Self-adaptive adjustment control system and method applied to natural gas flow calibration

Abstract

The invention provides a self-adaptive adjustment control system and method applied to natural gas flow calibration, and relates to the technical field of automatic control and adjustment systems.The method comprises the steps that real-time pipeline operation data are collected, and real-time flow state recognition is conducted to obtain real-time flow state data of a preset type; the instruction generation module is used for determining a trigger coefficient, judging whether a trigger condition is met or not, and generating a calibration trigger instruction if the trigger condition is met; the data optimization module is used for receiving a calibration trigger instruction, acquiring a current compensation parameter, generating a preliminary correction coefficient, calling reference data and determining a final optimization correction parameter set in combination with the calibration data packet; the calibration updating module is used for updating the real-time compensation algorithm, determining the deviation degree, and re-triggering the calibration process if the deviation degree exceeds a preset allowable range; and the process optimization module is used for extracting current key calibration data, calling historical calibration data, determining an optimal calibration period and optimizing the calibration process based on the optimal calibration period after the calibration process is finished. And the calibration precision and efficiency of the flowmeter are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic control and regulation systems, and particularly to an adaptive regulation control system and method applied to natural gas flow calibration. Background Art

[0002] During the use of natural gas flow meters, due to the change of gas flow patterns, traditional calibration methods usually rely on manual regular calibration, which is not only time-consuming and laborious, but also difficult to adapt to complex flow pattern changes and dynamic environments. Existing calibration methods are usually relatively fixed and lack real-time response to flow pattern changes, so they cannot effectively improve the calibration accuracy of flow meters and the stability of the system.

[0003] Currently, some existing calibration systems adopt calibration methods based on sensor data, but these systems mostly rely on static parameters and preset cycles and lack the ability to dynamically adjust compensation parameters according to real-time flow pattern data, resulting in certain errors in calibration results and affecting the long-term use effect and accuracy of natural gas flow meters.

[0004] The present invention provides an adaptive regulation control system and method applied to natural gas flow calibration. Summary of the Invention

[0005] The present invention provides an adaptive regulation control system applied to natural gas flow calibration to solve the problems in the prior art that it is impossible to adapt to flow pattern changes in real time and the calibration accuracy and efficiency are relatively low. By collecting flow pattern data in real time and combining with a dynamic optimization algorithm, it can automatically adjust compensation parameters and dynamically optimize the calibration process, thereby improving the calibration accuracy and efficiency of flow meters. During the long-term use of natural gas flow meters, it can adjust compensation parameters and calibration strategies in real time, reduce human intervention, and improve the stability and accuracy of the system.

[0006] The present invention provides an adaptive regulation control system applied to natural gas flow calibration, including: A data acquisition module: collecting real-time pipeline operation data through a multi-sensor array pre-placed on the natural gas pipeline, then extracting real-time dynamic flow pattern features, and performing real-time flow pattern recognition based on the real-time dynamic flow pattern features to obtain real-time flow pattern data of a preset type; An instruction generation module: determining a trigger coefficient, and judging whether a trigger condition is satisfied based on the real-time flow pattern data of the preset type and the trigger coefficient. If the trigger condition is satisfied, a calibration trigger instruction is generated, where the trigger coefficient includes: a flow pattern stability coefficient and an error accumulation coefficient; Data optimization module: Receives a calibration trigger instruction, obtains the current compensation parameters from a preset compensation parameter buffer, generates a preliminary correction coefficient in combination with real-time dynamic flow characteristics, packages the collected real-time pipeline operation data, real-time dynamic flow characteristics, and the preliminary correction coefficient into a calibration data packet, and calls reference data and combines the calibration data packet to determine the final optimized correction parameter set; Calibration update module: Updates the real-time compensation algorithm based on the final optimized correction parameter set. After the update, continuously collects actual flow data within a preset calibration period and compares it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, the calibration process is triggered again; Process optimization module: After the calibration process ends, extracts the current key calibration data, calls the historical calibration data, determines the optimal calibration period in combination with the current key calibration data, and optimizes the calibration process based on the optimal calibration period.

[0007] Preferably, the data acquisition module includes: Data acquisition unit: Collects real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline; Feature extraction unit: Inputs the collected real-time pipeline operation data into a preset sliding time window processor to extract real-time dynamic flow characteristics; Feature recognition unit: Inputs the real-time dynamic flow characteristics into a preset lightweight flow classifier for real-time flow state recognition to obtain real-time flow state data of a preset type.

[0008] Preferably, the flow state data of the preset type includes: steady flow state data, pulsating flow state data, and transient flow state data.

[0009] Preferably, the instruction generation module includes: Historical call unit: Calls the corresponding type of historical flow data and historical error accumulation data from a preset historical database based on the flow state data of the preset type; Stable threshold determination unit: Determines the flow state stability coefficient based on the historical flow data; Error threshold determination unit: Determines the error accumulation coefficient based on the historical error data; Data comparison unit: Compares the flow state stability coefficient with a preset reference stability threshold. At the same time, compares the error accumulation coefficient with a preset error tolerance upper limit; Process trigger unit: If the flow state stability coefficient is less than the preset reference stability threshold or the error accumulation coefficient breaks through the preset error tolerance upper limit, the calibration process is triggered again.

[0010] Preferably, the stable threshold determination unit includes: Feature analysis subunit: Extract the frequency domain features of the historical flow data in the continuous preset number of preset calibration periods before the current moment, and respectively determine the proportion of the flow state duration of the pulsating flow and the transient flow in each preset calibration period among the continuous preset number before the current moment; Index acquisition subunit: Perform exponentially weighted moving average calculations on the proportion of the flow state duration of the pulsating flow and the transient flow in each preset calibration period among the continuous preset number before the current moment to obtain the pulsating flow stability index and the transient flow stability index; Coefficient determination subunit: Determine the flow state stability coefficient based on the pulsating flow stability index and the transient flow stability index.

[0011] Preferably, the data optimization module includes: Parameter reading unit: Read the currently effective compensation parameters from the preset compensation parameter buffer; Data processing unit: Perform time series alignment processing on the currently effective compensation parameters and the subsequently collected real-time pipeline operation data; Matrix operation unit: According to the preset parameter adjustment rules, perform matrix operations on the real-time flow state features and the currently effective compensation parameters to obtain the initial correction coefficient; Data packaging unit: Package the real-time data, initial correction parameters, and real-time flow state features that have undergone time series alignment processing according to the preset standard protocol to obtain the current calibration data packet; Data matching unit: Based on the current calibration data packet, call the historical calibration database in the cloud for matching queries to obtain the historical calibration data sets corresponding to several historical reference cases; Data search unit: Construct a parameter optimization feasible region for the historical calibration data set, and search within the feasible region through the preset planning algorithm until the optimal historical calibration data set is obtained; Parameter update unit: Generate the current version number for the optimal historical calibration data set, and transmit the optimal historical calibration parameter set with the version number to the calibration update module through the encrypted channel, and immediately trigger the preset verification process until the verification result meets the preset requirements to determine the optimal historical calibration data set as the final optimized correction parameter set.

[0012] Preferably, the calibration update module includes: Calibration update unit: Update the real-time compensation algorithm based on the final optimized correction parameter set; Strategy adjustment unit: Set the basic calibration period according to the preset standard, and at the same time obtain the dynamic data of the natural gas pipeline working conditions, and adjust the sampling strategy based on the dynamic data of the natural gas pipeline working conditions; Data comparison subunit: Continuously collect the actual flow data based on the basic calibration period and the adjusted sampling strategy and compare it with the expected value; Deviation acquisition unit: Determine the instantaneous deviation, continuous deviation, and trend deviation based on the comparison data; Deviation analysis unit: Determine the deviation degree based on the instantaneous deviation, continuous deviation, and trend deviation; Process trigger unit: If the deviation degree exceeds the preset allowable range, re-trigger the calibration process.

