An intelligent method and system for instrument calibration

Through the combination of differential geometry theory and neural network model, the calibration path and parameters are optimized, and the problem of data processing in the existing technology is solved, the calibration accuracy and efficiency are improved, and the adaptability and stability of the system are enhanced.

CN118797246BActive Publication Date: 2024-11-15CNOOC INSPECTION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411281191.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-13
Publication Date
2024-11-15
Estimated Expiration
2044-09-13

AI Technical Summary

Technical Problem

The prior art is difficult to optimize data processing when dealing with high variability data and environmental changes, resulting in limited calibration accuracy and efficiency, and lack of real-time feedback mechanisms, which affects the stability and reliability of calibration results.

Method used

Differential geometry theory is used to analyze pressure data, generate parameter curvature graphs, optimize calibration paths through calibration path minimum curvature search and path adjustment, and dynamic parameter adjustment is carried out in real time with neural network model.

Benefits of technology

Improves the efficiency and accuracy of the calibration process, enhances the system's adaptability and flexibility to environmental changes, and ensures the accuracy and productivity of measurement results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118797246B_ABST
    Figure CN118797246B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of intelligent calibration technology, specifically to an intelligent instrument calibration method and system, comprising the following steps: collecting initial pressure data from a pressure transmitter, performing signal denoising and data standardization processing, and obtaining standardized pressure data. In the present invention, differential geometry theory is adopted, and a mathematical model is used to accurately describe data changes, and the key area with the minimum or maximum curvature can be accurately identified, and the search and optimization of the minimum curvature path are performed at the same time, which not only reduces manual intervention, but also improves the automation level of calibration, making the path adjustment more scientific and systematic, and dynamically adjusting the control point configuration in combination with a neural network model, allowing real-time feedback and parameter adjustment, greatly enhancing the system's adaptability to environmental changes and the flexibility of the calibration process, not only improving the efficiency and accuracy of the calibration process, but also enhancing the reliability and stability of the system, ensuring the accuracy of the measurement results, and improving production efficiency and product quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent calibration, and in particular to an intelligent method and system for instrument calibration. Background Art

[0002] The field of intelligent calibration technology involves the use of advanced algorithms (such as artificial intelligence, machine learning) and automated tools to improve the calibration accuracy of various instruments and equipment. It is usually used in industries that require extremely high precision, such as high-precision pressure transmitter verification. Intelligent calibration can optimize the performance of equipment, ensure the accuracy and reliability of measurement results, thereby improving production efficiency and product quality.

[0003] Among them, intelligent methods for instrument calibration usually involve the use of advanced algorithms and computer technology to automate the calibration process, including but not limited to the application of machine learning technology, the development of automated software tools, and the optimization of calibration parameters through data analysis. The purpose is to improve the accuracy of the calibration process through intelligent solutions, reduce manual intervention, and improve the repeatability and reliability of calibration operations. It aims to develop calibration devices that can meet the requirements of strict new regulations and support local business and technical needs, and provide support for independent innovation and technological independence in the domestic metrology and calibration industry.

[0004] Current technologies are often unable to cope with high-variability data and environmental changes, especially in the context of increasing demand for high-precision calibration. Traditional calibration methods often fail to achieve optimal data processing due to the lack of support for advanced mathematical modeling and automated tools, resulting in limited calibration accuracy and efficiency. Calibration systems that lack a real-time feedback mechanism also have difficulty adjusting calibration parameters in a timely manner to respond to changes in the environment and equipment status, which directly affects the stability and reliability of the calibration results. This not only prolongs the calibration cycle, but may also lead to reduced production efficiency and increased costs due to the accumulation of calibration errors. In severe cases, it may even affect the quality and safety of the final product. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose an intelligent method and system for instrument calibration.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: an intelligent method for instrument calibration, comprising the following steps:

[0007] S1: Collect initial pressure data from the pressure transmitter, perform signal denoising and data standardization processing, and obtain standardized pressure data;

[0008] S2: Analyze the standardized pressure data through differential geometry theory, perform parameter space curvature measurement, calculate geodesic distribution, and generate a parameter curvature map;

[0009] S3: Based on the parameter curvature map, perform a calibration path minimum curvature search, capture a path with minimum curvature, and perform path adjustment to obtain an optimized calibration path;

