CN117606578A

CN117606578A - Intelligent gas flow monitoring method

Info

Publication number: CN117606578A
Application number: CN202311573415.3A
Authority: CN
Inventors: 蔡剑锋; 陈信坛; 陈晓碧
Original assignee: Shenzhen Yihejia Intelligent Medical Technology Co ltd
Current assignee: Shenzhen Yihejia Intelligent Medical Technology Co ltd
Priority date: 2023-11-23
Filing date: 2023-11-23
Publication date: 2024-02-27

Abstract

The application provides an intelligent gas flow monitoring method, which comprises the following steps: detecting the current temperature of the gas, constructing a gas temperature prediction model, and obtaining a specific value of temperature change; a linear regression algorithm is utilized to establish a gas thermal expansion or contraction prediction model, and the degree of thermal expansion or contraction of the gas is calculated based on the gas temperature change data; predicting gas volume change according to real-time temperature and pressure monitoring; establishing a flow velocity and volume relation model according to the gas flow velocity and volume change data, and predicting gas flow velocity change based on gas volume change; calculating a difference value from a predicted flow rate through real-time flow measurement data; automatically adjusting and calibrating the flow measurement device according to the anomaly detection and stability assessment of the flow data; according to the history and the real-time monitoring data, automatically adjusting the calibration period of the flow measurement device; using the predicted data and the real-time data, a flow measurement calibration of the device is performed.

Description

Intelligent gas flow monitoring method

Technical Field

The invention relates to the technical field of information, in particular to an intelligent gas flow monitoring method.

Background

In modern industry and scientific research, accurate measurement of gas flow is an important component of many critical processes. It has wide application in various fields such as energy management, chemical industry, environmental monitoring and laboratory research. However, accurately measuring gas flow presents a number of technical challenges. The physical properties of the gas cause it to thermally expand or contract as the temperature changes, thereby affecting its volume and flow rate. This phenomenon complicates accurate flow measurement in environments with large temperature fluctuations. In addition to temperature, other environmental parameters such as pressure, humidity can also affect the flow characteristics of the gas. Maintaining measurement accuracy and consistency under varying environmental conditions is a challenge. Periodic calibration is required because of changes in environmental conditions and long-term use of the device, and the flow measurement device may experience performance bias. However, determining when to perform the calibration and how to accurately perform the calibration operation is a technical challenge. With the complexity of industrial processes and the increase of the requirements for data accuracy, the need for purely relying on traditional measurement methods cannot be met. How to efficiently process and analyze measurement data, and how to accurately predict flow changes based on environmental changes, becomes a critical issue. Under the trend of fast-developing industrial automation and intellectualization, the traditional manual monitoring and calibration method gradually appears to be inefficient and lagging. How to realize the automation and the intellectualization of the flow measurement process, and improve the efficiency and the accuracy is a current technical problem. Thus, a major problem faced in the field of gas flow measurement is how to accurately measure and predict gas flow under changing environmental conditions while increasing the level of automation and intelligence of the measurement device. The resolution of these problems is critical to ensuring the efficiency, safety and environmental compliance of industrial processes.

Disclosure of Invention

The invention provides an intelligent gas flow monitoring method, which mainly comprises the following steps:

detecting the current temperature of the gas, constructing a gas temperature prediction model, and obtaining a specific value of temperature change; a linear regression algorithm is utilized to establish a gas thermal expansion or contraction prediction model, and the degree of thermal expansion or contraction of the gas is calculated based on the gas temperature change data; predicting gas volume change according to real-time temperature and pressure monitoring; establishing a flow velocity and volume relation model according to the gas flow velocity and volume change data, and predicting gas flow velocity change based on gas volume change; calculating a difference value from a predicted flow rate through real-time flow measurement data; automatically adjusting and calibrating the flow measurement device according to the anomaly detection and stability assessment of the flow data; according to the history and the real-time monitoring data, automatically adjusting the calibration period of the flow measurement device; using the predicted data and the real-time data, a flow measurement calibration of the device is performed.

Further, the detecting the current temperature of the gas, constructing a gas temperature prediction model, and obtaining a specific value of the temperature change, including:

acquiring real-time temperature data of gas through a temperature sensor, and continuously recording the gas temperature data provided by the sensor by adopting a data acquisition system; acquiring a time stamp of each temperature data point through sensor data, and performing time sequence arrangement on the acquired temperature data; according to historical gas temperature data, performing model training by adopting a linear regression algorithm, constructing a gas temperature prediction model, predicting the trend of the gas temperature data, and determining the maximum value, the minimum value and the average value in the temperature data to obtain a specific value of temperature change; judging whether the temperature change exceeds a preset normal operation range; if the temperature change is beyond the normal range, an abnormal report is obtained, including a temperature change value and a time stamp.

Further, the method for establishing a gas thermal expansion or contraction prediction model by using a linear regression algorithm, calculating the degree of thermal expansion or contraction of the gas based on the gas temperature change data, includes:

acquiring real-time temperature data through a gas temperature sensor; performing data preprocessing on the real-time temperature data, including removing noise and abnormal values; establishing a gas thermal expansion or cold contraction prediction model by using the processed temperature data through a linear regression algorithm; according to the current temperature data, a preliminary prediction result of thermal expansion or cold contraction of the gas is obtained; obtaining a prediction error of each data point by calculating a difference value between a prediction value and an actual observation value, and calculating an overall error of the whole prediction result by adopting a method for calculating a mean square error; determining the accuracy of the model according to the distribution condition of errors; if the error value is smaller than the preset threshold value and the distribution is uniform, the model is considered to be accurate, and if the error value is larger than the preset threshold value or the distribution is uneven, the model is indicated to need further optimization; if the model needs to be adjusted, parameter adjustment is carried out to improve the gas thermal expansion or cold contraction prediction model; and obtaining the optimized prediction result of the thermal expansion or the cold contraction of the gas according to the improved thermal expansion or cold contraction prediction model of the gas.

Further, the predicting gas volume change based on real-time temperature and pressure monitoring includes:

acquiring a temperature prediction result of the gas at a certain moment or in a certain time period in the future by using a gas temperature prediction model; calculating an expected gas volume change by substituting the predicted temperature value into the ideal gas law using the ideal gas law pv=nrt, wherein P is the pressure, V is the volume, n is the amount of the substance of the gas, R is the ideal gas constant, and T is the temperature; acquiring current pressure data of the gas measured in real time through a pressure sensor; combining the pressure data measured in real time with the predicted temperature data to recalculate the gas volume change; determining an initial predicted value of the gas volume change after combining the temperature and pressure data; acquiring historical volume change data of the same type of gas under similar conditions; comparing the current prediction result with historical data; based on the comparison of the historical data, adjusting and correcting the current volume change prediction; continuously monitoring the temperature and pressure of the gas, updating data in real time, further refining and updating the prediction of the volume change according to the latest monitoring data, and determining the final prediction of the volume change of the gas.

Further, the establishing a flow rate and volume relation model according to the data of the gas flow rate and volume change, predicting the gas flow rate change based on the gas volume change, includes:

acquiring historical gas flow rate and volume change data, processing initial volume change data by using a Navier-Stokes equation, and establishing a flow rate and volume relation model; setting initial parameters of a relation model based on historical data, and predicting the flow velocity change of gas; comparing a flow velocity prediction result obtained through a flow velocity and volume relation model with volume change data measured in real time; if the deviation between the predicted result and the actual data is greater than a threshold value, dynamically adjusting parameters in an algorithm; continuously monitoring the volume change and the flow rate change of the gas, and if the volume change or the difference between the flow rate and the predicted value which are actually measured is larger than a preset threshold value, correcting the flow rate prediction by using a preset adjustment strategy; according to the gas flow rate data, using an autoregressive model to evaluate the stability of the flow rate change prediction; if the flow speed change prediction is abnormal, carrying out real-time dynamic correction on the flow speed prediction by using a Kalman filter; acquiring flow velocity prediction data corrected by a Kalman filter, and performing real-time monitoring again to verify the effectiveness of correction; if the corrected flow rate change data still has deviation, continuously using a Navier-Stokes equation to dynamically adjust parameters, and continuously adjusting until the flow rate change prediction data is stable; further comprises: the flow rate of the multiphase fluid is calculated and predicted based on the mixed fluid characteristics and the real-time data adjustments.