[0013] Preferably, the process optimization module includes: Record retrieval unit: Start the historical data retrieval program and retrieve the calibration records of the most recent preset number of times from the preset distributed time-series database; Data integration unit: Construct a calibration data cube based on the retrieved calibration records; Index determination subunit: Determine several key indicators based on the feature importance of the data in the calibration data cube; Record screening unit: Determine several historical successful calibration records based on the key indicators in the calibration records of the most recent preset number of times; Cycle recommendation unit: Analyze the interval cycles of all historical successful calibration records to determine several basic recommended cycles; Data extraction unit: After the calibration process ends, extract the current key calibration data; Cycle screening unit: Determine the optimal calibration cycle from all the basic recommended cycles based on the current key calibration data; Process optimization unit: Optimize the calibration process based on the optimal calibration cycle.

[0014] An adaptive adjustment control method applied to natural gas flow calibration, including: Step 1: Collect real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline, and then extract real-time dynamic flow state characteristics. Based on the real-time dynamic flow state characteristics, perform real-time flow state recognition to obtain real-time flow state data of a preset type; Step 2: Determine the trigger coefficient, and judge whether the trigger condition is met based on the real-time flow state data of the preset type and the trigger coefficient. If it is met, generate a calibration trigger instruction, where the trigger coefficient includes: flow state stability coefficient and error accumulation coefficient; Step 3: Receive the calibration trigger instruction, obtain the current compensation parameters from the preset compensation parameter buffer, generate a preliminary correction coefficient in combination with the real-time dynamic flow state characteristics, package the collected real-time pipeline operation data, real-time dynamic flow state characteristics, and the preliminary correction coefficient into a calibration data packet, call the reference data, and determine the final optimized correction parameter set in combination with the calibration data packet; Step 4: Update the real-time compensation algorithm based on the final optimized correction parameter set. After the update, continuously collect the actual flow data within the preset calibration cycle and compare it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, re-trigger the calibration process; Step 5: After the calibration process ends, extract the current key calibration data, call the historical calibration data, determine the optimal calibration period in combination with the current key calibration data, and optimize the calibration process based on the optimal calibration period.

[0015] Compared with the prior art, the beneficial effects of the present application are as follows: By collecting the flow state data in real time and combining with the dynamic optimization algorithm, it is possible to automatically adjust the compensation parameters and dynamically optimize the calibration process, thereby improving the accuracy and efficiency of the flowmeter calibration. During the long-term use of the natural gas flowmeter, the compensation parameters and calibration strategy are adjusted in real time, reducing human intervention and improving the stability and accuracy of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 FIG. is a schematic structural diagram of an adaptive adjustment control system applied to natural gas flow calibration provided by an embodiment of the present invention.

[0018] Figure 2 FIG. is a schematic flowchart of an adaptive adjustment control method applied to natural gas flow calibration provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0020] Embodiment 1

[0021] An embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration, as Figure 1 shown, including: Data acquisition module: Collect real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline, and then extract real-time dynamic flow state characteristics. Based on the real-time dynamic flow state characteristics, perform real-time flow state recognition to obtain real-time flow state data of a preset type; Instruction generation module: Determine the triggering coefficient, and judge whether the triggering condition is met based on the real-time preset type of flow state data and the triggering coefficient. If the condition is met, generate a calibration trigger instruction, where the triggering coefficient includes: flow state stability coefficient and error accumulation coefficient; Data optimization module: Receive the calibration trigger instruction, obtain the current compensation parameters from the preset compensation parameter buffer, generate a preliminary correction coefficient in combination with the real-time dynamic flow state characteristics, package the collected real-time pipeline operation data, real-time dynamic flow state characteristics and the preliminary correction coefficient into a calibration data packet, call the reference data and determine the final optimized correction parameter set in combination with the calibration data packet; Calibration update module: Update the real-time compensation algorithm based on the final optimized correction parameter set. After the update, continuously collect the actual flow data within the preset calibration period and compare it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, trigger the calibration process again; Process optimization module: After the calibration process is completed, extract the current key calibration data, call the historical calibration data, determine the optimal calibration period in combination with the current key calibration data, and optimize the calibration process based on the optimal calibration period.

[0022] In this embodiment, the multi-sensor array includes a pressure sensor, a temperature sensor, and a basic flow sensor.

[0023] In this embodiment, the current compensation parameters are the set of correction coefficients stored after the system's last effective calibration, including flow characteristic correction factors, temperature drift compensation coefficients, and pressure influence correction parameters; In this embodiment, the final optimized correction parameter set not only includes the instantaneous correction coefficient calculated on the edge side, but also adds a stability correction factor based on long-term operation data, an equipment aging compensation coefficient, and a working condition adaptation parameter.

[0024] Advantages of the above technical solution: By collecting flow state data in real time and combining with a dynamic optimization algorithm, it is possible to automatically adjust the compensation parameters and dynamically optimize the calibration process, thereby improving the accuracy and efficiency of flowmeter calibration. During the long-term use of the natural gas flowmeter, the compensation parameters and calibration strategy are adjusted in real time, reducing human intervention and improving the stability and accuracy of the system.

[0025] Embodiment 2

[0026] The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration. The data acquisition module includes: Data acquisition unit: Collect real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline; Feature extraction unit: Input the collected real-time pipeline operation data into a preset sliding time window processor to extract real-time dynamic flow state characteristics; Feature recognition unit: Input the real-time dynamic flow state features into a preset lightweight flow state classifier for real-time flow state recognition to obtain real-time flow state data of preset types.

[0027] In this embodiment, parameter data such as flow rate, pressure, temperature, and gas composition in the pipeline are collected in real time through a multi-sensor array pre-installed on the natural gas pipeline. For example, the internal pressure of the pipeline is measured using a pressure sensor, the gas temperature is collected by a temperature sensor, and the specific composition data of natural gas is obtained through a gas composition sensor; these data will serve as the basic input for subsequent processing.

[0028] In this embodiment, the feature extraction unit includes: Input the collected real-time pipeline operation data into a preset sliding time window processor. The processor divides the data into multiple time periods according to the set time window size and extracts the dynamic flow state features within that time period. For example, according to a set 10-second sliding window, the data flow characteristics within every 10 seconds are extracted, such as the fluctuation amplitude and change rate of the flow rate; these features provide the necessary input for subsequent flow state recognition.

[0029] In this embodiment, the feature recognition unit includes: Input the real-time dynamic flow state features into a preset lightweight flow state classifier for real-time flow state recognition to obtain real-time flow state data of preset types. For example, the flow state classifier identifies flow state types such as laminar flow and turbulent flow according to different airflow characteristics, generates corresponding flow state identifiers, and outputs corresponding flow state data, providing a basis for determining the real-time flow state of the system.

[0030] Beneficial effects of the above technical solution: By collecting pipeline operation data in real time through a multi-sensor array and using a sliding time window processor and a lightweight flow state classifier to extract and recognize real-time flow state features, the flow state type can be accurately judged, enabling precise calibration of the flowmeter, improving the real-time performance and accuracy of the calibration process, dynamically adjusting compensation parameters during pipeline operation, significantly improving the measurement accuracy and calibration efficiency of the natural gas flowmeter, reducing errors, and optimizing system performance.

[0031] Embodiment 3

[0032] The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration. The preset type of flow state data includes: steady flow state data, pulsating flow state data, and transient flow state data.

[0033] In this embodiment, the flow state classification results are stored in real time in a preset flow state feature database, and the database stores historical flow state feature data using a time series data structure; In this embodiment, steady flow state data includes: for example, in a stable natural gas pipeline, the flow rate and pressure of the gas change little, the fluid presents a uniform and stable flow state, and the flow meter can continuously and stably measure the flow data with small fluctuations; for example, if the flow change remains within a set small range during the measurement process and the reading of the flow sensor is stable, the data is identified as steady flow state data.

