degree of signal drift caused by individual differences in the sensing unit. The larger the weight value, the more severely the unit is affected by differential mode drift, and the stronger the suppression needs to be given in subsequent filtering.

[0093] Step 303: Determine the benchmark detection from the environmental parameter sequence of multiple historical detections and the current detection.

[0094] Among them, the benchmark test is the reference benchmark for common mode drift analysis. It is necessary to select test data that are stable and representative under environmental conditions to ensure the reliability of the correlation analysis between changes in environmental parameters and changes in response signals.

[0095] In one possible implementation, this application can compare the environmental parameter sequences of multiple historical tests and the current test with standard environmental parameters to determine the environmental parameter difference values ​​between the environmental parameter sequences of multiple historical tests and the current test and the standard environmental parameters. Then, based on the environmental parameter difference values ​​between the environmental parameter sequences of multiple historical tests and the current test and the standard environmental parameters, a benchmark test is determined from the multiple historical tests and the current test.

[0096] Among them, the environmental parameter difference value is used to quantify the degree of deviation between the environmental conditions of each test and the standard environmental conditions. The smaller the deviation, the more stable the environmental conditions of the test are, and the more suitable it is as a benchmark test.

[0097] In some embodiments, standard environmental parameters can be set according to the calibration environment of the sensor array, typically set to a temperature of 20°C and a humidity of 65%. These environmental conditions are typical and representative of most industrial testing scenarios. The comparison between the environmental parameter sequence and the standard environmental parameters is performed using Euclidean distance calculation. Since the environmental parameter sequence includes both temperature and humidity sequences, the Euclidean distances of the two types of sequences need to be calculated separately and then summed to obtain the environmental parameter difference value.

[0098] It should be noted that the standard environmental parameters can be adjusted according to the actual application scenario. For example, in a high-temperature industrial environment, the standard environmental parameters can be set to a temperature of 40°C and a humidity of 50% to ensure that the environmental conditions of the benchmark test match the actual application scenario and improve the pertinence of common mode drift analysis.

[0099] This application can use the detection with the smallest difference in environmental parameters as the benchmark detection, that is, the environmental conditions are closest to the standard environmental parameters, to ensure that the correlation between environmental changes and response signal changes is more accurate in subsequent common-mode drift analysis.

[0100] In some embodiments, if there are multiple environmental parameter differences that are the same and the smallest, the latest detection time can be selected as the benchmark detection, because the newer detection data can better reflect the current basic performance of the sensor array and reduce the impact of long-term aging on the accuracy of the benchmark.

[0101] Step 304: Based on the correlation analysis of the benchmark detection, multiple historical detections, and current detection data, determine the common mode drift of the acetylene sensor array.

[0102] This application can establish a mapping relationship between changes in environmental parameters and changes in response signals. This mapping relationship can be used to quantify the common-mode drift caused by environmental factors. Since the impact of common-mode drift on all sensing units is consistent, the correlation analysis results of each sensing unit can be integrated to improve the accuracy of the common-mode drift.

[0103] In some embodiments, this application can determine the common-mode drift based on the correlation between the benchmark detection and other detections in terms of environmental parameters and response signals. For each sensing unit, the environmental parameter difference sequence (such as temperature change, humidity change) and response signal difference sequence between each detection and the benchmark detection can be calculated, and then a mapping relationship between environmental parameter changes and response signal changes can be established through function fitting. Since the impact of common-mode drift on all sensing units is similar, the fitting results of each unit can be weighted and fused to obtain the common-mode drift amount characterizing the degree of common-mode drift of the entire sensor array.

[0104] It should be noted that the influence of differential mode drift should be considered during the fitting process. Data points that are significantly affected by differential mode drift should be assigned lower weights to improve the accuracy of common mode drift calculation. This application, based on historical data correlation analysis, can effectively capture the overall influence of environmental factors on the sensor array, providing a basis for the accurate quantification of common mode drift.

[0105] Based on the above technical solution, this application effectively isolates the interference of environmental factors on differential mode drift analysis by clustering historical detection data according to environmental parameters. Then, by analyzing the difference in response signals between the current detection and historical detections in each cluster, the differential mode drift weight of each sensing unit is accurately quantified. By selecting a benchmark detection and performing correlation analysis, the common mode drift of the sensor array is precisely calculated. In this way, this application can fully consider the different characteristics of differential mode drift and common mode drift and their mutual influence, providing high-quality parameter support for subsequent collaborative suppression and significantly improving the effectiveness and stability of drift suppression.