The method for calculating and predicting the flow rate of the multiphase fluid according to the characteristics of the mixed fluid and the real-time data adjustment specifically comprises the following steps:

based on the effect of the combination of multiphase fluids on the flow rate, a multiphase fluid flow rate prediction model v=k× (w _oil ×V _oil +w _water ×V _water +w _gas ×V _gas ) Wherein V represents the total flow rate, w _oil Is the volume ratio of oil in the mixture, w _water Is the volume ratio of water in the mixture, w _gas Is the volume ratio of the gas in the mixture, V _oil Is the basic flow rate of oil, V _water Is the basic flow rate of water, V _gas Is the base flow rate of the gas and k is an adjustment factor. Measuring basic flow velocity V in pure oil, pure water and pure gas environments respectively _oil ，V _water ，V _gas Flow rate data for each phase of fluid at different pressures and temperatures was recorded. The total flow velocity V of the mixed fluid is measured by mixing different proportions of oil, water, gas. A nonlinear regression analysis is used to determine the coefficient k in the multiphase fluid flow rate prediction model. In an actual operation scene, a sensor is used for monitoring the proportion w of oil, water and gas in real time _oil ，w _water ，w _gas . Applying the adjusted multiphase flow according to the real-time monitored fluid proportionThe volumetric flow rate prediction model predicts the flow rate. The value of k is dynamically adjusted to account for changes in fluid proportions based on real-time and historical data analysis. The difference between the predicted flow rate and the actual measured flow rate is compared periodically, and the accuracy and reliability of the multiphase fluid flow rate prediction model is evaluated. And further adjusting parameters in the formula according to the performance evaluation result, and optimizing the multiphase fluid flow velocity prediction model. A sophisticated monitoring system is implemented to continuously acquire data on the flow rate, proportion and environmental conditions of the multiphase fluid. And establishing a feedback mechanism, and continuously optimizing a prediction formula according to comparison of the monitoring data and the prediction result.

Further, the calculating the difference value from the predicted flow rate according to the real-time flow measurement data comprises:

recording the fluid flow in real time by using a sensor, and processing the acquired flow data by adopting a filtering and noise elimination technology; according to the gas volume data, a flow velocity and volume relation model is used for obtaining a predicted value of the gas flow velocity; calculating a difference between the flow rate monitored in real time and the flow rate predicted by the flow rate and volume relationship model; detecting whether the difference value of the flow velocity exceeds a preset safety threshold value, and triggering an abnormality detection mechanism if the difference value exceeds the threshold value; according to the abnormal detection result, based on the real-time flow measurement data and an analysis algorithm, automatically adjusting parameters of a flow velocity and volume relation model; updating flow measurement data in real time, and keeping the real-time property and the latest state of the data; continuously monitoring and analyzing the flow speed difference value, and if the difference value is kept within a preset range, not adjusting model parameters; automatically recording and storing all flow and flow rate data; further comprises: the fluid flow rate measurement and prediction process is continuously optimized according to the real-time environment variable and the flow rate data.

The measuring and predicting process for continuously optimizing the fluid flow rate according to the real-time environment variable and the flow rate data specifically comprises the following steps:

According to the density, temperature and pressure of the fluid, a flow velocity prediction model is established, and the flow velocity prediction model formula is F=a×D ^b ×T ^c ×P ^d Wherein F represents flow rate, D represents fluid density, T represents temperature, P represents pressure, a, b, c, D are according toThe parameters determined by the experimental data represent the extent to which these variables affect the flow rate. Data of D, T, P and actual flow rate F of the fluid are acquired under different environmental conditions. The parameters a, b, c, d in the formula are determined using regression analysis. D, T, P are measured in real time and the formula f=a×d is applied ^b ×T ^c ×P ^d And carrying out flow rate prediction. Recording the predicted flow rate F _pred And actually measuring the flow rate F _actual Differences between them. According to F _pred And F is equal to _actual The difference between the values of a, b, c, d are dynamically adjusted. Changes in environmental conditions and flow data are continuously monitored. The flow rate prediction model parameters are updated periodically with newly acquired data, ensuring that the model remains accurate and effective in a constantly changing environment. The performance of the predictive model is periodically assessed under extreme environmental conditions, including extreme temperatures, high or low pressures, abnormal humidity levels, chemicals or pollutants. And adjusting model parameters according to the performance evaluation result so as to cope with extreme environment conditions. And (3) compiling a comprehensive performance report, summarizing the performances of the model under different conditions, and providing decision support based on the report.

Further, the automatic adjustment and calibration of the flow measurement device according to anomaly detection and stability assessment of flow data comprises:

classifying the flow data according to the difference between the flow measurement data and the predicted flow rate, wherein the flow data comprises normal flow data and abnormal flow data, and judging the flow data to be the abnormal flow data if the difference between the flow measurement data and the predicted flow rate exceeds a preset threshold value; if the judgment result is abnormal flow data, judging whether the abnormality is caused by errors of the flow measurement equipment or not by comparing the data of different time points or the readings of other sensors; if the abnormality is caused by the equipment error, acquiring the state and performance parameters of the equipment, and evaluating the current working condition of the equipment; historical parameter data acquired by flow measurement equipment, including flow rate, temperature and pressure, trains a multi-layer sensor model; continuously acquiring readings under different flows, temperatures and pressures by using a flow measurement device, and evaluating the stability and accuracy of the output of the flow measurement device by using a pre-trained multi-layer sensor model; if the output of the multi-layer sensor analysis display equipment is unstable or inaccurate, the equipment automatically enters calibration, and the internal parameters of the equipment are automatically adjusted to correct the output error; after the calibration is finished, the equipment automatically returns to a normal working mode, and the equipment continues to perform flow measurement and data output; after calibration, the output data of the equipment is again evaluated by using the equipment diagnosis program, whether the equipment is successfully calibrated or not is judged, if the calibration is successful, the calibration flow is ended, and if the calibration is unsuccessful, the calibration is again carried out; further comprises: device performance monitoring and dynamic calibration adjustments are implemented based on environmental changes.

The device performance monitoring and dynamic calibration adjustment are implemented according to environmental changes, and specifically comprise:

and acquiring flow, temperature and pressure data of different temperatures, different fluid pressures and different fluid types through the sensor. And according to the data acquired under each environmental condition, model training is carried out by using a decision tree algorithm, the special data mode and abnormality of the environment are identified, and the influence of temperature and pressure changes on flow measurement is determined. Using the history and real-time data, a deep learning model based on a multi-layer perceptron is established, and the data characteristics under different environmental conditions are identified and adapted to various environmental conditions. And dynamically adjusting a calibration strategy according to the output of the multi-layer sensor model so as to adapt to the current environmental condition. And automatically adjusting the calibration setting of the equipment according to the analysis result of the multi-layer sensor model. Testing the calibrated device under different environmental conditions, recording performance data of the device, and comparing with the output of the predictive model to verify the calibration effect. And continuously monitoring the performance of the equipment under various environmental conditions, and periodically feeding back real-time data to the multi-layer perceptron model to perform iterative optimization of the model. Based on the test data under various circumstances, a comprehensive device performance assessment report is created that includes the performance of the device under various conditions, calibration effects, and any areas where further improvement is desired. Based on the performance evaluation report, maintenance and software updating of the device are periodically performed according to the problems and deficiencies indicated by the report.

Further, the automatic adjustment of the calibration period of the flow measurement device according to the historical and real-time monitoring data comprises:

acquiring equipment history calibration records, flow measurement data and environmental parameters including temperature and pressure; analyzing calibration frequency and performance fluctuation in historical data through an autoregressive model, and identifying key factors causing calibration requirement change, wherein the key factors comprise environmental condition change or equipment use frequency increase and decrease; according to the history and real-time data, a multi-layer sensor is adopted to construct a device calibration prediction model, and future calibration requirements are predicted; inputting historical calibration data and real-time monitoring parameters, and training a device to calibrate a prediction model; according to flow data and environmental parameters of the equipment, using an equipment calibration prediction model to determine the current equipment state and possible calibration requirements; judging whether the equipment needs to be calibrated according to the output of the equipment calibration prediction model and real-time data analysis, and determining the optimal time point of the calibration; when the equipment calibration prediction model predicts that the equipment needs to be calibrated, automatically triggering a calibration flow, and ensuring that the calibration flow meets the calibration requirement and the operation specification of the equipment; after calibration is completed, recording equipment performance data, and evaluating the calibration effect by comparing the flow measurement accuracy before and after calibration; and according to the calibration result and the operation data of the equipment, periodically adjusting the parameters of the equipment calibration prediction model.