[0034] In this embodiment, the pulsating flow state data includes: for example, the gas flow in the pipeline is affected by external factors (such as pipeline bends, pressure fluctuations, etc.), the flow velocity fluctuates periodically, and regular high and low fluctuations occur, and the flow data pulsates; for example, if the flow meter detects periodic changes in flow within a certain time interval, and this change has a certain regularity, then the data is pulsating flow state data.

[0035] In this embodiment, transient flow state data include: for example, in a natural gas pipeline, the flow rate fluctuates greatly due to certain emergencies (such as sudden opening or closing of the pipeline, airflow impact, etc.), and the flow velocity and pressure are in a state of rapid change, resulting in the flow data to fluctuate violently in an instant; for example, if the flow meter detects that the flow rate fluctuates greatly in a short period of time and the fluctuation duration is short, then this data is transient flow state data.

[0036] The beneficial effects of the above technical solution are: by presetting different types of flow state data such as steady flow, pulsating flow and transient flow, accurate flow state identification and calibration can be performed for different pipeline operating conditions. This diversified flow state classification method improves the adaptability of the system, enabling it to cope with the changing working conditions of natural gas pipelines. By identifying and classifying different flow state data, the system can adjust the compensation parameters more accurately, thereby improving the calibration accuracy and stability of the natural gas flow meter, ensuring that the flow meter can maintain efficient and accurate working performance under various working conditions, and optimizing energy monitoring and management.

[0037] Example 4

[0038] The embodiment of the present invention provides an adaptive regulation control system for natural gas flow calibration, and an instruction generation module, including: History calling unit: based on preset flow data of type, calling corresponding type of historical flow data and historical error accumulation data from preset historical database; Stability threshold determination unit: determines the flow stability coefficient based on historical flow data; Error threshold determination unit: determines the error accumulation coefficient based on historical error data; Data comparison unit: compares the flow stability coefficient with the preset reference stability threshold, and at the same time, compares the error accumulation coefficient with the preset error tolerance upper limit; Process trigger unit: If the flow state stability coefficient is less than the preset reference stability threshold or the error accumulation coefficient exceeds the preset error tolerance upper limit, the calibration process is re-triggered.

[0039] In this embodiment, the error accumulation coefficient is essentially a dynamically adjusted quality control boundary, which is implemented by using a moving average algorithm. First, the error sequence within the most recent K calibration cycles is extracted from the error accumulation database, and the mean absolute errors for the short term (the most recent 3 times) and the long term (all K times) are calculated respectively.

[0040] In this embodiment, if the flow state stability coefficient is less than the preset reference stability threshold or the error accumulation coefficient exceeds the preset error tolerance upper limit, these two indicators are compared with the preset error tolerance upper limit (stored in the compensation strategy knowledge base). When the short-term error exceeds a specific multiple of the long-term error or breaks through the tolerance upper limit, it is determined as the state of accelerated error accumulation. At this time, even if the flow state stability has not deteriorated, the system will actively generate a calibration instruction. This dual-criterion mechanism makes full use of the flow state recognition result in step 1 and the historical error accumulation data, enabling the trigger decision to consider both the real-time working conditions and the equipment performance decay trend, and realizing preventive maintenance. When any threshold condition is triggered, the system will encapsulate metadata such as the current flow state characteristic parameters, sensor calibration status, and compensation algorithm version, form a standard initialization parameter package, and transmit it to the cloud calibration system through an encrypted channel to provide a complete context environment for the calibration execution in the subsequent step 3; In this embodiment, when the flow state stability coefficient is lower than the threshold, it indicates that the pipeline operation is in a high-dynamic interference state. At this time, calibration must be triggered immediately to compensate for the measurement error caused by the flow state mutation.

[0041] Beneficial effects of the above technical solution: By calling the corresponding flow state data and error accumulation data from the historical database, it provides a rich historical reference basis for the instruction generation module. Through the determination of the stability threshold and the error threshold, the operation status of the flowmeter can be monitored in real time, and the flow state stability coefficient and the error accumulation coefficient can be compared in a timely manner to ensure that they are always within a reasonable range. When the flow state stability is insufficient or the error accumulation exceeds the standard, the calibration process can be automatically triggered, thereby effectively reducing the calibration error, improving the accuracy and stability of the natural gas flowmeter, and ensuring long-term efficient operation.

[0042] Embodiment 5

[0043] The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration. The stability threshold determination unit includes: Feature analysis sub-unit: Extract the frequency domain features of the historical flow data within the previous consecutive preset number of preset calibration cycles before the current moment, and respectively determine the proportion of the flow state duration of the pulsating flow and the transient flow within each of the previous consecutive preset number of preset calibration cycles before the current moment; Index acquisition subunit: perform exponentially weighted moving average calculations on the proportion of the flow state duration of pulsating flow and transient flow in each preset calibration period for a continuous preset number before the current moment, respectively, to obtain a pulsating flow stability index and a transient flow stability index; Coefficient determination subunit: determine a flow state stability coefficient based on the pulsating flow stability index and the transient flow stability index.

[0044] In this embodiment, to determine the proportion of the flow state duration of pulsating flow and transient flow in each preset calibration period for a continuous preset number before the current moment, specifically, in each time window, the ratio of the cumulative duration of pulsating flow / transient flow to the total window duration is statistically calculated, and then the ratio sequence of N windows is weighted and averaged.

[0045] In this embodiment, in the flow state stability threshold calculation stage, pulsating flow (Pulsating Flow) and transient flow (Transient Flow) should be statistically calculated separately. Pulsating flow is caused by periodic fluctuations (such as reciprocating pump and compressor pulsations), usually presenting high-frequency oscillations near the steady-state mean value, and the fluctuation amplitude is relatively stable. Transient flow is caused by sudden events (such as rapid valve opening and closing or pipeline breakage), manifested as a sharp flow / pressure step or decay within a short time, which is a non-periodic and dynamic interference. The error of pulsating flow mainly comes from high-frequency noise interference and can be corrected by band-pass filtering or periodic compensation. The error of transient flow stems from the lag of non-linear dynamic response and requires compensation relying on transient models. Specific statistical methods (step-by-step description): Step 1: Independently calculate the proportion of each flow state. In each time window (such as 5 minutes), respectively statistically calculate: ; Among them, is the proportion of pulsating flow in the current window, is the proportion of transient flow in the current window, is the duration of pulsating flow within the current window, is the total duration of the time window (for example, 5 minutes); is the duration of transient flow within the current window; Step 2: Dynamically weighted fusion, perform exponentially weighted moving average (EWMA) calculations on and for a continuous preset number of windows: ; ;

[0046] Among them, is the pulsating flow stability coefficient, is the transient flow stability coefficient, and the weight coefficient and dynamically adjust according to the influence degree of flow regime on the error (for example, transient flow may be given a higher weight), and are the EWMA values of the pulsating flow ratio and transient flow ratio in the previous window of the current window respectively (the initial value can be set to 0 or the historical average); Step 3: Joint trigger logic: The final flow regime stability coefficient is synthesized by weighting the two: ;

[0047] wherein, are the preset weights corresponding to the pulsating flow stability coefficient and transient flow stability coefficient respectively, determine the influence degree of the pulsating flow in the final stability coefficient and determine the influence degree of the transient flow in the final stability coefficient ; and will be dynamically adjusted to ensure that the model is more sensitive to a certain specific flow regime. For example, transient flow may have a greater impact on the stability of the system, so a higher weight can be given, while when the pulsating flow has a smaller impact on the system, a lower weight can be given. When exceeds the set threshold, calibration is triggered. For example, in the statistics of a certain natural gas transmission station: the pulsating flow ratio is stably maintained at 5% - 8% for a long time (caused by the periodic exhaust of the compressor); the transient flow ratio is usually 0.1%, but suddenly rises to 2% during a certain event (related to the action of the downstream emergency cut-off valve). In this case, the model will significantly increase the weight of the transient flow so as to preferentially trigger calibration to cope with possible step errors.