[0106] This application can quantify the differential mode drift weight in a variety of ways. The following describes the various quantization methods provided in this application.

[0107] As a possible embodiment of this application, step 302 above can be implemented through the following steps:

[0108] Step 401: For each cluster and each sensing unit, determine the response feature difference value between the sensing unit in the current detection and the cluster based on the response signal sequence of the sensing unit in each historical detection and the current detection in the cluster.

[0109] Among them, the response characteristic difference value reflects the degree of performance change of each sensing unit under specific environmental conditions between current detection and historical detection, and is an important basis for quantifying differential mode drift.

[0110] For example, the response feature difference value satisfies the following formula:

[0111]

[0112] in, For the first The sensing unit is currently detecting and the first... The response feature difference values ​​between the clusters For the first The sensing unit in the first The coefficient of variation (COP) is the coefficient of variation of the baseline values ​​in the response signal sequence of each historical detection within a cluster. The baseline values ​​can be represented by the minimum mean obtained by moving average of the response signal sequence, reflecting the signal baseline level of the sensing unit. The COP is calculated by dividing the standard deviation by the mean. For the first The number of historical tests in each cluster.

[0113] For the first The sensing unit is the first in this cluster. The absolute value of the difference between the response rise rate of the previous historical detection and the response rise rate of the current detection. The response rise rate can be represented by the ratio of the maximum value in the response signal sequence to the detection time corresponding to the maximum value, reflecting the sensitivity of the sensing unit to acetylene gas.

[0114] For the first The sensing unit is the first in this cluster. The absolute value of the difference between the recovery speed of the previous historical detection and the recovery speed of the current detection. The recovery speed can be represented by the ratio of the maximum value in the response signal sequence to the recovery time. The recovery time is the difference between the detection time corresponding to the maximum value and the detection time corresponding to the first minimum value after the maximum value, reflecting the signal recovery capability of the sensing unit.

[0115] For the first The sensing unit is the first in this cluster. The absolute value of the difference between the peak amplitude of the previous historical detection and the peak amplitude of the current detection. The peak amplitude can be represented by the average of all peaks in the response signal sequence, reflecting the amplitude level of the sensor unit's response signal.

[0116] This application can quantify the response characteristics through the above-mentioned dimensions such as response rise rate, recovery rate, peak amplitude and baseline value, and then evaluate the difference in response characteristics between each sensing unit and each cluster at the current detection.

[0117] It should be noted that, in order to eliminate the influence of dimensional differences, each characteristic parameter should be normalized before calculating the differences, such as using the maximum value normalization method to map it to the [0,1] interval.

[0118] Step 402: Determine the feature difference coefficient between the current detection and multiple historical detections of the sensing unit based on the response feature difference value between the current detection and each cluster.

[0119] Among them, the feature difference coefficient is a comprehensive quantification of the response feature difference value of a single sensing unit in all clusters, reflecting the overall differential mode drift of the sensing unit compared to all historical detections.

[0120] For example, the feature difference coefficient satisfies the following formula:

[0121]

[0122] in, For the first The feature difference coefficient of each sensing unit between the current detection and multiple historical detections. The number of clusters, For the first The sensing unit is currently detecting and the first... The response feature difference values ​​between the clusters For all sensing units in the current detection and the first The coefficient of variation (COP) of the response characteristic differences between clusters measures the consistency of differences among sensing units within a cluster. The COP is calculated by dividing the standard deviation by the mean. If the dispersion of the difference values ​​among sensing units within a cluster is small, it indicates that the cluster has high reference value. A larger value indicates a greater weight in the contribution of the characteristic difference coefficient.

[0123] Step 403: For each sensing unit, determine the differential mode drift weight of the sensing unit based on the characteristic difference coefficient between the current detection and multiple historical detections of each sensing unit in the acetylene sensor array.

[0124] Among them, the differential mode drift weight is used to measure the proportion of the differential mode drift of a single sensing unit relative to the entire sensor array. The larger the weight, the more severe the individual drift of the sensing unit is, and the greater the interference to the overall detection results.