Further, the performing flow measurement calibration of the device using the predicted data and the real-time data includes:

acquiring real-time data of flow measurement equipment, including flow rate, fluid density and temperature; if the predicted data is within the preset error range, performing model training by using a cyclic neural network algorithm according to the real-time data, constructing a device performance change prediction model, and detecting the change of the device performance or identifying possible abnormality; obtaining a preliminary data quality report through the analysis result of the cyclic neural network algorithm, wherein the report comprises evaluation of data consistency, any abnormal mode identified and overall evaluation of equipment performance; according to the flow rate, fluid density and temperature data of the equipment in a period of time in the future and the current real-time flow rate, fluid density and temperature data obtained from the flow measurement equipment, comparing the difference between the predicted data and the real-time data at different time points through a long-period memory network algorithm, and determining the change trend of the difference along with time; obtaining analysis results of differences between the predicted data and the real-time data, including the magnitude, direction and possible reasons of the differences; judging whether the equipment needs to be calibrated according to the analysis result, and if the difference exceeds a preset threshold value, determining that the equipment needs to be calibrated; if the calibration is determined to be needed, calling a calibration program of the equipment, wherein the program automatically adjusts equipment setting according to an analysis result so as to reduce the difference between the prediction and the real-time data; applying the determined calibration parameters to the flow measurement device; after calibration is completed, real-time flow measurement data of the equipment are acquired again, data before and after calibration are compared, and the calibration effect is evaluated; using a device performance change prediction model, and evaluating the performance of the device after calibration according to the real-time data after calibration, so as to ensure that all adjustments reach the expected effect; the performance of the flow measurement device is continuously monitored, and the device performance change prediction model is periodically used to analyze and optimize the data.

The technical scheme provided by the embodiment of the invention can have the following beneficial effects:

the invention provides an intelligent gas flow monitoring method. The accuracy and the efficiency of gas flow measurement in the industrial and scientific fields are remarkably improved. The method overcomes the measurement challenges caused by thermal expansion or cold contraction of gas under temperature change by integrating an advanced temperature prediction model and a flow calculation method, and ensures the measurement accuracy and consistency under fluctuating environmental conditions. The method can predict the volume change of the gas through real-time monitoring and data analysis, and accurately predict the flow rate change of the gas according to the volume change, so that reliable flow data is provided under changing environmental conditions. In addition, the invention obviously reduces the requirement and frequency of equipment maintenance and improves the operation efficiency through an automatic abnormality detection and calibration mechanism. The intelligent calibration period adjustment function further optimizes the long-term performance of the device and reduces the need for manual intervention. The method not only improves the safety and efficiency of the industrial process, but also is beneficial to realizing higher environment compliance, and has important significance for modern industrial automation and intelligent transformation.

Drawings

FIG. 1 is a flow chart of an intelligent gas flow monitoring method of the present invention.

Fig. 2 is a schematic diagram of an intelligent gas flow monitoring method according to the present invention.

Fig. 3 is a schematic diagram of an intelligent gas flow monitoring method according to the present invention.

Detailed Description

The technical solutions of the present invention will be clearly and completely described in connection with the embodiments, and it is obvious that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The intelligent gas flow monitoring method in this embodiment may specifically include:

s101, detecting the current temperature of the gas, and constructing a gas temperature prediction model to obtain a specific value of temperature change.

And acquiring real-time temperature data of the gas through a temperature sensor, and continuously recording the gas temperature data provided by the sensor by adopting a data acquisition system. And acquiring a time stamp of each temperature data point through the sensor data, and performing time sequence arrangement on the acquired temperature data. According to historical gas temperature data, a linear regression algorithm is adopted for model training, a gas temperature prediction model is constructed, the trend of the gas temperature data is predicted, and the maximum value, the minimum value and the average value in the temperature data are determined to obtain a specific value of temperature change. And judging whether the temperature change exceeds a preset normal operation range. If the temperature change is beyond the normal range, an abnormal report is obtained, including a temperature change value and a time stamp.

For example, a temperature sensor measures the gas temperature once per minute and continuously records data, resulting in a sensor recording a temperature change from 120 ℃ to 150 ℃ over one hour. Each temperature data point has a time stamp, e.g., the temperature of 12:00PM is 120℃and the temperature of 12:30PM is 150 ℃. The linear regression model was trained using the gas temperature data of the past week, and the model revealed that the temperature generally increased at noon, e.g., from 120 ℃ to 150 ℃. The model predicts the temperature trend for the next few hours, finding that the predicted temperature will reach 160 ℃. If the safe operating temperature range of the reactor is 100 ℃ to 150 ℃, the maximum temperature 160 ℃ predicted by the model is outside this range. The system automatically generates an exception report as the predicted temperature exceeds the normal operating range. The report indicates that the temperature would be expected to reach 160 ℃ beyond the safe range at about 1:00 pm.

S102, a linear regression algorithm is utilized to establish a gas thermal expansion or contraction prediction model, and the degree of thermal expansion or contraction of the gas is calculated based on the gas temperature change data.

And acquiring real-time temperature data through a gas temperature sensor. And carrying out data preprocessing on the real-time temperature data, including removing noise and abnormal values. And establishing a gas thermal expansion or cold shrinkage prediction model by using the processed temperature data through a linear regression algorithm. And obtaining a preliminary prediction result of thermal expansion or cold contraction of the gas according to the current temperature data. Obtaining a prediction error of each data point by calculating a difference value between a prediction value and an actual observation value, and calculating an overall error of the whole prediction result by adopting a method for calculating a mean square error; determining the accuracy of the model according to the distribution condition of errors; if the error value is smaller than the preset threshold value and the distribution is uniform, the model is considered to be accurate, and if the error value is larger than the preset threshold value or the distribution is uneven, the model is indicated to need further optimization. And if the model needs to be adjusted, parameter adjustment is carried out to improve the gas thermal expansion or cold contraction prediction model. According to the improved gas thermal expansion or contraction prediction model, obtaining an optimized gas thermal expansion or contraction prediction result;

For example, a gas temperature sensor is used to measure the real-time temperature of the gas and obtain real-time temperature data

[25,28,1,32,29]; first, noise and abnormal values are removed from the real-time temperature data, and the temperature data of the third data point is detected to be 1, and is removed as an abnormal value. The processed temperature data is [25,28,32,29]; next, a linear regression algorithm is used to build a gas thermal expansion or contraction prediction model, and a linear regression model y=43x+15 is built from the processed temperature data, where y represents the degree of thermal expansion or contraction of the gas, and x represents the temperature. Obtaining a corresponding gas thermal expansion or contraction prediction result [1090,1219,1391,1262] according to the current temperature data [25,28,32,29]; calculating the difference value between the predicted value and the actual observed value to obtain the prediction error of each data point; then, the overall error of the whole prediction result is calculated by adopting a method for calculating the mean square error. According to the distribution condition of the errors, whether the thermal expansion or contraction prediction model of the gas has the condition of overestimating or underestimating the expansion or contraction degree of the gas can be judged. If the error value is smaller than a preset threshold value and is uniformly distributed, if the mean square error is smaller than 2 and the error is distributed within +/-5, the model is considered to be accurate, and if the error value is larger than the preset threshold value or is unevenly distributed, the model is indicated to need further optimization; if the mean square error is less than 5, the error distribution is within + -5, which indicates that the model is accurate. If the model is found to need to be adjusted, parameter adjustment or algorithm adjustment can be performed to improve the gas thermal expansion or contraction prediction model. Finally, according to the improved gas thermal expansion or contraction prediction model, the optimized gas thermal expansion or contraction prediction result can be obtained.

