CN-121979025-A - Real-time monitoring control system and controller based on thermal flow sensor

CN121979025ACN 121979025 ACN121979025 ACN 121979025ACN-121979025-A

Abstract

The invention discloses a real-time monitoring control system and a controller based on a thermal flow sensor, and relates to the technical field of sensor monitoring. The real-time monitoring control system and the controller based on the thermal flow sensor comprise S1, preprocessing sensor monitoring data and historical statistical data, S2, extracting lag and response time dynamic characteristics, carrying out modal decomposition, phase estimation, peak value calculation and steady-state noise analysis, S3, carrying out phase compensation, peak value recovery and enhancement compensation judgment, recording residual sequences and compensation parameters, outputting optimized flow data, S4, constructing a reliability interval and dynamic consistency score synthesis control reliability mark, and carrying out safety constraint, prediction filtration and archiving. The problems that the existing thermal flow sensor generates obvious phase lag and peak flattening distortion under the influence of thermal inertia of thermal diffusion under the rapid fluctuation and pulsation working conditions, and the real-time monitoring precision and the closed-loop control stability are difficult to guarantee due to the lack of an online dynamic response analysis and compensation mechanism are solved.

Inventors

CHEN XINZHUN
ZHANG BIN
LI NINGZI
XU WENJI
LI NA

Assignees

广州奥松电子股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (10)