[0048] The beneficial effects of the above technical solution: By extracting the frequency domain characteristics of historical flow data, accurately identifying the duration ratios of pulsating flow and transient flow, reflecting the dynamic process of flow regime changes, using the exponential weighted moving average method to smooth the data, improving the accuracy of stability evaluation, calculating the flow regime stability coefficient based on the stability indexes of pulsating flow and transient flow, ensuring efficient and stable operation during the calibration process, reducing calibration errors, and optimizing system performance.

[0049] Example 6: The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration, and a data optimization module, including: Parameter reading unit: Read the currently effective compensation parameters from the preset compensation parameter buffer; Data processing unit: performs time series alignment processing on the currently effective compensation parameters and the real-time pipeline operation data collected later; Matrix operation unit: According to the preset parameter adjustment rules, the real-time flow characteristics and the currently effective compensation parameters are matrix operated to obtain the initial correction coefficient; Data packaging unit: packages the real-time data, initial correction parameters and real-time flow characteristics after time series alignment processing according to the preset standard protocol to obtain the current calibration data packet; Data matching unit: Based on the current calibration data packet, the historical calibration database in the cloud is called for matching query to obtain historical calibration data sets corresponding to several historical reference cases; Data search unit: constructs a feasible domain for parameter optimization of the historical calibration data set, and searches within the feasible domain through a preset planning algorithm until the optimal historical calibration data set is obtained; Parameter update unit: Generates the current version number for the optimal historical calibration data set, and transmits the optimal historical calibration parameter set carrying the version number to the calibration update module through an encrypted channel, immediately triggering the preset verification process until the verification result meets the preset requirements and determines the optimal historical calibration data set as the final optimization correction parameter set.

[0050] In this embodiment, time series alignment is performed: the sampling data of different devices such as pressure sensors, temperature probes and flow meters are timestamped and matched through preset homologous clock markers. The system uses a weighted moving average algorithm to eliminate the sampling time difference between devices to ensure that all data are in the same time base. Due to the differences in sampling frequency and delay of each sensor device in the pipeline (such as the sampling rate of the pressure sensor is 100Hz and the temperature probe is only 10Hz), the system needs to use a special timing synchronization algorithm to align multi-source data. In specific implementation, firstly, the clock deviation of all devices is controlled within 50 microseconds through the hardware clock synchronization protocol (using the PTPv2 precise time protocol); secondly, high-frequency data (such as pressure values) are downsampled, and low-frequency data (such as temperature values) are upsampled using cubic spline interpolation, and finally unified to the standard timing base of 20Hz. For the disorder of data packets caused by network jitter, the system reorders them through a sliding time window mechanism, and the window size is dynamically adjusted to 3 times the typical transmission delay (the default is 300 milliseconds). More importantly, this step detects and removes outliers caused by sensor failures, using an adaptive threshold algorithm based on moving standard deviation: five consecutive sampling points outside the 3σ range are considered abnormal. After the timing alignment is completed, the data set will be marked as "synchronized" and accompanied by metadata containing the signal-to-noise ratio indicators of each channel for weighted calculation in the subsequent optimization stage. The processing accuracy of this process directly determines the effectiveness of subsequent parameter optimization.

[0051] In this embodiment, matrix operations are performed. In the operation process, the standardized covariance matrix is used to calculate the dynamic adjustment weights of each parameter, and an initial correction vector composed of 12 dimensions is generated. The correction value of each dimension is checked against preset boundary conditions to ensure that out-of-range corrections do not occur. According to the preset multi-parameter coupling model, the real-time flow state characteristics (including 8 dynamic indicators such as pulsating flow intensity coefficient and flow velocity profile distortion degree) are calculated collaboratively with the current compensation parameters. Specifically, when executing, first construct a 24×24 standardized covariance matrix, and the matrix elements are determined by the physical correlation equations between the parameters (for example, the cross term between the temperature compensation coefficient and the pressure correction value is derived from the van der Waals equation of state). Then, use the singular value decomposition (SVD) algorithm to solve the eigenvectors of the matrix. Based on the magnitude of the eigenvectors, the system automatically assigns the dynamic adjustment weights of 12 dimensions, and 70% of the correction weights are obtained in the first 3 principal component directions (usually corresponding to the sensitive dimensions of flow state mutation, temperature gradient, and pressure fluctuation). During the calculation process, the correction amplitude of each dimension is restricted by preset physical constraint conditions (such as the single adjustment of the temperature correction coefficient does not exceed ±0.5%FS). These constraints are embedded in the optimization objective function in the form of inequalities, and the boundary conditions are forced to be satisfied by the Lagrange multiplier method. The finally generated set of initial correction coefficients includes the adjustment values of 12 dimensions and their corresponding confidence interval evaluations, and this result will be used as the input of the initial population for subsequent global optimization.

[0052] In this embodiment, it is packaged according to the preset standard protocol: First, convert the real-time data (including 12 types of original measurement values such as pressure, temperature, and flow rate) after time alignment into a fixed-length binary array according to the data frame structure defined by the ISO 31-20 standard; then add a 128-byte message header containing information such as the data generation time, device ID, and data quality flag; finally, attach the initial correction coefficient and the flow state characteristic matrix to the data body in the form of a structured array. To ensure data integrity, the system uses the SHA-256 algorithm to generate the hash digest of the entire data packet and embeds the digest value in the tail check segment of the message. The total length of the data packet is strictly controlled within 4KB, and for cases exceeding the limit, the Zstandard compression algorithm will be automatically enabled (the compression level is set to 3 to balance speed and ratio). Each successfully encapsulated data packet will obtain a unique serial number, which is jointly composed of a timestamp (42 bits), a node number (10 bits), and a packet counter (12 bits), and can achieve a globally unique identifier. Before the data packet is sent into the transmission queue, it will also undergo a memory mapping check to verify whether all pointer references and array boundaries conform to safety specifications to prevent buffer overflow attacks.

[0053] In this embodiment, the historical calibration database in the cloud is called based on the current calibration data packet for matching and querying. The matching process is divided into three steps: First, perform rough screening of working conditions. Filter out 90% of the irrelevant historical data according to static features such as pipeline diameter (classified by DN200 - DN600) and medium composition (methane percentage in the range of 85% - 99%); Then, calculate the dynamic similarity. Use the improved DTW (Dynamic Time Warping) algorithm to compare the morphological similarity between the current flow state feature curve and historical cases. The matching weight distribution is: pressure fluctuation pattern accounts for 40%, temperature gradient change accounts for 30%, and flow mutation feature accounts for 30%; Finally, perform result sorting. Retain the top 50 cases with similarity scores (the minimum similarity threshold is set to 0.72). These cases are grouped by dimensions such as working condition environment (winter / summer) and operation years (new / old pipelines) to form a reference case set for subsequent optimization. The entire process follows the principle of "data does not leave the domain". All matching calculations are completed at the edge computing node, and only the desensitized reference data index is returned. The original historical data is always stored in the security area of the central database.

[0054] In this embodiment, generate the current version number for the optimal historical calibration data set, and transmit the optimal historical calibration parameter set carrying the version number to the calibration update module through an encrypted channel. Immediately trigger the preset verification process until the verification result meets the preset requirements, and determine the optimal historical calibration data set as the final optimized correction parameter set. The selected optimal parameter set is first standardized and encapsulated by the version management system: The version number adopts the semantic encoding rule (major version. feature version. revision version + build number). For example, v2.3.1_20240615 represents the 3rd major improvement under the 2nd generation core algorithm architecture; Subsequently, it is transmitted to each level of calibration update module through an encrypted channel based on the national secret SM4 algorithm. The transmission process adopts a shard verification mechanism similar to BitTorrent to ensure the integrity and anti-interference ability of data transmission. The parameter set needs to undergo four-level verification before deployment: basic verification (parameter range check), simulation verification (digital twin system test), physical verification (standard table comparison), and on-site verification (48-hour trial operation). Each level of verification includes 27 specific test items (such as sudden load addition and subtraction test, fast start and stop test, etc.). Only the parameter set that passes all verifications (comprehensive score ≥ 95 points) will be marked as the "approved" status and officially written into the non-volatile memory of the device (the Flash memory adopts a dual-Bank alternating writing mechanism to ensure data security in case of accidental power-off). The complete optimization process (from parameter reading to final deployment) is completed within 150 seconds, meeting the real-time requirements of on-site calibration. All process data is archived in the form of a blockchain to form an immutable calibration traceability chain.