S103, predicting the gas volume change according to real-time temperature and pressure monitoring.

And acquiring a temperature prediction result of the gas at a certain moment or in a certain time period in the future by using a gas temperature prediction model. Using ideal gas law pv=nrt, the predicted temperature value is substituted into ideal gas law, where P is the pressure, V is the volume, n is the amount of substance of the gas, R is the ideal gas constant, and T is the temperature, and the expected gas volume change is calculated. And acquiring the current pressure data of the gas measured in real time through the pressure sensor. Combining the pressure data measured in real time with the predicted temperature data to recalculate the gas volume change; an initial predicted value of the gas volume change after combining the temperature and pressure data is determined. Historical volume change data for the same type of gas under similar conditions is obtained. And comparing the current prediction result with the historical data. Based on the comparison of the historical data, the current volume change prediction is adjusted and corrected. Continuously monitoring the temperature and pressure of the gas, updating data in real time, further refining and updating the prediction of the volume change according to the latest monitoring data, and determining the final prediction of the volume change of the gas.

For example, the gas temperature prediction model predicts that the temperature of the gas will rise from the current 27 ℃ to 77 ℃ at tomorrow. The volume change is calculated using the ideal gas law: ideal gas law pv=nrtp is used. To apply this law we need to convert the temperature from degrees celsius to K, since the temperature unit in the ideal gas law is K. The conversion formula is T (K) =t (°c) +273.15. Initial temperature t1=27 ℃ =300.15K, predicted temperature t2=77 ℃ = 350.15K. Under initial conditions, the pressure p=1 atm, the quantity of substance n=1 mol, the gas constant r=0.0821 l/cdotpatm/(K/cdotpmol); the initial volume V1 can be calculated from v1=nrt1p

V1=1× 0.0821 × 300.151 ≡24.65L. The predicted volume V2 can be calculated from v2=nrt2p

V2=1× 0.0821 × 350.151 ≡28.75L. The predicted results indicate that the volume of gas will increase from about 24.65L to about 28.75L due to the temperature rise. This indicates that the gas will expand and the volume will increase.

S104, establishing a flow velocity and volume relation model according to the gas flow velocity and volume change data, and predicting the gas flow velocity change based on the gas volume change.

Historical gas flow rate and volume change data are acquired, initial volume change data are processed by using a Navier-Stokes equation, and a flow rate and volume relation model is established. The initial parameters of the relational model are set based on the historical data, and the flow rate change of the gas is predicted. The flow rate prediction result obtained by the flow rate and volume relation model is compared with the volume change data measured in real time. And if the deviation between the predicted result and the actual data is greater than a threshold value, dynamically adjusting parameters in the algorithm. And continuously monitoring the volume change and the flow rate change of the gas, and if the actually measured volume change or the difference between the flow rate and the predicted value is larger than a preset threshold value, correcting the flow rate prediction by using a preset adjustment strategy. Based on the gas flow rate data, the stability of the flow rate change predictions is evaluated using an autoregressive model. And if the flow velocity change prediction is abnormal, carrying out real-time dynamic correction on the flow velocity prediction by using a Kalman filter. And acquiring flow velocity prediction data corrected by the Kalman filter, and carrying out real-time monitoring again to verify the effectiveness of correction. And if the corrected flow rate change data still has deviation, continuously using a Navier-Stokes equation to dynamically adjust parameters, and continuously adjusting until the flow rate change prediction data is stable.

For example, a change in the flow rate of a gas in a certain gas pipe is to be predicted. According to the analysis of the historical data, initial parameters are obtained, the average value of the historical gas flow velocity is 10m/s, and the average value of the historical volume change is 5 m/3/s. First, a flow rate and volume relation model is established by using a Navier-Stokes equation, and the flow rate in the next time period is predicted, and the predicted value is 12m/s. Then, the volume change of the gas pipeline is measured in real time to obtain the actual value of 6m3/s. According to the preset threshold, if the deviation between the predicted result and the actual data is greater than the threshold value of 1m/s, parameters in the algorithm need to be dynamically adjusted. Next, the volume change and the flow rate change of the gas are continuously monitored. In the next time period, the actual measured volume change is 7m 3/s, and there is a large difference between the predicted value of 6m 3/s. And correcting the flow rate prediction according to a preset adjustment strategy to obtain a corrected predicted flow rate of 15m/s. Then, the stability of the flow rate change prediction is evaluated by using an autoregressive model according to the gas flow rate data, and the evaluation result shows that the flow rate prediction is continuous and stable, so that finer flow rate change data is obtained. However, in a certain period of time, abnormality occurs suddenly in the flow rate prediction, and stable prediction is not possible. And at the moment, the Kalman filter is used for carrying out real-time dynamic correction on the flow velocity prediction, so that corrected flow velocity prediction data is 18m/s. Then, real-time monitoring is performed again to verify the validity of the correction, and the deviation between the corrected flow rate change data and the actual measurement data is small. And if the corrected flow rate change data still has deviation, continuously using a Navier-Stokes equation to dynamically adjust parameters, and continuously adjusting until the flow rate change prediction data is stable.

The flow rate of the multiphase fluid is calculated and predicted based on the mixed fluid characteristics and the real-time data adjustments.

Establishing a multiphase fluid flow velocity prediction model according to the influence of the combination of multiphase fluid on the flow velocity

V＝k×(w _oil ×V _oil +w _water ×V _water +w _gas ×V _gas ) Wherein V represents the total flow rate, w _oil Is the volume ratio of oil in the mixture, w _water Is the volume ratio of water in the mixture, w _gas Is the volume ratio of the gas in the mixture, V _oil Is the basic flow rate of oil, V _water Is the basic flow rate of water, V _gas Is the base flow rate of the gas and k is an adjustment factor. Measuring basic flow velocity V in pure oil, pure water and pure gas environments respectively _oil ，V _water ，V _gas Flow rate data for each phase of fluid at different pressures and temperatures was recorded. The total flow velocity V of the mixed fluid is measured by mixing different proportions of oil, water, gas. A nonlinear regression analysis is used to determine the coefficient k in the multiphase fluid flow rate prediction model. In an actual operation scene, a sensor is used for monitoring the proportion w of oil, water and gas in real time _oil ，w _water ，w _gas . And according to the fluid proportion monitored in real time, applying the regulated multiphase fluid flow velocity prediction model to conduct flow velocity prediction. The value of k is dynamically adjusted to account for changes in fluid proportions based on real-time and historical data analysis. The difference between the predicted flow rate and the actual measured flow rate is compared periodically, and the accuracy and reliability of the multiphase fluid flow rate prediction model is evaluated. And further adjusting parameters in the formula according to the performance evaluation result, and optimizing the multiphase fluid flow velocity prediction model. A sophisticated monitoring system is implemented to continuously acquire data on the flow rate, proportion and environmental conditions of the multiphase fluid. And establishing a feedback mechanism, and continuously optimizing a prediction formula according to comparison of the monitoring data and the prediction result.

For example, in a pure water environment, the basic flow velocity V is measured _water 10m/s. In pure oil environment, the basic flow velocity V is measured _oil 8m/s. In a pure gas environment, the basic flow velocity V is measured _gas 6m/s. According to realityMixing different proportions of oil, water and gas in the laboratory setting to obtain total flow velocity V, when w _oil ＝5,w _water ＝3,w _gas When=2, the total flow velocity V was measured to be 12m/s. When w is _oil ＝4,w _water ＝4,w _gas At=2, a total flow V of 15m/s was measured. When w is _oil ＝2,w _water ＝3,w _gas At=5, a total flow V of 18m/s was measured. Nonlinear regression analysis was performed based on the above data to yield the formula v=k× (w _oil × _Voil +w _water ×V _water +w _gas ×V _gas ) The adjustment coefficient k of (2) is 9. In an actual operation scene, a sensor is used for monitoring the proportion w of oil, water and gas in real time _oil ,w _water ,w _gas . If the ratio of oil, water and gas monitored in real time is w _oil ＝3,w _wate r＝4,w _gas =3. And carrying out flow rate prediction according to the adjusted formula:

v=9× (3×8+4×10+3×6) =9×82=738 m/s; the difference between the predicted flow rate and the actual measured flow rate is compared periodically, and the accuracy and reliability of the multiphase fluid flow rate prediction model is evaluated. And further adjusting parameters in the formula according to the performance evaluation result, and optimizing the multiphase fluid flow velocity prediction model. A sophisticated monitoring system is implemented to continuously acquire data on the flow rate, proportion and environmental conditions of the multiphase fluid. And establishing a feedback mechanism, and continuously optimizing a prediction formula according to comparison of the monitoring data and the prediction result.