1. Real-time monitoring control system based on thermal type flow sensor, its characterized in that includes following steps: S1, collecting sensor monitoring data, and acquiring historical statistical data; S2, constructing an edge window based on sensor monitoring data, extracting lag and response time dynamic characteristics, and performing modal decomposition, phase estimation, peak value calculation and steady-state noise analysis; s3, generating a trigger threshold value according to dynamic characteristic statistics, and executing phase compensation, peak value recovery and enhancement compensation judgment; s4, constructing a credible interval based on the residual sequence, synthesizing a control credibility mark with the dynamic consistency score, and executing safety constraint and prediction filtering according to the control credibility mark to complete the whole process archiving.
2. The real-time monitoring control system based on the thermal flow sensor according to claim 1, wherein the specific process of acquiring the historical statistical data is as follows: Collecting sensor monitoring data, wherein the sensor monitoring data comprise instantaneous flow data, execution quantity sequence, sensor temperature data, heating driving data, pipeline pressure data, valve opening instruction data, air pump driving signal data and sampling time stamp data; Acquiring historical statistical data and creating a historical statistical database, wherein the historical statistical data comprises historical instantaneous flow data, historical sensor temperature data, historical valve opening instruction data, historical air pump driving signal data and historical steady-state change rate; The heating driving data refer to sampling values of heating driving current and heating driving voltage, and a sampling time interval is obtained by means of time stamp difference.
3. The real-time monitoring control system based on thermal flow sensor according to claim 1, wherein the specific process of preprocessing the sensor monitoring data and the historical statistical data is as follows: The method comprises the steps of performing deburring processing on instantaneous flow data, pipeline pressure data and sensor temperature data through a sliding window median algorithm, removing isolated peaks and communication jitter anomalies, performing time alignment on data sequences in sensor monitoring data and historical statistical data by taking sampling timestamp data as a reference, and performing standardization and normalization processing on the sensor monitoring data and the historical statistical data through a distribution standardization and linear normalization algorithm.
4. The real-time monitoring control system based on the thermal flow sensor according to claim 1, wherein the specific process of constructing an edge window based on sensor monitoring data and extracting the dynamic characteristics of hysteresis and response time is as follows: When a valve actuator exists, valve opening instruction data is used as a control input reference, when the air pump actuator exists, air pump driving signal data is used as a control input reference, bayesian online variable point detection is carried out on a control input reference sequence and pipeline pressure data, edge occurrence time stamps are extracted, subsection monotonicity constraint edge detection is carried out on instantaneous flow data, flow response edge time stamps are extracted, difference is carried out on the two types of edge time stamps under the same time base of sampling time stamp data, monitoring edge lag time is obtained, edge window is obtained by combining edge detection results of the control input reference data and the pipeline pressure data under the same time base of the sampling time stamp data, percentage response estimation based on quantile regression is further carried out in the edge window, response rising time and response falling time are obtained by locating percentage response points of instantaneous flow data in the same window, and response rising time and response falling time are calculated.
5. The real-time monitoring control system based on the thermal flow sensor according to claim 1, wherein the specific processes of performing modal decomposition, phase estimation, peak value calculation and steady-state noise analysis are as follows: Performing variation modal decomposition on the instantaneous flow data in a pulsation window, separating a main pulsation mode and a high-frequency noise mode, wherein the pulsation window is jointly determined by the instantaneous flow data and control input reference data under the same time base of sampling time stamp data through periodic feature detection; the peak clipping distortion is calculated by comparing the reference peak amplitude with the actual peak amplitude, a steady-state window is determined according to transient flow data, short-time variance of pipeline pressure data, first-order differential amplitude and a variable point detection result of control input reference data, a static amplitude mapping coefficient from control input reference to transient flow is established in the steady-state window by using historical transient flow data, historical valve opening instruction data and historical air pump driving signal data through recursive least square with forgetting factors; And outputting a monitoring accuracy analysis result, wherein the monitoring accuracy analysis result comprises monitoring edge lag time, response rising time, response falling time, period phase lag time, peak value maintaining coefficient and steady-state noise intensity.
6. The real-time monitoring control system based on thermal flow sensor according to claim 1, wherein the specific process of generating trigger threshold value and executing phase compensation, peak value recovery and enhancement compensation judgment according to dynamic feature statistics is as follows: Based on a criterion of a steady-state window and the stability of a periodic structure, a strategy stability judging window constructed in a continuous pulsation window is used for obtaining a steady window, based on statistical distribution of monitoring accuracy analysis results, a Bayesian quantile estimation algorithm is adopted for respectively generating dynamic response optimization thresholds, wherein the dynamic response optimization thresholds comprise a hysteresis time optimization threshold, a coefficient optimization threshold and a response time optimization threshold, and the cycle phase hysteresis time, the peak value keeping coefficient, the response rising time, the response falling time and the dynamic response optimization threshold are compared in real time: When the period phase lag time is smaller than or equal to the lag time optimizing threshold, the existing phase compensation state is kept unchanged and monitoring is continuously executed; When the peak value maintaining coefficient is larger than or equal to the coefficient optimizing threshold value, the original peak value compensation strategy is kept unchanged and monitoring is continuously executed; The method comprises the steps of triggering an enhanced compensation strategy when the response rising time and the response falling time are simultaneously larger than a response time optimizing threshold value, and maintaining the current compensation strategy unchanged when the response rising time and the response falling time are simultaneously smaller than or equal to the response time optimizing threshold value.
7. The real-time monitoring control system based on the thermal flow sensor according to claim 1, wherein the specific process of correcting the dynamic distortion, recording the residual sequence and updating the compensation parameters, and outputting the monitoring accuracy optimized flow data is as follows: The phase compensation strategy adopts a state observation framework of unscented Kalman filtering, wherein instantaneous flow data is taken as an observed quantity, control input reference data, heating driving current data and heating driving voltage data are taken as exogenous driving quantities, a dynamic equation containing a hysteresis state is established, the dynamic equation adopts a two-state structure, the main flow state is updated at each sampling time stamp according to the change trend of the control input reference data, the main flow state at the last time and the hysteresis energy state at the last time, the hysteresis energy state is updated according to the heating power formed by the hysteresis energy state at the last time, the heating driving current data and the heating driving voltage data, a hysteresis correction flow estimated value is output at each sampling time stamp, the flow estimated value after hysteresis correction is taken as a predicted quantity, the instantaneous flow data is taken as the observed quantity, an unscented Kalman filtering updating process is established, and an unscented Kalman filtering residual sequence and an updating compensation parameter are recorded; The peak value recovery strategy adopts regularized deconvolution reconstruction with amplitude constraint, namely constructing a pulse transfer function for deconvolution by utilizing a reference peak value amplitude and a peak value holding coefficient, recovering peak value information in instantaneous flow data, and taking the peak value holding coefficient of an equivalent amplitude response of the pulse transfer function; The enhanced compensation strategy adopts feedforward-feedback coupling, wherein a feedforward item is obtained by controlling input reference data to be subjected to recursive least square online updating with forgetting factors, a feedback item is obtained by unscented Kalman filtering residual error closed loop correction, compensation parameters are subjected to sensor temperature segmentation correction, temperature segmentation boundaries are obtained by clustering historical sensor temperature data through a Gaussian mixture model, and output monitoring accuracy optimizes flow data, optimizes triggering marks and compensation parameter records.
8. The real-time monitoring control system based on the thermal flow sensor according to claim 1, wherein the specific process of constructing a trusted interval based on a residual sequence and synthesizing a control reliability mark with a dynamic consistency score is as follows: Inputting unscented Kalman filtering residual sequences in the compensation parameter records, constructing a flow residual distribution model, taking the unscented Kalman filtering residual sequences as training samples, adopting order-preserving quantile regression to fit a residual condition quantile function, outputting a residual credible upper bound and a residual credible lower bound, and superposing the residual credible upper bound and the residual credible lower bound on monitoring accuracy optimization flow data to obtain a flow credible interval upper bound and a flow credible interval lower bound; And (3) aligning and calculating the monitoring accuracy optimized flow data with a main pulse mode of the control input reference data through dynamic time warping to obtain a dynamic consistency score, constructing the dynamic consistency score based on the cycle phase lag time and the peak value maintaining coefficient, synthesizing a control reliability mark from the flow reliability interval, the dynamic consistency score and the monitoring accuracy analysis result, and synchronously outputting the control reliability mark.
9. The real-time monitoring control system based on thermal flow sensor according to claim 1, wherein the specific process of executing safety constraint and predictive filtering and completing whole process archiving according to the control reliability mark is as follows: When the control reliability mark enters a first-level risk zone, a model prediction safety filter is started, a limited prediction domain model prediction control structure is adopted by the model prediction safety filter, soft constraint penalty is carried out on the flow deviation and the variation of the execution quantity by a deviation penalty matrix and a variation penalty matrix, the length of a prediction domain is consistent in a range from a to b sampling periods according to a control period and a working condition dynamic characteristic, the sampling period and a monitoring data sampling interval are kept consistent, the current execution quantity variation rate and the safety boundary construct a control barrier function, the control barrier function is greater than zero and serves as a variation rate safety constraint, the output execution quantity variation rate is updated in the safety boundary; Creating a real-time monitoring database, executing event-level archiving, wherein the archiving content is indexed by a sampling time stamp and comprises monitoring accuracy optimization flow data, a flow credible interval upper bound, a flow credible interval lower bound, a dynamic consistency score, a control credibility mark, a protection trigger mark, a safety execution amount correction amplitude and a corresponding monitoring accuracy analysis result.
10. A thermal flow sensor based real-time monitoring controller, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor executing the program to implement the monitoring control system of the controller according to any one of claims 1-9.