[0125] For example, the differential mode drift weights satisfy the following formula:

[0126]

[0127] in, For the first Differential mode drift weights of individual sensing units For the first The feature difference coefficient of each sensing unit between the current detection and multiple historical detections. This is the sum of the feature difference coefficients of all sensing units between the current detection and multiple historical detections. This is a safety parameter used to correct for denominators of 0; its specific value can be determined based on... The value is determined by the range of possible values, such as taking the empirical value of 1.

[0128] The larger the proportion of the characteristic difference coefficient of a single sensing unit to the total difference coefficient, the greater the impact of its differential mode drift on the overall signal. Therefore, the differential mode drift weight is positively correlated with the characteristic difference coefficient.

[0129] Based on the above technical solution, this application extracts key feature parameters of the response signal, constructs response feature difference values ​​to quantify the differences within a single cluster, and obtains feature difference coefficients by fusing the difference values ​​of all clusters. This comprehensively reflects the overall drift degree of the sensing unit. Finally, based on the feature difference coefficients of each sensing unit between the current detection and multiple historical detections, the differential mode drift weight is obtained, achieving accurate quantification of differential mode drift. This solution effectively separates the influence of environmental factors from individual factors, making the calculation of differential mode drift weights more targeted and reliable, and providing a precise basis for subsequent filter parameter adjustments.

[0130] As another possible embodiment of this application, step 302 above can be implemented by the following steps:

[0131] Step 501: For each cluster and each sensing unit, determine the characteristic coefficients between the current detection and each historical detection in the cluster based on the response signal sequence of the sensing unit in each historical detection in the cluster.

[0132] Among them, the characteristic coefficients include the response rise rate characteristic coefficient, the recovery rate characteristic coefficient, and the peak amplitude characteristic coefficient.

[0133] In some embodiments, the response rise rate characteristic coefficient is a normalized result of the response rise rate, the recovery rate characteristic coefficient is a normalized result of the recovery rate, and the peak amplitude characteristic coefficient is a normalized result of the peak amplitude. The normalization method adopts maximum-minimum normalization, and the extreme values ​​are determined based on the theoretical performance parameters of the sensor. The relevant calculations of the response rise rate, recovery rate, and peak amplitude can be referred to in the above embodiments, and will not be repeated here.

[0134] The characteristic coefficient is a quantification of the key features of the response signal. Unlike the above embodiments, the characteristic coefficient in this embodiment directly reflects the response characteristic level of the sensing unit, rather than the difference from historical detection. Through subsequent anomaly detection, characteristic coefficients that deviate significantly from most samples can be identified, thereby determining the degree of differential mode drift.

[0135] Step 502: For each cluster, determine the feature matrix corresponding to the cluster based on the feature coefficients between the current detection and each historical detection in the cluster by each sensing unit.

[0136] The feature matrix integrates the feature coefficients of all sensing units and all historical detections within a single cluster, providing a data foundation for subsequent anomaly detection. Each row of the feature matrix represents a sensing unit, and each column corresponds to a type of feature coefficient (response rise rate feature coefficient, recovery rate feature coefficient, peak amplitude feature coefficient). The elements in the matrix are the feature coefficient values ​​of the corresponding sensing unit in the corresponding sample.

[0137] For example, the first Each cluster contains The sensor array contains the second historical detection. If there are 10 sensing units, then the feature matrix of the cluster is: OK Column, among which The columns correspond to the three feature coefficients for each historical detection.

[0138] Step 503: Perform local anomaly detection on the feature matrix corresponding to each cluster to determine the differential mode drift weight of each sensing unit in the acetylene sensor array.

[0139] Local Outlier Factor (LOF) detection is used to identify outliers in the feature matrix, i.e., samples whose feature coefficients deviate significantly from those of most sensing units. These outliers correspond to sensing units or detection data with severe differential mode drift. The LOF value quantifies the degree of anomaly of a single sensing unit, thereby determining the differential mode drift weight. For example, the number of nearest neighbors for LOF detection can be set to 5. The LOF value is obtained by calculating the local reachability density of each data point relative to its neighbors; the larger the LOF value, the more severe the anomaly.