S105, calculating a difference value with the predicted flow rate according to the real-time flow measurement data.

The sensor is used for recording the fluid flow in real time, and filtering and noise elimination technology is adopted to process the acquired flow data. And obtaining a predicted value of the gas flow rate by using a flow rate and volume relation model according to the gas volume data. The difference between the flow rate monitored in real time and the flow rate predicted by the flow rate and volume relationship model is calculated. Detecting whether the difference value of the flow velocity exceeds a preset safety threshold value, and triggering an abnormality detection mechanism if the difference value exceeds the threshold value. And automatically adjusting parameters of the flow velocity and volume relation model based on the real-time flow measurement data and an analysis algorithm according to the abnormality detection result. And updating flow measurement data in real time, and keeping the real-time property and the latest state of the data. And continuously monitoring and analyzing the flow velocity difference value, and if the difference value is kept within a preset range, not adjusting the model parameters. All flow and velocity data is automatically recorded and stored.

For example, there is a sensor that records in real time the flow of gas in an industrial pipeline. The sensor records data once per minute and the raw flow data it measures may contain noise, and the data is processed to eliminate the noise using a filtering technique such as moving average. If the flow values measured five times in succession are 100,105,98,102 and 99L/min, respectively, the filtered value is the average of these five numbers, i.e., 100.8L/min. Next, a pre-established flow rate and volume relationship model is used to predict flow rate based on the volumetric change data of the gas. If the model predicts that the flow rate at a particular volume should be 105L/min. Then, the actually measured flow rate was compared with the predicted flow rate of 105L/min at 100.8L/min, and a difference of 4.2L/min was found. If the set safety threshold is 5L/min, the difference of 4.2L/min does not exceed the threshold, so that an abnormality detection mechanism is not triggered. The sensor continues to record data while the system updates the flow data in real time. If the future measured difference remains within the threshold range, no adjustment of the model parameters is required. All flow and velocity data is automatically recorded and stored for future analysis and reference.

The fluid flow rate measurement and prediction process is continuously optimized according to the real-time environment variable and the flow rate data.

According to the density, temperature and pressure of the fluid, a flow velocity prediction model is established, and the flow velocity prediction model formula is F=a×D ^b ×T ^c ×P ^d Where F denotes flow rate, D denotes fluid density, T denotes temperature, P denotes pressure, a, b, c, D are parameters determined from experimental data, and the extent of influence of these variables on flow rate is expressed. Data of D, T, P and actual flow rate F of the fluid are acquired under different environmental conditions. The parameters a, b, c, d in the formula are determined using regression analysis. D, T, P are measured in real time and the formula f=a×d is applied ^b ×T ^c ×P ^d And carrying out flow rate prediction. Recording the predicted flow rate F _pred And actually measuring the flow rate F _actual Differences between them. According to F _pred And F is equal to _actual The difference between the values of a, b, c, d are dynamically adjusted. Changes in environmental conditions and flow data are continuously monitored. The flow rate prediction model parameters are updated periodically with newly acquired data, ensuring that the model remains accurate and effective in a constantly changing environment. The performance of the predictive model is periodically assessed under extreme environmental conditions, including extreme temperatures, high or low pressures, abnormal humidity levels, chemicals or pollutants. And adjusting model parameters according to the performance evaluation result so as to cope with extreme environment conditions. And (3) compiling a comprehensive performance report, summarizing the performances of the model under different conditions, and providing decision support based on the report.

For example, monitoring the flow rate of gas J during a chemical plant process, it is desirable to be able to predict changes in gas flow rate under different temperature and pressure conditions. Under standard laboratory conditions, the flow rate F of the gas J was measured to be 15m/s and the density D was measured to be 0.09kg/m3. Determining parameters by regression analysis to obtain the result

a=1.2, b=1, c= -1, d=1a=1.2, b=1, c= -1, d=1. If the laboratory conditions are changed to 320K and 1.1atm, measuring D, T and P in real time, and applying the formula to perform flow rate prediction to obtain a prediction result F _pred 16m/s. Actually measured flow rate F _actual 15.8m/s. This difference is recorded and the parameters are fine-tuned, e.g. a is adjusted to 1.25. Environmental changes, such as temperature and pressure, are continuously monitored and the corresponding flow rates are recorded. Under extreme conditions, such as a temperature rise to 40 ℃ and a pressure rise to 1.5atm, the model performance was re-evaluated. If the performance is not in accordance with the expectations, the model parameters are further adjusted. Reports were made summarizing the behavior of the flow rate prediction model under different temperature and pressure conditions.

S106, automatically adjusting and calibrating the flow measurement device according to the abnormality detection and stability evaluation of the flow data.

Classifying the flow data according to the difference between the flow measurement data and the predicted flow rate, wherein the flow data comprises normal flow data and abnormal flow data, and judging the flow data to be abnormal flow data if the difference between the flow measurement data and the predicted flow rate exceeds a preset threshold value. If the judgment result is abnormal flow data, judging whether the abnormality is caused by errors of the flow measurement equipment or not by comparing the data of different time points or the readings of other sensors. If the abnormality is caused by the equipment error, acquiring the state and performance parameters of the equipment, and evaluating the current working condition of the equipment. Historical parameter data acquired by flow measurement equipment, including flow rate, temperature and pressure, trains a multi-layer sensor model; and continuously acquiring readings under different flows, temperatures and pressures by using the flow measurement equipment, and evaluating the stability and accuracy of the output of the flow measurement equipment by using a pre-trained multi-layer sensor model. If the output of the multi-layer sensor analysis display equipment is unstable or inaccurate, the equipment automatically enters calibration, and the internal parameters of the equipment are automatically adjusted to correct the output errors. After calibration is completed, the device automatically returns to the normal working mode, and the device continues to perform flow measurement and data output. After calibration, the output data of the equipment is again evaluated by using the equipment diagnosis program, whether the equipment is successfully calibrated is judged, if the calibration is successful, the calibration flow is ended, and if the calibration is unsuccessful, the calibration is re-performed.

For example, the predicted flow rate of the flow measurement device is 100m/s, and the actual measured flow rate data is 105m/s. According to the set threshold value of 5m/s, the difference value exceeds the threshold value, which indicates that data abnormality exists, and the data is classified into normal and abnormal types. By comparing the data at different time points, the flow data at other time points are found to be 100m/s under the same operation condition, which indicates that the abnormality may be caused by equipment errors. Historical parameter data acquired by flow measurement equipment, including flow rate, temperature and pressure, trains a multi-layer sensor model; the current operating conditions of the device may be evaluated for its status and performance parameters. If the temperature parameter of the equipment is 35 ℃, the pressure parameter is 2MPa, and the parameters are in a normal working range according to historical data analysis. The stability and accuracy of the flow measurement device was evaluated using a pre-trained multi-layer perceptron model. According to the data continuously acquired by the equipment, the stability of the output of the multi-layer sensor analysis display equipment is 98%, and the accuracy is 96%. According to the analysis result of the multi-layer sensor, the equipment output has certain instability and inaccuracy. The device automatically enters a calibration flow, and automatically adjusts internal parameters to correct the output error. After calibration is completed, the output data of the device is again evaluated using the device diagnostic program. After calibration, the stability of the data output by the equipment is improved to 99%, and the accuracy is improved to 98%. And according to the evaluation result of the equipment diagnosis program, the equipment is successfully calibrated, and the calibration flow is ended. The device is restored to the normal operation mode, and flow measurement and data output are continued. And if the evaluation result shows that the equipment is not successfully calibrated, the calibration process is carried out again.

Device performance monitoring and dynamic calibration adjustments are implemented based on environmental changes.