[0010] S4: According to the optimized calibration path, the control points of the pressure transmitter are spatially reconfigured, with high-density layout in the high curvature area and low-density layout in the low curvature area, and the control points are adjusted according to the feedback of the calibration effect to obtain the adjusted control point configuration;

[0011] S5: combining the neural network model, dynamically adjusting the parameters of the adjusted control point configuration, optimizing the real-time calibration process, and adjusting the calibration parameters according to the output error to generate dynamically optimized calibration parameters;

[0012] S6: Use the dynamically optimized calibration parameters to complete the final calibration of the pressure transmitter, and perform error analysis and performance verification. As a further solution of the present invention, the standardized pressure data is specifically a pure signal with noise removed and a data value that has been normalized. The parameter curvature map includes color coding of curvature values ​​and curvature contours. The optimized calibration path includes minimizing the path length, optimizing the path curvature, and improving the path smoothness. The adjusted control point configuration includes optimizing the control point spacing, classifying the control point sensitivity, and adjusting the control point response time. The dynamically optimized calibration parameters include the pressure threshold, temperature compensation coefficient, and pressure response time adjusted by real-time feedback.

[0013] As a further solution of the present invention, the standardized pressure data is analyzed by differential geometry theory, parameter space curvature measurement is performed, and geodesic distribution is calculated to generate a parameter curvature map in the following steps:

[0014] S201: Analyze the standardized pressure data using differential geometry theory, identify local maximum and minimum values ​​by calculating the second-order derivatives of the data points, mark the high points and low points of the curvature, and generate local curvature marks;

[0015] S202: Based on the local curvature mark, apply spatial analysis technology to map the curvature change of the overall parameter space, determine the area with the maximum curvature change, and generate a curvature change map;

[0016] S203: extracting a continuous path with minimal curvature according to the curvature change graph, ensuring that each segment on the connection path achieves curvature minimization, capturing the optimal geodesic, and obtaining a parameter curvature graph.

[0017] As a further solution of the present invention, based on the parameter curvature map, the minimum curvature search of the calibration path is performed, the path with the minimum curvature is captured, and the path is adjusted. The steps of obtaining the optimized calibration path are as follows:

[0018] S301: Based on the parameter curvature graph, identifying the local minimum point of curvature in the entire graph, marking all potential minimum curvature regions, and obtaining a potential minimum curvature point set;

[0019] S302: According to the potential minimum curvature point set, a curvature comparison algorithm is applied to compare the curvature values ​​of adjacent data points point by point, minimizing the curvature difference between each point and adjacent points, connecting to form a continuous path, and obtaining a preliminary minimum curvature path;

[0020] S303: Based on the preliminary minimum curvature path, perform path adjustment, adjust the position of each point on the path through a local optimization adjustment algorithm, and generate an optimized calibration path.

[0021] As a further solution of the present invention, the local optimization adjustment algorithm is according to the formula:

[0022]

[0023] Calculate the new position of each point on the path ,in, is the position of the current point on the path, is the learning rate, is the curvature function About Dot The gradient of is the adjustment factor, Yes The cosine value of the current direction and the optimized direction, is a distance-based attenuation factor, Yes The distance to the starting point of the path, So The square of is an exponential decay function.

[0024] As a further solution of the present invention, according to the optimized calibration path, the control points of the pressure transmitter are spatially reconfigured, high-density layout is performed in the high curvature area, low-density layout is performed in the low curvature area, and the control points are adjusted according to the feedback of the calibration effect. The steps of obtaining the adjusted control point configuration are:

[0025] S401: Based on the optimized calibration path, using curvature analysis technology, scanning the entire calibration path, identifying and marking high curvature areas and low curvature areas, accurately dividing each area, and generating an area marking map;

[0026] S402: deploying dense control points in the high curvature area indicated by the regional marking map, reducing the density of control points in the low curvature area, optimizing the layout of each control point, and obtaining a preliminary control point layout plan;

[0027] S403: Based on the preliminary control point layout plan, calibration data is automatically collected and analyzed, and according to the analysis results, the position and density parameters of the control points are adjusted, and each control point is adjusted through loop feedback to optimize the calibration performance and obtain the adjusted control point configuration.