[0140] For example, the differential mode drift weights satisfy the following formula:

[0141]

[0142] in, For the first Differential mode drift weights of individual sensing units For the first The anomaly metric of the first sensing unit can be determined based on the LOF value and characteristic coefficients. For example, for the first... For each sensing unit, this application can calculate the mean of the absolute values ​​of the differences between the current LOF value of the sensing unit and the LOF values ​​of all historical detections within each cluster, and simultaneously calculate the mean of the absolute values ​​of the differences between the sensing unit and the characteristic coefficients of all other sensing units in the current detection. Multiplying these two parameters yields the anomaly metric for the sensing unit. . This is the sum of the abnormal metrics of all sensing units. The larger the proportion of the abnormal metrics of a single sensing unit to the total abnormal metrics, the more severe its differential mode drift, and the higher the differential mode drift weight.

[0143] Based on the above technical solution, this application constructs a feature matrix by extracting feature coefficients and uses local anomaly factor detection to identify abnormal features caused by differential mode drift, thereby quickly identifying differential mode drift that deviates significantly from the normal level. It is suitable for application scenarios that are sensitive to abnormal drift.

[0144] As a possible embodiment of this application, step 304 above can be implemented through the following steps:

[0145] Step 601: For each sensing unit, based on the response signal sequence and environmental parameter sequence of the sensing unit in the benchmark detection, multiple historical detections and the current detection, determine the environmental parameter difference sequence and response signal difference sequence of the sensing unit in each historical detection and the current detection.

[0146] Among them, the environmental parameter difference sequence and the response signal difference sequence reflect the environmental and signal changes relative to the benchmark detection, and are the basic data for establishing the mapping relationship between the environment and the response signal.

[0147] For example, the environmental parameter difference sequence may include a temperature difference sequence and a humidity difference sequence. Each element of the temperature difference sequence is the difference between the temperature data of a certain detection and the temperature data of the benchmark detection at the corresponding time. The humidity difference sequence is similar. Each element of the response signal difference sequence is the difference between the response signal data of a certain detection and the response signal data of the benchmark detection.

[0148] Step 602: Based on the differential sequences of environmental parameters and response signals of the sensing unit in each historical detection and the current detection, perform weighted multivariate function fitting to determine the common-mode drift fitting function corresponding to the sensing unit.

[0149] Among them, the common-mode drift fitting function is used to characterize the influence of changes in environmental parameters on the response signal.

[0150] In one possible implementation, this application can determine the changes in environmental parameters and response signals of the sensing unit between the detection and the benchmark detection for each detection and each sensing unit in multiple historical detections and the current detection. Then, a weighted multivariate function is fitted based on the changes in environmental parameters and response signals of the sensing unit between the historical detection and the benchmark detection in the current detection and the benchmark detection, respectively, to determine the common-mode drift fitting function corresponding to the sensing unit.

[0151] For example, this application can use the mean of the elements of the response signal difference sequence as the dependent variable and the mean of the elements of the temperature difference sequence and the humidity difference sequence as the independent variables, as data points for function fitting. For each sensing unit, the data points of the sensing unit in all historical detections and current detections are obtained in the above manner to form a fitting dataset.

[0152] During the fitting process, the fitting weight for each data point can be determined based on the differential mode drift weight of the corresponding sensing unit. For example, the fitting weight satisfies the following formula:

[0153]

[0154] in, For the first The first sensing unit Fitting weights for each data point For the first The sensing unit in the first Differential drift weights for each data point during detection. For the first The differential mode drift weights are the sum of the differential mode drift weights corresponding to all data points of each sensing unit during detection. It should be noted that the differential mode drift weights in the above embodiments of this application correspond to the differential mode drift weights of each sensing unit during the current detection. For each sensing unit in each historical detection, the differential mode drift weights can be determined and stored in a database during historical detection. This application can obtain the differential mode drift weights of each sensing unit in each historical detection by retrieving data from the database.

[0155] Differential mode drift weight The larger the value (indicating more severe differential mode drift), the higher the fitting weight. The smaller the value, the higher the value of the sensing unit. The smaller the contribution of each data point to the common-mode drift fitting, the greater the contribution; conversely, the larger the contribution of the differential-mode drift weight. The smaller the value, the higher the fitting weight. The larger the value, the higher the value of the sensing unit. More reliable data points contribute more to the common-mode drift fitting. The weighted fitting strategy described above effectively suppresses the interference of differential-mode drift on common-mode drift quantization and improves the accuracy of the fitting function.