For example, data of flow, temperature and pressure at different temperatures, different fluid pressures and different fluid types are obtained through a sensor, and when the temperature is 25 ℃, the fluid pressure is 2bar, and the fluid type is water, the flow measured by the sensor is 10L/min, the temperature is 27 ℃, and the pressure is 2bar. According to the data acquired under each environmental condition, a decision tree algorithm is used for model training, the special data mode and abnormality of the environment are identified, the influence of temperature and pressure changes on flow measurement is determined, and the fact that the flow measurement is larger at high temperature and high pressure can be found. The method is characterized in that a deep learning model based on a multi-layer sensor is established by using historical and real-time data, data characteristics under different environmental conditions are identified, the data characteristics are adapted to various environmental conditions, and complex relations among flow, temperature and pressure under different temperatures, pressures and fluid types can be learned. And dynamically adjusting a calibration strategy according to the output of the multi-layer sensor model so as to adapt to the current environmental condition. When the output flow measurement value of the multi-layer sensor model has larger deviation from the actual measurement value, the calibration setting of the sensor can be automatically adjusted. When the multi-layer sensor model analyzes that the measured value of the sensor is larger in a high-temperature and high-pressure environment, the calibration setting can be automatically adjusted to correct the deviation. Testing the calibrated equipment under different environmental conditions, recording performance data of the equipment, comparing the performance data with the output of the prediction model to verify the calibration effect, for example, placing the calibrated equipment in an environment with the temperature of 30 ℃ and the fluid pressure of 3bar and the fluid type of oil, and comparing the performance data with the flow value output by the prediction model to verify the calibration effect. And continuously monitoring the performance of the equipment under various environmental conditions, and periodically feeding back real-time data to the multi-layer perceptron model to perform iterative optimization of the model. Performance data for the device at different temperatures, pressures and fluid types are collected daily and used to update the multi-layer perceptron model to improve accuracy of the performance predictions. And (3) establishing a comprehensive equipment performance evaluation report according to the test data in various environments. The report includes the performance of the device at different temperatures, pressures and fluid types, calibration effects, and any field where further improvements are needed. The report indicates that there is a large deviation in the flow measurement of the device in high temperature environments, and that further improvements in the calibration strategy of the sensor are needed. Based on the performance evaluation report, maintenance and software updating of the device are periodically performed according to the problems and deficiencies indicated by the report. And according to the deviation problem under the high-temperature environment indicated by the report, calibrating or replacing the sensor, and updating related parameters in the multi-layer sensor model to improve the accuracy of performance prediction.

And S107, automatically adjusting the calibration period of the flow measurement device according to the history and the real-time monitoring data.

Acquiring equipment history calibration records, flow measurement data and environmental parameters including temperature and pressure; by means of an autoregressive model, calibration frequency and performance fluctuations in historical data are analyzed, and key factors causing calibration requirement changes are identified, including changes in environmental conditions or increases and decreases in equipment usage frequency. And according to the historical and real-time data, constructing a device calibration prediction model by adopting a multi-layer sensor, and predicting future calibration requirements. And inputting historical calibration data and real-time monitoring parameters, and calibrating a prediction model by training equipment. Based on the flow data and environmental parameters of the device, a device calibration prediction model is used to determine the current device state and possible calibration requirements. And judging whether the equipment needs to be calibrated according to the output of the equipment calibration prediction model and real-time data analysis, and determining the optimal time point of the calibration. When the equipment calibration prediction model predicts that the equipment needs to be calibrated, the calibration flow is automatically triggered, and the calibration flow is ensured to meet the calibration requirement and the operation specification of the equipment. After calibration is completed, device performance data is recorded and the effect of the calibration is assessed by comparing the flow measurement accuracy before and after calibration. And according to the calibration result and the operation data of the equipment, periodically adjusting the parameters of the equipment calibration prediction model.

For example, there is a flowmeter device that requires calibration to ensure measurement accuracy. Historical calibration records of the device, flow measurement data, and environmental parameters, including temperature and pressure, are first collected. Calibration frequency and performance fluctuations in the historical data can be analyzed by an autoregressive model. The calibration frequency was found to be on average every three months over the past year, and the performance fluctuations remained within 5% before and after calibration. Next, it is necessary to identify key factors that lead to variations in calibration requirements. By analysing the historical data, it was found that the change in calibration requirements was related to a change in environmental conditions and an increase or decrease in the frequency of use of the device, which may increase as the temperature increases in summer. And constructing a device calibration prediction model by adopting the multi-layer perceptron according to the historical and real-time data. Historical calibration data and real-time monitoring parameters are input, and a model is trained to predict future calibration requirements. From the historical data, the model may predict that calibration is required in the next quarter. Flow data and related environmental parameters of the device are continuously monitored and current device status and possible calibration requirements are determined using a device calibration predictive model. According to flow data, temperature and pressure parameters of the equipment, the model can judge that the current equipment needs to be calibrated. Based on the output of the predictive model and the real-time data analysis, it is possible to determine whether the device needs calibration and to determine the optimal point in time for the calibration. The model predicts that the device needs to be calibrated in the next month, can automatically trigger the calibration flow at the optimal time point, and ensures that the calibration meets the requirements and the operation specifications. After calibration is complete, device performance data is recorded and the effect of the calibration is assessed by comparing the flow measurement accuracy before and after calibration. The measurement accuracy after calibration is improved to 3%. And periodically adjusting parameters of the prediction model according to the calibration result and the operation data of the equipment. Parameter adjustment can be performed according to data of the last year so as to improve accuracy of the model. If the flow measurement data of the device has an average value of 1000L/min in the past week, the performance fluctuation range after the last calibration is 50L/min. Meanwhile, the average temperature in the past week was 25 ℃ and the average pressure in the past week was 2bar in the environmental parameters. From historical data analysis, it was found that the calibration requirements of the device increased by 10% when the temperature was increased to 30 ℃. Based on the model predictions, the next quarter the device has a 50% probability of needing calibration. According to real-time data analysis, the current temperature is 28 ℃, the pressure is 5bar, and the model judges that the equipment needs to be calibrated. And according to the optimal time point, the calibration flow is automatically triggered, and the measurement accuracy after calibration is improved to 3%. And according to the calibration result and the equipment operation data, periodically adjusting parameters of the prediction model to improve accuracy.

S108, performing flow measurement calibration of the equipment by using the prediction data and the real-time data.

Acquiring real-time data of flow measurement equipment, including flow rate, fluid density and temperature; if the predicted data is within the preset error range, performing model training by using a cyclic neural network algorithm according to the real-time data, constructing a device performance change prediction model, and detecting the change of the device performance or identifying possible abnormality. By cycling through the analysis of the neural network algorithm, a preliminary data quality report is obtained that includes an assessment of data consistency, any abnormal patterns identified, and an overall assessment of device performance. According to the flow rate, fluid density and temperature data of the device in a certain period of time in the future and the current real-time flow rate, fluid density and temperature data obtained from the flow measurement device, the difference between the predicted data and the real-time data at different time points is compared through a long-short-period memory network algorithm, and the change trend of the difference along with time is determined. The analysis of the differences between the predicted data and the real-time data is obtained, including the magnitude, direction and possible cause of the differences. And judging whether the equipment needs to be calibrated according to the analysis result, and if the difference exceeds a preset threshold value, determining that the equipment needs to be calibrated. If it is determined that calibration is required, a calibration program of the device is invoked, which automatically adjusts the device settings based on the analysis results to reduce the difference between the predicted and real-time data. The determined calibration parameters are applied to the flow measurement device. And after the calibration is finished, acquiring real-time flow measurement data of the equipment again, comparing the data before and after the calibration, and evaluating the calibration effect. And (3) using a device performance change prediction model, and evaluating the performance of the device after calibration according to the real-time data after calibration, so as to ensure that all adjustments reach the expected effect. The performance of the flow measurement device is continuously monitored, and the device performance change prediction model is periodically used to analyze and optimize the data.