[0055] Beneficial effects of the above technical solution: By reading the currently effective compensation parameters from the compensation parameter buffer, the latest compensation information is ensured to be used during the calibration process. Through time series alignment processing of real-time data, the consistency and accuracy of the data are improved. Preliminary correction factors are generated according to preset rules, effectively reducing errors. Through the data packing unit and the data matching unit, the system can efficiently match with the cloud historical database, further optimizing the calibration results. Finally, the parameter update unit ensures the security and reliability of the calibration results by encrypting and transmitting the optimal calibration data set. This optimization process improves the accuracy, reliability, and automation level of the system.

[0056] Embodiment 7: The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration. The calibration update module includes: Calibration update unit: Update the real-time compensation algorithm based on the final optimized correction parameter set; Strategy adjustment unit: Set the basic calibration period according to the preset standard, and at the same time obtain the dynamic data of the natural gas pipeline working conditions, and adjust the sampling strategy based on the dynamic data of the natural gas pipeline working conditions; Data comparison sub-unit: Continuously collect the actual flow data based on the basic calibration period and the adjusted sampling strategy and compare it with the expected value; Deviation acquisition unit: Determine the instantaneous deviation, continuous deviation, and trend deviation based on the comparison data; Deviation analysis unit: Determine the deviation degree based on the instantaneous deviation, continuous deviation, and trend deviation; Process trigger unit: If the deviation degree exceeds the preset allowable range, re-trigger the calibration process.

[0057] In this embodiment, the final optimized correction parameter set not only includes the instantaneous correction coefficient calculated on the edge side, but also adds a stability correction factor based on long-term operation data, an equipment aging compensation coefficient, and a working condition adaptation parameter.

[0058] In this embodiment, obtaining the dynamic data of the natural gas pipeline working conditions: By deploying high-frequency sensors (such as pressure transmitters, temperature sensors, ultrasonic flow meters, etc.) at key nodes, the system can capture the change trend of the pipeline working conditions in real time and generate a dynamic data stream, including core indicators such as instantaneous flow rate, pressure gradient, and temperature distribution; In this embodiment, the sampling strategy is adjusted based on the dynamic data of the natural pipeline working conditions. A machine learning model is used to predict the possible changes in the working conditions within the next 6 hours (such as the flow rate changes caused by the throttle valve adjustment), and the sampling frequency is optimized accordingly: the basic sampling interval (such as once every 5 minutes) is maintained under steady-state working conditions, while the sampling frequency is automatically increased to the second level (such as once every 2 seconds) under transient working conditions (such as when the flow rate growth rate exceeds 5% per minute). This dynamic adjustment mechanism can ensure that the data not only meets the needs of statistical analysis but also does not affect the system performance due to over-sampling. In addition, the system can also identify special working conditions (such as pipeline pigging, regulating station maintenance, etc.) and intelligently match the preset sampling plans to avoid abnormal data interfering with the calibration results. During steady-state operation, equal-interval sampling is used (recording a complete set of data every 5 minutes), while event-triggered sampling is started under transient working conditions (increasing to 10 sets per second when the detected pressure change rate > 0.5% / s).

[0059] In this embodiment, actual flow data is continuously collected based on the basic calibration period and the adjusted sampling strategy. The data acquisition network adopts a hierarchical architecture: the on-site sensor layer transmits the original measurement values through the PROFIBUS-DP bus, the regional collector performs preliminary filtering (using a zero-phase IIR filter with a cut-off frequency set to 0.8 times the Nyquist frequency), and the central processing unit finally integrates the synchronous data sets of the entire pipe network. The key innovation lies in the introduction of a redundant measurement mechanism: two sets of sensors, a primary and a backup, are configured for each measurement point. The real-time data is processed by the Byzantine fault-tolerant algorithm, which can automatically identify and isolate faulty channels (the typical criterion is that the measurement deviation exceeds 2σ continuously for 3 times). The system specifically sets up a data quality assessment module, which conducts a percentile scoring from three dimensions: accuracy (the error compared with the standard meter), integrity (the missing rate < 0.1%), and timeliness (the transmission delay < 200 ms). Data with a score lower than 80 points will trigger automatic re-sampling. All valid data is accompanied by a complete metadata description, including 18 types of auxiliary information such as the sensor calibration certificate number, the environmental temperature compensation value, and the power quality monitoring results, forming a traceable data lineage map. The first level of data acquisition is the theoretical reference value (the theoretical flow calculated in real time based on the AGA8 state equation), the second level is the equipment reference value (the independent measurement result of the ultrasonic standard meter), and the third level is the statistical reference value (the moving average of the data in the same period in the past 30 days). The comparison engine uses the weighted difference analysis method to calculate the absolute deviation for instantaneous measurement points, the root mean square error (RMSE) for time series (such as a 15-minute segment), and conducts a Spearman correlation test on the trend term. The system establishes a deviation classification model: random deviations (Gaussian distribution) only trigger data quality alarms, while systematic deviations (continuously larger or smaller) activate the dynamic fine-tuning of compensation parameters. Each comparison generates a structured evaluation report, including the deviation type (such as a negative offset caused by insufficient temperature compensation), the impact quantification (the percentage deviation and the cumulative error amount), and the confidence score (based on the sensor health status and the flow pattern complexity); In this embodiment, the comparison with the expected value is based on the comparison engine running a multi-reference system analysis algorithm while maintaining three baseline lines: the physical model baseline (theoretical calculated value based on the AGA8 equation), the device baseline (ultrasonic standard meter reading), and the historical statistical baseline (average value of the same period in the past 30 days). The comparison process uses an adaptive window technique: the root mean square error is calculated using a 15-minute sliding window in the steady state, and the extreme value deviation is analyzed using a 3-second short window in the transient state. The system implements a three-level deviation determination: primary detection (single point exceeding the threshold), intermediate verification (exceeding the limit continuously for 3 times), and advanced confirmation (cross-checking with the device baseline). Only when all three levels of conditions are met simultaneously is it confirmed as a valid deviation. For the confirmed deviation, the system starts root cause tracing: the dominant factors are identified through Granger causality analysis (for example, temperature compensation failure is manifested as lag correlation > 0.7), and the accuracy of flow regime identification is judged using frequency domain coherence detection (when the coherence coefficient < 0.6, it is determined as a misjudgment of the flow regime). Each comparison generates a structured report, including the deviation level (classified according to the ISO 7066 standard), the influence factor weight (TOP3 factors and their contribution degrees), and repair suggestions (mapped to the preset response strategy library). This information is pushed to the decision support system, and at the same time, the whole network deviation heat map is updated.

[0060] In this embodiment, the instantaneous deviation, continuous deviation, and trend deviation are determined based on the comparison data: three core deviation indicators are quantified through multi-time scale analysis: the instantaneous deviation is defined as the difference between the latest single measurement value and the theoretical baseline, expressed as a normalized percentage (such as -1.2% FS); the continuous deviation is calculated through sliding window integration (window length 1 hour), reflecting the long-term cumulative effect of the deviation, and its statistics include the maximum value, minimum value, and standard deviation; the trend deviation is obtained through linear regression analysis, fitting the slope of the data in the past 30 minutes and calculating its significance (when the p value < 0.05, it is determined as a valid trend). The extraction process of the three types of deviations adopts anti-interference design: the original data is first denoised by wavelet, then different frequency domain components are separated through variational mode decomposition (VMD), and finally the deviation characteristics are calculated for each component respectively.