[0156] Step 603: Based on the differential sequence of environmental parameters detected by each sensing unit and the corresponding common-mode drift fitting function, determine the common-mode drift of the acetylene sensor array.

[0157] In one possible implementation, this application can input the change in environmental parameters between the current detection and the benchmark detection of each sensing unit into the common-mode drift fitting function corresponding to the sensing unit to obtain the unit drift amount corresponding to the sensing unit. Then, based on the unit drift amounts corresponding to each sensing unit, a weighted fusion is performed to obtain the common-mode drift amount of the acetylene sensor array.

[0158] For example, the common-mode drift satisfies the following formula:

[0159]

[0160] in, This represents the common-mode drift of the acetylene sensor array. This represents the number of sensing units in the acetylene sensor array. For the first The fusion weights of individual sensing units, For the first The common-mode drift fitting function of each sensing unit. For the first The change in environmental parameters of each sensing unit between the current detection and the benchmark detection. For the first Each sensing unit is in response to changes in environmental parameters of... The change in the response signal at that time.

[0161] In some embodiments, for the above-mentioned fusion weights, this application can calculate the th among all detections. The ratio of the average differential mode drift weight of each sensing unit to the sum of the average differential mode drift weights of all sensing units in all detections is used, and the difference between 1 and this ratio is taken as the value of the first... The fusion weights of individual sensing units are adjusted inversely using the mean of differential mode drift weights; sensing units with more severe differential mode drift have smaller fusion weights.

[0162] In other embodiments, since differential mode drift weights have been introduced during multivariate fitting, the influence weight of coordinate points with excessive differential mode drift during fitting is reduced. During the function fitting process of a certain sensing unit, if the sensing unit exhibits differential mode drift, the irreversibility of this drift will lead to increasingly poor function fitting quality. Therefore, this application can... The ratio of the determination coefficient of the common-mode drift fitting function of the first sensing unit to the sum of the determination coefficients of the common-mode drift fitting functions of all sensing units is used as the value of the first... The fusion weights of each sensing unit. The coefficient of determination is used to characterize the degree of fit of the common-mode drift fitting function to the data, and can be expressed as the difference between 1 and the ratio of the residual sum of squares (RSS) to the total sum of squares (TSS). The closer the coefficient of determination is to 1, the better the fitting function is, and the larger the fusion weights are.

[0163] Based on the above technical solutions, this application eliminates the influence of signal offset in benchmark detection through differential sequences, reduces the interference of differential mode drift on common mode drift analysis through weighted fitting, and integrates the fitting results of each sensing unit through weighted fusion, thereby achieving accurate quantification of the common mode drift. These technical solutions fully consider the impact of individual differences on common mode drift analysis, solve the error problem caused by fitting of a single sensing unit, and improve the accuracy of common mode drift suppression.