For example, acquiring real-time data of a flow measurement device, including flow rate, fluid density, temperature; if the predicted data is in the preset rangeAnd in the error range, performing model training by using a cyclic neural network algorithm according to the real-time data, constructing a device performance change prediction model, and detecting the change of the device performance or identifying possible abnormality. By cycling through the analysis of the neural network algorithm, a preliminary data quality report is obtained that includes an assessment of data consistency, any abnormal patterns identified, and an overall assessment of device performance. Real-time data is acquired from the flow measurement device, and the flow rate is: 50m/s, fluid density: 1000kg/m ³ Temperature: 25 ℃; next, these data are model trained using a recurrent neural network algorithm and a device performance change prediction model is constructed. And analyzing the obtained preliminary data quality report by the model to obtain data consistency assessment, wherein the real-time data is consistent with the previous data according to the historical data. No abnormal pattern is currently identified. According to the model analysis, the equipment performance is normal. Next, the predicted data and the real-time data are compared using a long-short term memory network algorithm to determine if there is a significant difference between the two. If the predicted data is: 51m/s, fluid density: 1001kg/m ³ Temperature: 26 ℃; the difference results obtained by the analysis may be that the flow rate was increased by 1m/s and the fluid density was increased by 1kg/m ³ The temperature was increased by 1 ℃. The real-time data has a slight increase in flow rate, fluid density, and temperature over the predicted data. An increase in ambient air temperature results in an increase in fluid density and temperature, while an increase in flow rate may be due to other factors. Based on the analysis result, it is determined that the device needs to be calibrated because the difference exceeds a predetermined threshold. Next, a calibration procedure of the device is invoked, and device settings are automatically adjusted based on the analysis results to reduce the difference between the predicted and real-time data. After calibration is completed, real-time flow measurement data of the equipment are acquired again, the data before and after calibration are compared, the calibration effect is evaluated, and the real-time data after calibration are: flow rate and speed: 55m/s, fluid density: 1005kg/m ³ Temperature: 25 ℃; by contrast evaluation, it can be determined that the calibration is good and that the difference between the predicted and real-time data is significantly reduced. Finally, using the equipment performance change prediction model according to the calibratedAnd (3) real-time data, evaluating the performance of the equipment after calibration, and ensuring that all adjustments reach the expected effect. The performance of the flow measurement device is continuously monitored, and the device performance change prediction model is periodically used to analyze and optimize the data.

The above description of the embodiments is only for helping to understand the technical solution of the present application and its core ideas; those of ordinary skill in the art will appreciate that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An intelligent gas flow monitoring method, the method comprising:

2. The method of claim 1, wherein the detecting the current temperature of the gas, constructing a gas temperature prediction model, and deriving the specific value of the temperature change, comprises:

acquiring real-time temperature data of gas through a temperature sensor, and continuously recording the gas temperature data provided by the sensor by adopting a data acquisition system; acquiring a time stamp of each temperature data point through sensor data, and performing time sequence arrangement on the acquired temperature data; according to historical gas temperature data, performing model training by adopting a linear regression algorithm, constructing a gas temperature prediction model, predicting the trend of the gas temperature data, and determining the maximum and minimum values and the average value in the temperature data to obtain a specific value of temperature change; judging whether the temperature change exceeds a preset normal operation range; if the temperature change is beyond the normal range, an abnormal report is obtained, including a temperature change value and a time stamp.

3. The method of claim 1, wherein the establishing a predictive model of thermal expansion or contraction of the gas using a linear regression algorithm, calculating the degree of thermal expansion or contraction of the gas based on the data of the change in the temperature of the gas, comprises:

4. The method of claim 1, wherein predicting gas volume changes based on real-time temperature and pressure monitoring comprises:

Acquiring a temperature prediction result of the gas at a certain moment or in a certain time period in the future by using a gas temperature prediction model; calculating an expected gas volume change using an ideal gas law pv=nrt, substituting a predicted temperature value into the ideal gas law, where P is pressure, V is volume, n is the amount of a substance of the gas, R is an ideal gas constant, and T is temperature; acquiring current pressure data of the gas measured in real time through a pressure sensor; combining the pressure data measured in real time with the predicted temperature data to recalculate the gas volume change; determining an initial predicted value of the gas volume change after combining the temperature and pressure data; acquiring historical volume change data of the same type of gas under similar conditions; comparing the current prediction result with historical data; based on the comparison of the historical data, adjusting and correcting the current volume change prediction; continuously monitoring the temperature and pressure of the gas, updating data in real time, further refining and updating the prediction of the volume change according to the latest monitoring data, and determining the final prediction of the volume change of the gas.

5. The method of claim 1, wherein the modeling flow rate and volume relationships based on the gas flow rate and volume change data, predicting gas flow rate changes based on gas volume changes, comprises:

Acquiring historical gas flow rate and volume change data, processing initial volume change data by using a Navier-Stokes equation, and establishing a flow rate and volume relation model; setting initial parameters of a relation model based on historical data, and predicting the flow velocity change of gas; comparing a flow velocity prediction result obtained through a flow velocity and volume relation model with volume change data measured in real time; if the deviation between the predicted result and the actual data is greater than a threshold value, dynamically adjusting parameters in an algorithm; continuously monitoring the volume change and the flow rate change of the gas, and if the volume change or the difference between the flow rate and the predicted value which are actually measured is larger than a preset threshold value, correcting the flow rate prediction by using a preset adjustment strategy; according to the gas flow rate data, using an autoregressive model to evaluate the stability of the flow rate change prediction; if the flow speed change prediction is abnormal, carrying out real-time dynamic correction on the flow speed prediction by using a Kalman filter; acquiring flow velocity prediction data corrected by a Kalman filter, and performing real-time monitoring again to verify the effectiveness of correction; if the corrected flow rate change data still has deviation, continuously using a Navier-Stokes equation to dynamically adjust parameters, and continuously adjusting until the flow rate change prediction data is stable; further comprises: calculating and predicting the flow rate of the multiphase fluid according to the characteristics of the mixed fluid and the real-time data adjustment;

The method for calculating and predicting the flow rate of the multiphase fluid according to the characteristics of the mixed fluid and the real-time data adjustment specifically comprises the following steps: based on the effect of the combination of multiphase fluids on the flow rate, a multiphase fluid flow rate prediction model v=k× (w _oil ×V _oil +w _water ×V _water+ w _gas ×V _gas ) Wherein V represents the total flow rate, w _oil Is the volume ratio of oil in the mixture, w _water Is the volume ratio of water in the mixture, w _gas Is the volume ratio of the gas in the mixture, V _oil Is the basic flow rate of oil, V _water Is the basic flow rate of water, V _gas Is the base flow rate of the gas, k is an adjustment factor; measuring basic flow velocity V in pure oil, pure water and pure gas environments respectively _oil ，V _water ，V _gas Recording flow velocity data of each phase of fluid at different pressures and temperatures; measuring the total flow velocity V of the mixed fluid by mixing oil, water and gas in different proportions; determining a coefficient k in the multiphase fluid flow rate prediction model using nonlinear regression analysis; in an actual operation scene, a sensor is used for monitoring the proportion w of oil, water and gas in real time _oil ，w _water ，w _gas The method comprises the steps of carrying out a first treatment on the surface of the According to the fluid proportion monitored in real time, applying the regulated multiphase fluid flow velocity prediction model to perform flow velocity prediction; dynamically adjusting the value of k according to real-time and historical data analysis to cope with the change of the fluid proportion; periodically comparing the difference between the predicted flow rate and the actual measured flow rate, and evaluating the accuracy and reliability of the multiphase fluid flow rate prediction model; further adjusting parameters in the formula according to the performance evaluation result, and optimizing a multiphase fluid flow velocity prediction model; implementing a perfect monitoring system to continuously obtain the flow rate, proportion and environmental condition data of multiphase fluid The method comprises the steps of carrying out a first treatment on the surface of the And establishing a feedback mechanism, and continuously optimizing a prediction formula according to comparison of the monitoring data and the prediction result.