[0061] In this embodiment, the deviation degree is determined based on the instantaneous deviation, the continuous deviation and the trend deviation: the calculation of the comprehensive deviation degree adopts a nonlinear aggregation algorithm: first, the three types of deviations are standardized (divided by their respective preset thresholds), and then fused through a weighted power average function (p=2), and finally a score of 0~100 is output. The weight distribution follows the principle of working condition adaptation: the continuous deviation accounts for 50% in steady state, and the instantaneous and trend are 25% each; the weight of the instantaneous deviation in the transient state is increased to 50%, and a correction factor for the complexity of the flow state is introduced (the total score increases by 10% during turbulence). The scoring model is optimized through historical data training to ensure that 85 points correspond to the maximum deviation allowed for actual trade settlement (such as ±1.5%). The system updates the deviation degree every 5 minutes, and predicts the score changes in the next 15 minutes through trend extrapolation. When the predicted value is about to enter the dangerous range (>80 points), defensive measures (such as limiting the rate of change of the opening of the regulating valve) are initiated in advance. The comprehensive deviation is visualized in various forms: the contribution ratio of the three types of deviations is displayed in a dynamic radar chart on the SCADA interface, and the overall status is intuitively reflected in the mobile terminal with color codes (green / yellow / red), while generating a deviation evolution curve for offline analysis. The scoring results are also linked to the equipment maintenance system. When the associated deviation of a sensor is higher than the warning line for three consecutive days, a calibration work order for the device is automatically generated.

[0062] In this embodiment, if the deviation exceeds the preset allowable range, the calibration process is re-triggered: when the comprehensive deviation exceeds the preset threshold (the default is 75 points, which can be set in stages), the system starts the intelligent recalibration decision process: first determine the calibration impact domain (usually 10 times the pipe diameter range upstream and downstream of the abnormal point) through the pipeline network topology analysis, then evaluate the calibration urgency (based on the deviation growth rate and trade impact), and finally select the optimal calibration mode (fast local calibration or full-line precision calibration). The recalibration trigger adopts a dual-channel confirmation mechanism: the hardware comparator monitors key indicators in real time (such as pressure mutation >15%), and the software decision engine comprehensively calculates economy and safety. The calibration process implements closed-loop management: the compensation effect is verified immediately after each calibration. If the new parameters fail to reduce the deviation to a safe range (<60 points), the calibration level is automatically upgraded (such as switching from single-point calibration to multi-condition calibration). The system retains the historical versions of the last three calibration parameters and supports rapid rollback to any stable state. A detailed report is generated for all recalibration events, including the triggering cause (such as "continuous deviation exceeding the limit"), execution process data (test operating points and results) and effect verification conclusions. The experience of handling similar events is associated through the enterprise-level knowledge graph to continuously optimize the adaptive capabilities of the calibration strategy.

[0063] Technical effects of the above technical solution: By optimizing and correcting the parameter set to update the real-time compensation algorithm, the timeliness and accuracy of the compensation algorithm are ensured. The strategy adjustment unit adjusts the sampling strategy based on dynamic pipeline operating conditions data, improving the pertinence and efficiency of sampling. It continuously compares the actual flow data with the expected value, promptly discovers deviations, and ensures that the flowmeter always remains in the optimal working state by accurately evaluating the deviation degree. If the deviation degree exceeds the preset range, the process trigger unit automatically restarts the calibration process, ensuring the stability and efficiency of the system. This method improves the accuracy and flexibility of the calibration process and optimizes the long-term performance of the flowmeter.

[0064] Embodiment 8: The embodiment of the present invention provides an adaptive adjustment control system applied to natural gas flow calibration. The process optimization module includes: Record retrieval unit: Start the historical data retrieval program, and retrieve the calibration records of the most recent preset number of times from the preset distributed time-series database; Data integration unit: Construct a calibration data cube based on the retrieved calibration records; Index determination subunit: Determine a number of key indicators based on the feature importance of the data in the calibration data cube; Record screening unit: Determine a number of historical successful calibration records based on the key indicators in the calibration records of the most recent preset number of times; Period recommendation unit: Analyze the interval periods of all historical successful calibration records to determine a number of basic recommended periods; Data extraction unit: After the calibration process ends, extract the current key calibration data; Period screening unit: Determine the optimal calibration period from all the basic recommended periods based on the current key calibration data; Process optimization unit: Optimize the calibration process based on the optimal calibration period.

[0065] In this embodiment, retrieve the calibration records of the most recent preset number of times: Adopt a time-partition index strategy, and only scan the time period closest to the current operating conditions (such as the interval matching the season and day-night mode). During data extraction, triple verification is performed - verifying integrity (ensuring that there are no missing required fields), consistency (checking whether the timestamps are continuous), and validity (eliminating abnormal records caused by sensor failures).

[0066] In this embodiment, constructing a calibration data cube based on the retrieved calibration records is an analysis-specific structure where, after obtaining the original calibration records, the system performs a multi-dimensional data integration operation to construct the calibration data cube. This cube organizes data along three dimensions: the time axis, the device axis, and the operating condition axis, with calibration events as the basic unit. The time axis retains the complete chronological chain of the calibration process, including the trigger moment, the execution timestamps of each stage, and the total elapsed time. The device axis associates hardware information such as the serial number of the flowmeter and the sensor calibration records. The operating condition axis integrates environmental parameters such as pressure, temperature, and components, as well as their derived features. The data aggregation process uses a star model, with the calibration event fact table as the core, and multiple dimension tables are associated through foreign keys. The system automatically detects and processes naming conflicts and unit differences across data sources, and uniformly converts all numerical values to the standard measurement system. To support efficient analysis, the cube establishes a multi-level index structure, including an inverted index based on the flow regime type, a B+ tree index for the time range, etc. After construction, the system generates a data quality report, statistically calculates indicators such as the field filling rate and the compliance of the value range. When the quality score is lower than the threshold, an alarm will be triggered and suspicious data will be marked.

[0067] In this embodiment, determining a number of key indicators based on the feature importance of the data in the calibration data cube uses a two-stage feature selection method: In the first stage, a pre-trained random forest model is used to calculate the Gini importance scores of each field, and the top 20% of the features are selected into the candidate set. In the second stage, through time series correlation analysis, redundant features in the candidate set with a fluctuation coefficient lower than the threshold or with insignificant correlation with the calibration result are removed. The finally retained key indicators are divided into three categories: performance indicators (such as the deviation improvement rate, calibration time), operating condition indicators (such as the pressure fluctuation amplitude, temperature gradient), and correction indicators (such as the adjustment range of compensation parameters). The system dynamically maintains the feature importance list and periodically retrains the selection model by combining new data. For the selected key indicators, the system automatically generates a visual distribution report, showing their statistical characteristics and the comparison with the historical baseline, and marking the out-of-normal-range outliers for manual review. All key indicators and their metadata (definition formula, calculation method, reasonable range) are stored in the indicator library for standardized management.

[0068] In this embodiment, determining a number of historical successful calibration records based on the calibration records of the key indicators in the most recent preset number of times is based on the predefined successful calibration judgment rules, and valid samples are screened from the historical calibration records. The judgment rule is a multi-condition combination logic: the core condition is that the deviation improvement rate is not less than the preset threshold (such as 30%), and the auxiliary conditions include that the stable operation time after calibration meets the requirements (such as no alarm for 4 hours), the verification test pass rate meets the standard (such as more than 90% of the test items are qualified), etc. The screening process adopts a streaming processing architecture, evaluates the calibration records one by one in reverse chronological order, and performs composite condition matching through the rule engine. In order to eliminate accidental interference, the system additionally applies a sliding window consistency check, requiring the key indicators of successful calibration to be stable in adjacent time windows. The records that pass the screening are marked as "historical successful calibration", and their complete data (including original input and corrected output) are stored in a dedicated analysis pool. The system will statistically analyze the sample composition of the pool in real time to ensure that the sample distribution of each flow type and working condition interval is relatively balanced. If there are insufficient samples of a certain type of working condition, the query time range will be automatically expanded or the screening threshold will be adjusted to supplement the data.