Description

Real-time monitoring control system and controller based on thermal flow sensor Technical Field The invention relates to the technical field of sensor monitoring, in particular to a real-time monitoring control system and a controller based on a thermal flow sensor. Background The thermal type gas flow sensor has the characteristics of compact structure, wide range coverage, capability of realizing direct measurement of mass flow and the like, and is widely applied to medical equipment, small-sized gas circuit systems, precise industrial process control and various embedded gas supply devices. In such systems, flow monitoring typically works in conjunction with a valve actuator, an air pump drive unit, a line pressure measurement device, and an ambient temperature acquisition unit to form a real-time detection and regulation link around the gas flow conditions. With the development of the gas circuit system to the rapid adjustment, small flow resolution, periodic pulsation conveying and multi-scene coupling control direction, flow data, control instruction data, pressure data and temperature data acquired in real time show higher dynamic property and multisource property, and the dependence of the system on data alignment, change trend identification, working condition association analysis and loop stability is continuously enhanced. In order to ensure the reliability of the flow monitoring and controlling process, related applications generally need to comprehensively judge the flow variation behavior by utilizing historical statistical data, real-time multi-source signals and gas path operation characteristics, thereby supporting closed-loop regulation, operation monitoring and continuous and stable work. For example, the invention patent with publication number CN120670710A discloses a flushing and cooling method, a system and equipment with an automatic identification route, the method collects equipment surface space point cloud data through a three-dimensional laser scanning sensor and builds a three-dimensional structure model, combines a material identification sensor and a temperature collection network to obtain material properties, a heat conduction coefficient and multi-source temperature gradient distribution, builds a thermal load distribution diagram and a heat dissipation demand model on the basis of sensor data fusion, determines a flushing point sequence and a shortest route according to a data-driven path planning algorithm, then generates control parameters such as flushing pressure, flow, duration and the like through a parameter self-adaptive algorithm according to the area thermal load density, monitors temperature change data in real time in the flushing execution process, dynamically adjusts the flushing parameters based on sensor feedback, and realizes automatic flushing and cooling control based on sensing and data decision. The invention patent with publication number CN116854156B discloses a factory sewage quality monitoring and purifying method and system based on the Internet of things, the method collects monitoring data such as pollutant concentration, flow rate and flow rate in the sewage discharge process through a distributed water quality and flow sensor network, a diffusion model of pollutants along a discharge path is constructed by using a sensor data analysis result, a control instruction is sent to a purifying end based on the model to determine interception time, purifying amount and disinfection material adding amount, meanwhile, whether the operation condition reaches a working limit or not is judged by combining discharge state data recorded by a sensor and operation conditions fed back by the purifying end, and purifying operation is adjusted when the operation condition reaches the limit, and the whole system takes sensor collection, data fusion, diffusion modeling and control linkage as a core to realize collaborative management of the sewage monitoring and purifying process. Although the existing sensor monitoring and data processing technology can realize multi-source acquisition, state identification and simple dynamic analysis, the existing thermal flow monitoring method generally lacks accurate quantification of phase lag and peak clipping distortion under the rapid fluctuation working condition, the online dynamic response analysis often depends on fixed threshold or experience correction, real-time accurate separation and compensation of lag, peak clipping and noise are difficult to realize, and meanwhile, an online judging mechanism based on residual error, dynamic consistency and trusted boundary is lacking, closed-loop constraint cannot be formed on data reliability and control safety in the compensation process, so that monitoring precision and control stability are not facilitated to be ensured. Therefore, in view of the above problems, there is a need for a thermal flow sensor-based real-time monitoring control system and controller. Disclosure o