6. The method of claim 1, wherein said calculating a difference from a predicted flow rate from real-time flow measurement data comprises:

recording the fluid flow in real time by using a sensor, and processing the acquired flow data by adopting a filtering and noise elimination technology; according to the gas volume data, a flow velocity and volume relation model is used for obtaining a predicted value of the gas flow velocity; calculating a difference between the flow rate monitored in real time and the flow rate predicted by the flow rate and volume relationship model; detecting whether the difference value of the flow velocity exceeds a preset safety threshold value, and triggering an abnormality detection mechanism if the difference value exceeds the threshold value; according to the abnormal detection result, based on the real-time flow measurement data and an analysis algorithm, automatically adjusting parameters of a flow velocity and volume relation model; updating flow measurement data in real time, and keeping the real-time property and the latest state of the data; continuously monitoring and analyzing the flow speed difference value, and if the difference value is kept within a preset range, not adjusting model parameters; automatically recording and storing all flow and flow rate data; further comprises: continuously optimizing a fluid flow rate measurement and prediction process according to the real-time environment variable and the flow rate data;

The measuring and predicting process for continuously optimizing the fluid flow rate according to the real-time environment variable and the flow rate data specifically comprises the following steps: according to the density, temperature and pressure of the fluid, a flow velocity prediction model is established, and the flow velocity prediction model formula is F=a×D ^b ×T ^c ×P ^d Wherein F represents flow rate, D represents fluid density, T represents temperature, P represents pressure, a, b, c, D are parameters determined from experimental data, representing the extent of influence of these variables on flow rate; acquiring data of D, T, P and actual flow rate F of the fluid under different environmental conditions; determining parameters a, b, c and d in the formula by using a regression analysis method; d, T, P are measured in real time and the formula f=a×d is applied ^b ×T ^c ×P ^d Carrying out flow rate prediction; recording the predicted flow rate F _pred And actually measuring the flow rate F _actual Differences betweenThe method comprises the steps of carrying out a first treatment on the surface of the According to F _pred And F is equal to _actual The difference between the values of a, b, c and d are dynamically adjusted; continuously monitoring changes in environmental conditions and flow data; updating flow velocity prediction model parameters by using newly acquired data regularly to ensure that the model is kept accurate and effective in a continuously changing environment; periodically evaluating the performance of the predictive model under extreme environmental conditions, including extreme temperatures, high or low pressures, abnormal humidity levels, chemicals or contaminants; adjusting model parameters according to the performance evaluation result to cope with extreme environment conditions; and (3) compiling a comprehensive performance report, summarizing the performances of the model under different conditions, and providing decision support based on the report.

7. The method of claim 1, wherein the automatically adjusting and calibrating the flow measurement device based on anomaly detection and stability assessment of the flow data comprises:

classifying the flow data according to the difference between the flow measurement data and the predicted flow rate, wherein the flow data comprises normal flow data and abnormal flow data, and judging the flow data to be the abnormal flow data if the difference between the flow measurement data and the predicted flow rate exceeds a preset threshold value; if the judgment result is abnormal flow data, judging whether the abnormality is caused by errors of the flow measurement equipment or not by comparing the data of different time points or the readings of other sensors; if the abnormality is caused by the equipment error, acquiring the state and performance parameters of the equipment, and evaluating the current working condition of the equipment; historical parameter data acquired by the flow measurement device, including flow rate, temperature and pressure, trains a multi-layer perceptron model; continuously acquiring readings under different flows, temperatures and pressures by using a flow measurement device, and evaluating the stability and accuracy of the output of the flow measurement device by using a pre-trained multi-layer sensor model; if the output of the multi-layer sensor analysis display equipment is unstable or inaccurate, the equipment automatically enters calibration, and the internal parameters of the equipment are automatically adjusted to correct the output error; after the calibration is finished, the equipment automatically returns to a normal working mode, and the equipment continues to perform flow measurement and data output; after calibration, the output data of the equipment is again evaluated by using the equipment diagnosis program, whether the equipment is successfully calibrated or not is judged, if the calibration is successful, the calibration flow is ended, and if the calibration is unsuccessful, the calibration is again carried out; further comprises: performing device performance monitoring and dynamic calibration adjustment according to environmental changes;

The device performance monitoring and dynamic calibration adjustment are implemented according to environmental changes, and specifically comprise: acquiring flow, temperature and pressure data of different temperatures, different fluid pressures and different fluid types through a sensor; according to the data acquired under each environmental condition, model training is carried out by using a decision tree algorithm, the special data mode and abnormality of the environment are identified, and the influence of temperature and pressure changes on flow measurement is determined; using history and real-time data, establishing a deep learning model based on a multi-layer sensor, identifying data characteristics under different environmental conditions and adapting to various environmental conditions; dynamically adjusting a calibration strategy according to the output of the multi-layer sensor model to adapt to the current environmental conditions; automatically adjusting the calibration setting of the device according to the analysis result of the multi-layer sensor model; testing the calibrated equipment under different environmental conditions, recording performance data of the equipment, and comparing the performance data with the output of a prediction model to verify the calibration effect; continuously monitoring the performance of the equipment under various environmental conditions, and periodically feeding back real-time data to the multi-layer perceptron model to perform iterative optimization of the model; based on the test data in various environments, a comprehensive device performance assessment report is established, wherein the device performance assessment report comprises the performance of the device under various conditions, the calibration effect and any field needing further improvement; based on the performance evaluation report, maintenance and software updating of the device are periodically performed according to the problems and deficiencies indicated by the report.

8. The method of claim 1, wherein the automatically adjusting the calibration period of the flow measurement device based on historical and real-time monitoring data comprises:

acquiring equipment history calibration records and flow measurement data, and environmental parameters including temperature and pressure; analyzing calibration frequency and performance fluctuation in historical data through an autoregressive model, and identifying key factors causing calibration requirement change, wherein the key factors comprise environmental condition change or equipment use frequency increase and decrease; according to the history and real-time data, a multi-layer sensor is adopted to construct a device calibration prediction model, and future calibration requirements are predicted; inputting historical calibration data and real-time monitoring parameters, and training a device to calibrate a prediction model; according to flow data and environmental parameters of the equipment, using an equipment calibration prediction model to determine the current equipment state and possible calibration requirements; judging whether the equipment needs to be calibrated according to the output of the equipment calibration prediction model and real-time data analysis, and determining the optimal time point of the calibration; when the equipment calibration prediction model predicts that the equipment needs to be calibrated, automatically triggering a calibration flow, and ensuring that the calibration flow meets the calibration requirement and the operation specification of the equipment; after calibration is completed, recording equipment performance data, and evaluating the calibration effect by comparing the flow measurement accuracy before and after calibration; and according to the calibration result and the operation data of the equipment, periodically adjusting the parameters of the equipment calibration prediction model.

9. The method of claim 1, wherein using the predicted data and the real-time data to perform flow measurement calibration of the device comprises:

acquiring real-time data of flow measurement equipment, including flow rate, fluid density and temperature; if the predicted data is within the preset error range, performing model training by using a cyclic neural network algorithm according to the real-time data, constructing a device performance change prediction model, and detecting the change of the device performance or identifying possible abnormality; obtaining a preliminary data quality report through the analysis result of the cyclic neural network algorithm, wherein the report comprises evaluation of data consistency, any abnormal mode identified and overall evaluation of equipment performance; according to the flow rate, fluid density and temperature data of the equipment in a certain period of time in the future and the current real-time flow rate, fluid density and temperature data obtained from the flow measurement equipment, comparing the difference between the predicted data and the real-time data at different time points through a long-period memory network algorithm, and determining the change trend of the difference along with time; obtaining analysis results of differences between the predicted data and the real-time data, including the magnitude, direction and possible reasons of the differences; judging whether the equipment needs to be calibrated according to the analysis result, and if the difference exceeds a preset threshold value, determining that the equipment needs to be calibrated; if the calibration is determined to be needed, calling a calibration program of the equipment, wherein the program automatically adjusts equipment setting according to an analysis result so as to reduce the difference between the prediction and the real-time data; applying the determined calibration parameters to the flow measurement device; after calibration is completed, real-time flow measurement data of the equipment are acquired again, data before and after calibration are compared, and the calibration effect is evaluated; using a device performance change prediction model, and evaluating the performance of the device after calibration according to the real-time data after calibration, so as to ensure that all adjustments reach the expected effect; the performance of the flow measurement device is continuously monitored, and the device performance change prediction model is periodically used to analyze and optimize the data.