[0069] In this embodiment, the interval period of all historical successful calibration records is analyzed, and several basic recommended periods are determined for the historical successful calibration sample set, and the system performs period feature mining and analysis. The analysis process adopts a hybrid method: first, the kernel density estimation technology is used to draw the probability distribution curve of the calibration interval time under different flow states, and the lower limit of the 90% confidence interval is taken as the conservative recommended value; secondly, a hidden Markov model is established to identify the implicit state transition law of the calibration effect and the time interval; finally, combined with the pipeline fluid mechanics characteristics, the theoretical minimum allowable interval is calculated as a constraint condition. The analysis results generate multiple basic recommended cycle schemes, each of which corresponds to a specific flow state-operating condition combination mode (such as "steady flow + low temperature" mode recommends 12 hours, "pulsating flow + high load" mode recommends 6 hours). The system will verify the statistical significance of each scheme and eliminate recommendations with insufficient sample support. All basic recommended cycles and their derivation basis are stored in the cycle knowledge base, which supports semantic retrieval and allows filtering queries by equipment model, environmental parameters and other conditions. Abnormal interval patterns detected during the analysis process (such as two consecutive calibration intervals are too short) will automatically trigger the equipment health check process.

[0070] In this embodiment, after the calibration process ends, when the current key calibration data is extracted upon completion of the current calibration task, the system immediately starts the data extraction process to capture the complete information flow of this calibration from the real-time data bus. The extraction scope covers the original snapshot at the time of calibration trigger (operating condition data sampled in milliseconds), the compensation parameter adjustment sequence (including the coefficient changes in each iteration), and multi-dimensional performance indicators in the verification stage, etc. The data acquisition module adopts a double-buffer mechanism to ensure that high-frequency sampled data is not lost, and at the same time solves the time difference problem of multi-source data through a time alignment algorithm. The core data items extracted strictly correspond to the preset list of key indicators, including mandatory fields (such as initial / final deviation values) and optional fields (such as derived features under specific operating conditions). The system performs real-time data quality verification, interpolates and repairs abnormal sampling points or marks them for exclusion. After the extraction is completed, a data summary report is generated to compare the key indicator differences between this calibration and the historical benchmark, and prominently display the significantly changed items (such as the temperature correction coefficient suddenly increasing by more than 20%). All extracted data is assigned a unified transaction ID and stored in binding with the calibration task metadata.

[0071] In this embodiment, the optimal calibration cycle is determined based on the current key calibration data in all basic recommended cycles: the key indicator feature vector of the current calibration is input into the cycle decision model, which pre-loads all basic recommended cycle schemes. The matching process is divided into three steps: first, determine the mode to which the current operating condition belongs through a flow state classifier; second, calculate the Euclidean distance between the current indicator and the representative samples of each mode, and screen the 5 nearest neighbor candidate cycles; finally, apply a multi-criteria decision-making algorithm to comprehensively consider the calibration cost, risk aversion coefficient, and operation and maintenance strategy preferences to determine the optimal solution from the candidate cycles. The decision model has a built-in conflict detection mechanism. When the recommended result deviates from the recent actual cycle by more than the preset tolerance, an expert rule review process will be started. The final cycle determination result includes the main recommended value and the elastic interval (such as "8 hours ± 1 hour"), and is accompanied by a confidence score and a list of alternative solutions. The system will record the complete reasoning path of this decision, including the excluded candidate cycles and the reasons for their elimination, for subsequent optimization reference. The determined optimal cycle takes effect immediately and is synchronously updated to the calibration task scheduler.

[0072] In this embodiment, after obtaining the optimal calibration period, the system dynamically adjusts the entire process based on the optimized calibration process: at the scheduling level, the calibration trigger time window algorithm is reconstructed, changing the fixed period to an adaptive sliding window based on the current period value; at the resource allocation level, the computing power requirements are recalculated according to the new period parameters, and the task allocation strategy of the edge nodes is dynamically adjusted; at the business rule level, the calibration compensation parameter adjustment algorithm is optimized to better adapt to the parameter drift mode in the new period. All optimization measures are verified through the A / B test framework, first run in 5% of the calibration tasks, and gradually expanded to the full volume after confirming that the effect meets the standard. The system continuously monitors the optimized key performance indicators (such as calibration success rate, resource utilization rate), and when it detects that the indicator deterioration exceeds the preset threshold, it automatically rolls back to the previous stable configuration. Each process optimization generates a version change report, which details the adjustment content, expected benefits, and potential risks. This report is incorporated into the configuration management system for unified version control. The optimized new process takes effect officially at the beginning of the next cycle.

[0073] The beneficial effects of the above technical solution: Through the collaborative action of multiple units of the process optimization module, the efficiency and accuracy of the natural gas flowmeter calibration process are significantly improved. The record retrieval unit extracts the latest calibration records from the distributed time series database, provides basic data for the data integration unit, and constructs a calibration data cube for comprehensive analysis. The key index is determined by the index determination subunit, and the historical successful calibration records are screened to ensure optimization based on successful experience. The period recommendation unit provides a basic recommended period for screening by analyzing the interval period of the calibration records. Finally, the period screening unit determines the optimal calibration period based on the current calibration data, and the process optimization unit further optimizes the calibration process to ensure that each calibration is carried out under optimal conditions. This optimization method greatly improves the automation and accuracy of the calibration process, ensuring the efficient and stable operation of the system.

[0074] Embodiment 9: An embodiment of the present invention provides an adaptive adjustment control method applied to natural gas flow calibration, as Figure 2 shown, including: Step 1: Collect real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline, then extract real-time dynamic flow state characteristics, and perform real-time flow state recognition based on the real-time dynamic flow state characteristics to obtain real-time flow state data of a preset type; Step 2: Determine the trigger coefficient, and judge whether the trigger condition is met based on the real-time flow state data of the preset type and the trigger coefficient. If it is met, a calibration trigger instruction is generated. Among them, the trigger coefficient includes: a flow state stability coefficient and an error accumulation coefficient; Step 3: Receive the calibration trigger instruction, obtain the current compensation parameters from the preset compensation parameter buffer, generate a preliminary correction coefficient in combination with the real-time dynamic flow state characteristics, package the collected real-time pipeline operation data, real-time dynamic flow state characteristics, and preliminary correction coefficient into a calibration data packet, and call the reference data and determine the final optimized correction parameter set in combination with the calibration data packet; Step 4: Update the real-time compensation algorithm based on the final optimized correction parameter set. After the update, continuously collect the actual flow data within the preset calibration period and compare it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, trigger the calibration process again; Step 5: After the calibration process ends, extract the current key calibration data, call the historical calibration data, determine the optimal calibration period in combination with the current key calibration data, and optimize the calibration process based on the optimal calibration period.