[0164] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0165] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for collaborative suppression of differential-mode and common-mode drift in an acetylene sensor array, characterized in that, include: The detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array are obtained; the detection data includes the response signal sequence of each sensing unit in the acetylene sensor array and the environmental parameter sequence. Drift analysis is performed based on the detection data corresponding to the multiple historical detections and the current detection to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array. The differential mode drift weight is used to characterize the degree of signal drift caused by individual differences of the corresponding sensing unit. The common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array caused by environmental factors. Based on the differential mode drift weight and the common mode drift amount, the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array is subjected to collaborative filtering to obtain the target response signal sequence of each sensing unit after drift suppression. Determining the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array includes: The detection data from multiple historical detections are clustered and grouped according to the environmental parameter sequence in the detection data, resulting in multiple clusters; Differential analysis is performed on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array; A baseline detection is determined from the environmental parameter sequence of the multiple historical detections and the current detection. Based on the correlation analysis of the benchmark detection data, the multiple historical detection data, and the current detection data, the common mode drift of the acetylene sensor array is determined. The method involves performing difference analysis on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array, including: For each cluster and each sensing unit, the response feature difference value between the sensing unit in the current detection and the cluster is determined based on the response signal sequence of the sensing unit in each historical detection and the current detection in the cluster. Based on the response feature difference value of the sensing unit between the current detection and each cluster, the feature difference coefficient of the sensing unit between the current detection and the multiple historical detections is determined; For each sensing unit, the differential mode drift weight of the sensing unit is determined based on the characteristic difference coefficient between the current detection and the multiple historical detections of each sensing unit in the acetylene sensor array. Alternatively, a difference analysis can be performed on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array, including: For each cluster and each sensing unit, based on the response signal sequence of each historical detection and the current detection of the sensing unit in the cluster, the characteristic coefficients between the current detection and each historical detection in the cluster are determined; the characteristic coefficients include response rise rate characteristic coefficients, recovery rate characteristic coefficients and peak amplitude characteristic coefficients. For each cluster, the feature matrix corresponding to the cluster is determined based on the feature coefficients between the current detection and each historical detection in the cluster by each sensing unit. Local anomaly detection is performed on the feature matrix corresponding to each cluster to determine the differential mode drift weight of each sensing unit in the acetylene sensor array. Based on the environmental parameter sequence of the multiple historical detections and the current detection, a benchmark detection is determined from the multiple historical detections and the current detection, including: The environmental parameter sequences from the multiple historical tests and the current test are compared with standard environmental parameters to determine the environmental parameter differences between the multiple historical tests and the current test and the standard environmental parameters. Based on the environmental parameter differences between the environmental parameter sequences from the multiple historical tests and the current tests and the standard environmental parameters, a benchmark test is determined from the multiple historical tests and the current tests. Based on the correlation analysis of the benchmark detection data, the multiple historical detection data, and the current detection data, the common-mode drift of the acetylene sensor array is determined, including: For each sensing unit, based on the response signal sequence and environmental parameter sequence of the sensing unit in the reference detection, the multiple historical detections and the current detection, the environmental parameter difference sequence and response signal difference sequence of the sensing unit in each historical detection and the current detection are determined; Based on the weighted multivariate function fitting of the environmental parameter difference sequence and response signal difference sequence in each historical detection and the current detection by the sensing unit, the common-mode drift fitting function corresponding to the sensing unit is determined; the common-mode drift fitting function is used to characterize the influence relationship of environmental parameter changes on the response signal; The common-mode drift of the acetylene sensor array is determined based on the differential sequence of environmental parameters detected by each sensing unit and the corresponding common-mode drift fitting function.

2. The method for coordinated suppression of differential-mode and common-mode drift in an acetylene sensor array according to claim 1, characterized in that, Based on the weighted multivariate function fitting of the environmental parameter difference sequence and response signal difference sequence of each historical detection and the current detection by the sensing unit, the common-mode drift fitting function corresponding to the sensing unit is determined, including: For each of the multiple historical detections and the current detection, and for each sensing unit, determine the changes in environmental parameters and response signals of the sensing unit between the detection and the benchmark detection. The common-mode drift fitting function corresponding to the sensing unit is determined by performing a weighted multivariate function fitting on the changes in environmental parameters and response signals between the sensing unit in historical detection and current detection and the benchmark detection.

3. The method for coordinated suppression of differential-mode and common-mode drift in an acetylene sensor array according to claim 2, characterized in that, Based on the differential sequence of environmental parameters detected by each sensing unit and the corresponding common-mode drift fitting function, the common-mode drift of the acetylene sensor array is determined, including: For each sensing unit, the change in environmental parameters between the current detection and the benchmark detection is input to the common-mode drift fitting function corresponding to the sensing unit to obtain the unit drift amount corresponding to the sensing unit. The common-mode drift of the acetylene sensor array is obtained by weighted fusion based on the unit drift of each sensing unit.

4. The method for coordinated suppression of differential-mode and common-mode drift in an acetylene sensor array according to claim 1, characterized in that, Based on the differential mode drift weight and the common mode drift amount, the response signal sequences of each sensing unit in the currently detected data of the acetylene sensor array are subjected to collaborative filtering to obtain the target response signal sequences of each sensing unit after drift suppression, including: Based on the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array and the common mode drift, the preliminary correction signal sequence of each sensing unit is determined. For each sensing unit, based on the differential mode drift weight and the preliminary correction signal sequence of the sensing unit, the Kalman filter algorithm is used to filter the signal to obtain the target response signal sequence of the sensing unit after drift suppression.