[0075] Advantages of the above technical solution: By collecting flow state data in real time and combining with the dynamic optimization algorithm, it is possible to automatically adjust the compensation parameters and dynamically optimize the calibration process, thereby improving the accuracy and efficiency of flowmeter calibration. During the long-term use of the natural gas flowmeter, the compensation parameters and calibration strategy are adjusted in real time, reducing human intervention and improving the stability and accuracy of the system.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An adaptive regulation control system applied to natural gas flow calibration, characterized in that, Including: Data acquisition module: It acquires real-time pipeline operation data through a multi-sensor array pre-placed on the natural gas pipeline, then extracts real-time dynamic flow state characteristics, and performs real-time flow state recognition based on the real-time dynamic flow state characteristics to obtain real-time flow state data of a preset type; Instruction generation module: It determines a trigger coefficient, and judges whether the trigger condition is met based on the real-time flow state data of the preset type and the trigger coefficient. If it is met, a calibration trigger instruction is generated. Among them, the trigger coefficient includes: a flow state stability coefficient and an error accumulation coefficient; Data optimization module: It receives the calibration trigger instruction, obtains the current compensation parameter from the preset compensation parameter buffer, generates a preliminary correction coefficient in combination with the real-time dynamic flow state characteristics, packs the acquired real-time pipeline operation data, real-time dynamic flow state characteristics, and the preliminary correction coefficient into a calibration data packet, and calls the reference data and combines the calibration data packet to determine the final optimized correction parameter set; Calibration update module: It updates the real-time compensation algorithm based on the final optimized correction parameter set. After the update, it continuously acquires the actual flow data within a preset calibration period and compares it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, the calibration process is triggered again; Process optimization module: After the calibration process ends, it extracts the current key calibration data, calls the historical calibration data, determines the optimal calibration period in combination with the current key calibration data, and optimizes the calibration process based on the optimal calibration period.

2. The adaptive regulation control system applied to natural gas flow calibration according to claim 1, wherein Data acquisition module, including: Data acquisition unit: It acquires real-time pipeline operation data through a multi-sensor array pre-placed on the natural gas pipeline; Feature extraction unit: It inputs the acquired real-time pipeline operation data into a preset sliding time window processor to extract real-time dynamic flow state characteristics; Feature recognition unit: It inputs the real-time dynamic flow state characteristics into a preset lightweight flow state classifier for real-time flow state recognition to obtain real-time flow state data of a preset type.

3. The adaptive adjustment control system applied to natural gas flow calibration according to claim 2, wherein Flow state data of a preset type includes: steady flow state data, pulsating flow state data, and transient flow state data.

4. An adaptive regulation control system applied to natural gas flow calibration according to claim 1, characterized in that, Instruction generation module, including: Historical call unit: It calls the historical flow data and historical error accumulation data of the corresponding type from the preset historical database based on the flow state data of the preset type; Stable threshold determination unit: It determines the flow state stability coefficient based on the historical flow data; Error threshold determination unit: It determines the error accumulation coefficient based on the historical error data; Data comparison unit: It compares the flow state stability coefficient with the preset reference stability threshold. At the same time, it compares the error accumulation coefficient with the preset error tolerance upper limit; Process trigger unit: If the flow state stability coefficient is less than the preset reference stability threshold or the error accumulation coefficient exceeds the preset error tolerance upper limit, the calibration process is triggered again.

5. An adaptive regulation control system applied to natural gas flow calibration according to claim 4, characterized in that Stable threshold determination unit, including: Feature analysis sub-unit: It performs frequency domain feature extraction on the historical flow data within a continuous preset number of preset calibration periods before the current moment, and respectively determines the proportion of the flow state duration of pulsating flow and transient flow within each of the continuous preset number of preset calibration periods before the current moment; Index acquisition subunit: Perform exponentially weighted moving average calculations on the proportion of the flow state duration of pulsating flow and transient flow in each preset calibration period for a continuous preset number of periods before the current moment, respectively, to obtain the pulsating flow stability index and the transient flow stability index; Coefficient determination subunit: Determine the flow state stability coefficient based on the pulsating flow stability index and the transient flow stability index.

6. An adaptive adjustment control system applied to natural gas flow calibration according to claim 1, characterized in that, Data optimization module, including: Parameter reading unit: Read the currently effective compensation parameters from the preset compensation parameter buffer; Data processing unit: Perform time series alignment processing on the currently effective compensation parameters and the subsequently collected real-time pipeline operation data; Matrix operation unit: Perform matrix operations on the real-time flow state characteristics and the currently effective compensation parameters according to the preset parameter adjustment rules to obtain the initial correction coefficient; Data packaging unit: Package the real-time data, initial correction parameters, and real-time flow state characteristics that have undergone time series alignment processing according to the preset standard protocol to obtain the current calibration data packet; Data matching unit: Call the historical calibration database in the cloud based on the current calibration data packet for matching queries to obtain a historical calibration data set corresponding to a number of historical reference cases; Data search unit: Construct a parameter optimization feasible region for the historical calibration data set, and search within the feasible region through a preset planning algorithm until the optimal historical calibration data set is obtained; Parameter update unit: Generate the current version number for the optimal historical calibration data set, and transmit the optimal historical calibration parameter set with the version number to the calibration update module through an encrypted channel, immediately triggering a preset verification process until the verification result meets the preset requirements, and determining the optimal historical calibration data set as the final optimized correction parameter set.

7. An adaptive regulation control system applied to natural gas flow calibration according to claim 1, characterized in that, Calibration update module, including: Calibration update unit: Update the real-time compensation algorithm based on the final optimized correction parameter set; Strategy adjustment unit: Set the basic calibration period according to the preset standard, and at the same time obtain the dynamic data of the natural gas pipeline working conditions, and adjust the sampling strategy based on the dynamic data of the natural pipeline working conditions; Data comparison subunit: Continuously collect the actual flow data based on the basic calibration period and the adjusted sampling strategy and compare it with the expected value; Deviation acquisition unit: Determine the instantaneous deviation, continuous deviation, and trend deviation based on the comparison data; Deviation analysis unit: Determine the deviation degree based on the instantaneous deviation, continuous deviation, and trend deviation; Process trigger unit: If the deviation degree exceeds the preset allowable range, re-trigger the calibration process.

8. An adaptive adjustment control system applied to natural gas flow calibration according to claim 1, wherein, Process optimization module, including: Record retrieval unit: Start the historical data retrieval program and retrieve the calibration records for the most recent preset number of times from the preset distributed time series database; Data integration unit: Construct a calibration data cube based on the retrieved calibration records; Index determination subunit: Determine a number of key indicators based on the feature importance of the data in the calibration data cube; Record screening unit: Determine a number of historical successful calibration records based on the key indicators in the calibration records for the most recent preset number of times; Cycle recommendation unit: Analyze the interval cycles of all historical successful calibration records to determine a number of basic recommended cycles; Data extraction unit: After the calibration process ends, extract the current key calibration data; Period screening unit: Determine the optimal calibration period among all basic recommended periods based on the current key calibration data; Process optimization unit: Optimize the calibration process based on the optimal calibration period.

9. An adaptive adjustment control method applied to natural gas flow calibration, characterized in that, Including: Step 1: Collect real-time pipeline operation data through a multi-sensor array pre-installed on the natural gas pipeline, and then extract real-time dynamic flow state characteristics. Based on the real-time dynamic flow state characteristics, perform real-time flow state recognition to obtain real-time flow state data of a preset type; Step 2: Determine the trigger coefficient, and judge whether the trigger condition is met based on the real-time flow state data of the preset type and the trigger coefficient. If the trigger condition is met, generate a calibration trigger instruction. Among them, the trigger coefficient includes: flow state stability coefficient and error accumulation coefficient; Step 3: Receive the calibration trigger instruction, obtain the current compensation parameter from the preset compensation parameter buffer, generate a preliminary correction coefficient in combination with the real-time dynamic flow state characteristics, package the collected real-time pipeline operation data, real-time dynamic flow state characteristics and the preliminary correction coefficient into a calibration data packet, call the reference data and determine the final optimized correction parameter set in combination with the calibration data packet; Step 4: Update the real-time compensation algorithm based on the final optimized correction parameter set. After the update, continuously collect actual flow data within the preset calibration period and compare it with the expected value to determine the deviation degree. If the deviation degree exceeds the preset allowable range, trigger the calibration process again; Step 5: After the calibration process ends, extract the current key calibration data, call the historical calibration data, determine the optimal calibration period in combination with the current key calibration data, and optimize the calibration process based on the optimal calibration period.

Images