5. A differential-mode and common-mode drift cooperative suppression system for an acetylene sensor array, characterized in that, include: The data acquisition unit is used to acquire the detection data corresponding to multiple historical detections and the current detection of the acetylene sensor array; the detection data includes the response signal sequence of each sensing unit in the acetylene sensor array and the environmental parameter sequence. The drift analysis unit is used to perform drift analysis based on the detection data corresponding to the multiple historical detections and the current detection, to determine the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array; the differential mode drift weight is used to characterize the degree of signal drift of the corresponding sensing unit due to individual differences; the common mode drift is used to characterize the degree of overall signal drift of the acetylene sensor array due to environmental factors. The collaborative filtering unit is used to perform collaborative filtering on the response signal sequence of each sensing unit in the current detection data of the acetylene sensor array according to the differential mode drift weight and the common mode drift amount, so as to obtain the target response signal sequence of each sensing unit after drift suppression. Determining the differential mode drift weight of each sensing unit in the acetylene sensor array and the common mode drift of the acetylene sensor array includes: The detection data from multiple historical detections are clustered and grouped according to the environmental parameter sequence in the detection data, resulting in multiple clusters; Differential analysis is performed on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array; A baseline detection is determined from the environmental parameter sequence of the multiple historical detections and the current detection. Based on the correlation analysis of the benchmark detection data, the multiple historical detection data, and the current detection data, the common mode drift of the acetylene sensor array is determined. The method involves performing difference analysis on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array, including: For each cluster and each sensing unit, the response feature difference value between the sensing unit in the current detection and the cluster is determined based on the response signal sequence of the sensing unit in each historical detection and the current detection in the cluster. Based on the response feature difference value of the sensing unit between the current detection and each cluster, the feature difference coefficient of the sensing unit between the current detection and the multiple historical detections is determined; For each sensing unit, the differential mode drift weight of the sensing unit is determined based on the characteristic difference coefficient between the current detection and the multiple historical detections of each sensing unit in the acetylene sensor array. Alternatively, a difference analysis can be performed on the response signal sequences between the current detection and historical detections in multiple clusters to determine the differential mode drift weight of each sensing unit in the acetylene sensor array, including: For each cluster and each sensing unit, based on the response signal sequence of each historical detection and the current detection of the sensing unit in the cluster, the characteristic coefficients between the current detection and each historical detection in the cluster are determined; the characteristic coefficients include response rise rate characteristic coefficients, recovery rate characteristic coefficients and peak amplitude characteristic coefficients. For each cluster, the feature matrix corresponding to the cluster is determined based on the feature coefficients between the current detection and each historical detection in the cluster by each sensing unit. Local anomaly detection is performed on the feature matrix corresponding to each cluster to determine the differential mode drift weight of each sensing unit in the acetylene sensor array. Based on the environmental parameter sequence of the multiple historical detections and the current detection, a benchmark detection is determined from the multiple historical detections and the current detection, including: The environmental parameter sequences from the multiple historical tests and the current test are compared with standard environmental parameters to determine the environmental parameter differences between the multiple historical tests and the current test and the standard environmental parameters. Based on the environmental parameter differences between the environmental parameter sequences from the multiple historical tests and the current tests and the standard environmental parameters, a benchmark test is determined from the multiple historical tests and the current tests. Based on the correlation analysis of the benchmark detection data, the multiple historical detection data, and the current detection data, the common-mode drift of the acetylene sensor array is determined, including: For each sensing unit, based on the response signal sequence and environmental parameter sequence of the sensing unit in the reference detection, the multiple historical detections and the current detection, the environmental parameter difference sequence and response signal difference sequence of the sensing unit in each historical detection and the current detection are determined; Based on the weighted multivariate function fitting of the environmental parameter difference sequence and response signal difference sequence in each historical detection and the current detection by the sensing unit, the common-mode drift fitting function corresponding to the sensing unit is determined; the common-mode drift fitting function is used to characterize the influence relationship of environmental parameter changes on the response signal; The common-mode drift of the acetylene sensor array is determined based on the differential sequence of environmental parameters detected by each sensing unit and the corresponding common-mode drift fitting function.

Images