[0028] As a further solution of the present invention, in combination with a neural network model, the adjusted control point configuration is dynamically adjusted to optimize the real-time calibration process, and the calibration parameters are adjusted according to the output error. The step of generating dynamically optimized calibration parameters is as follows:

[0029] S501: Based on the adjusted control point configuration, in combination with the neural network model, configure the input layer of the neural network to receive the control point data and the real-time pressure reading, set the input parameters and start the real-time capture of the data stream, and generate the configuration data stream;

[0030] S502: According to the configuration data stream, activate the real-time data analysis function of the neural network, automatically process the input data, identify the control points with large deviations, use the gradient descent method to calculate the required parameter adjustment values, and generate a preliminary parameter adjustment plan; S503: Based on the preliminary parameter adjustment plan, refine the parameter settings of all control points one by one, optimize the position and response threshold of each point, monitor the calibration data during the adjustment process in real time, analyze the adjustment effect, and generate dynamically optimized calibration parameters.

[0031] As a further solution of the present invention, the gradient descent method is according to the formula:

[0032]

[0033]

[0034] Calculate parameter adjustment values ,in, is the current parameter value, is the learning rate, is the loss function About parameters The gradient of is the regularization coefficient, is the sample variance, is the gradient sign adjustment coefficient, is the sign of the gradient, is the square root of the scale of the gradient.

[0035] An intelligent instrument calibration system, the system comprising:

[0036] The data acquisition and processing module collects and records the initial pressure readings from the pressure transmitter, removes the noise exceeding the threshold in the low-pass filter, and normalizes the data to obtain standardized pressure data;

[0037] The curvature analysis module uses differential geometry theory to analyze the standardized pressure data, mark the high points and low points of the curvature, determine the area with the maximum curvature change, extract the continuous path with the minimum curvature, and obtain a parameter curvature map;

[0038] The path optimization module identifies the local minimum point of curvature in the entire graph according to the parameter curvature graph, applies a curvature comparison algorithm to compare the curvature values ​​of adjacent data points point by point, connects them to form a continuous path, adjusts the position of each point on the path, and generates an optimized calibration path;

[0039] The layout optimization module identifies and marks high curvature areas and low curvature areas based on the optimized calibration path, deploys dense control points in the high curvature areas, reduces the density of control points in the low curvature areas, optimizes the layout of each control point, and obtains a preliminary control point layout plan;

[0040] The configuration data acquisition module adjusts the position and density parameters of the control points based on the preliminary control point layout plan, combines the neural network model, configures the input layer of the neural network to receive the control point data and real-time pressure readings, sets input parameters and starts real-time capture of data streams, and generates a configuration data stream;

[0041] The calibration parameter optimization module uses a neural network to perform real-time data analysis based on the configuration data stream, identifies control points with large deviations, automatically calculates the required parameter adjustment values, optimizes the position and response threshold of each point, analyzes the adjustment effect, and generates dynamically optimized calibration parameters;

[0042] The deviation analysis and data collation module adjusts the response characteristics and thresholds of the control points based on the dynamically optimized calibration parameters, calculates the deviation between the actual output and the expected output, adjusts the calibration parameters to reduce the deviation, monitors the adjustment process and collects performance data for data collation to obtain the final calibration record.

[0043] Compared with the prior art, the advantages and positive effects of the present invention are:

[0044] In the present invention, differential geometry theory is adopted and mathematical models are used to accurately describe data changes. The key areas with the minimum or maximum curvature can be accurately identified, and the search and optimization of the minimum curvature path are performed at the same time, which not only reduces manual intervention but also improves the automation level of calibration, making the path adjustment more scientific and systematic. The control point configuration is dynamically adjusted in combination with a neural network model, allowing real-time feedback and parameter adjustment, which greatly enhances the system's adaptability to environmental changes and the flexibility of the calibration process, which not only improves the efficiency and accuracy of the calibration process, but also enhances the reliability and stability of the system, ensures the accuracy of the measurement results, and thus improves production efficiency and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0046] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0047] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0048] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0049] Figure 5 This is a detailed flow chart of S4 of the present invention;

[0050] Figure 6 This is a detailed flow chart of S5 of the present invention;

[0051] Figure 7 This is a detailed flow chart of S6 of the present invention;

[0052] Figure 8 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0054] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0055] Embodiment 1

[0056] See also Figure 1 The present invention provides a technical solution: an intelligent method for instrument calibration, comprising the following steps:

[0057] S1: Collect initial pressure data from the pressure transmitter, perform signal denoising and data standardization processing, and obtain standardized pressure data;

[0058] S2: Analyze the normalized pressure data through differential geometry theory, perform parameter space curvature measurement, and calculate the geodesic distribution to generate the parameter curvature map;

[0059] S3: Based on the parameter curvature graph, perform a calibration path minimum curvature search, capture the minimum curvature path, and adjust the path to obtain the optimized calibration path; S4: According to the optimized calibration path, spatially reconfigure the control points of the pressure transmitter, perform high-density layout in high-curvature areas, and perform low-density layout in low-curvature areas, and adjust the control points according to the calibration effect feedback to obtain the adjusted control point configuration;

[0060] S5: Combined with the neural network model, dynamically adjust the parameters of the adjusted control point configuration, optimize the real-time calibration process, and adjust the calibration parameters according to the output error to generate dynamically optimized calibration parameters;

[0061] S6: Complete the final calibration of the pressure transmitter using dynamically optimized calibration parameters, and perform error analysis and performance verification.

[0062] The standardized pressure data is specifically a pure signal with noise removed and a normalized data value. The parameter curvature map includes color coding of the curvature value and curvature contour lines. The optimized calibration path includes minimization of path length, optimization of path curvature and improvement of path smoothness. The adjusted control point configuration includes optimization of control point spacing, classification of control point sensitivity and adjustment of control point response time. The dynamically optimized calibration parameters include pressure threshold, temperature compensation coefficient and pressure response time adjusted with real-time feedback.

[0063] See also Figure 2 , the specific steps of S1 are:

[0064] S101: Collecting initial pressure readings from the pressure transmitter, recording the pressure value at the measurement time point, and generating an original pressure record;

[0065] First, the data acquisition system is initialized to ensure that the pressure sensor is calibrated and functioning properly. The continuous signal from the pressure transmitter is collected by the acquisition unit and its timestamp is associated with the pressure value and stored as a data record. The recorded data format is {timestamp, pressure value} and the data is saved as a CSV file for subsequent processing. During the acquisition process, each data point is monitored in real time to detect any potential abnormal readings or equipment failures.

[0066] S102: Based on the original pressure record, the set frequency threshold is applied in the low-pass filter to process the signal, and the noise exceeding the threshold is eliminated to obtain the denoised pressure data; The collected original pressure data is sent to the pre-set low-pass filter, which is configured with a specific frequency threshold and only allows frequency components below the threshold to pass. In this process, each data point is compared with its adjacent data points and processed by the filter operator. The calculation formula is ,in, Represents the pressure value after filtering, is the original data point, is the length of the filter. After each calculation, the filtered data set is updated to provide a more accurate data basis for the next step.

[0067] S103: Based on the denoised pressure data, convert the data into a standard format, normalize the data, ensure that the data is within a predetermined range, and obtain standardized pressure data;

[0068] After the denoising process is completed, the data format is converted to a standardized form, including unifying the timestamp format and converting the units of the pressure values. Normalization is performed to facilitate data comparison. The normalization formula is: ,in, is the normalized pressure value, is the denoised pressure data, and are the minimum and maximum pressure values ​​in the data set, respectively. In this way, all pressure values ​​will be converted to the range of 0 to 100 for subsequent data processing and analysis.

[0069] See also Figure 3 , the specific steps of S2 are:

[0070] S201: using differential geometry theory to analyze the standardized pressure data, identifying local maxima and minima by calculating the second-order derivatives of the data points, marking the high and low points of the curvature, and generating local curvature marks;

[0071] First, each data point is extracted from the standardized pressure data set, and its second-order derivative is calculated to evaluate the curvature change. The specific calculation formula is: ,in, Representing time point The pressure value, is the measurement interval. Through this formula, local extreme points in data fluctuations can be detected, indicating the severity of pressure changes. For each local maximum or minimum value, its timestamp and pressure value are marked and recorded to form high and low point marks of local curvature, which are used for further spatial analysis.

[0072] S202: Based on the local curvature markers, apply spatial analysis technology to map the curvature changes of the overall parameter space, determine the area with the largest curvature change, and generate a curvature change map; merge the local curvature markers with the entire data set, and use spatial analysis techniques (such as Kriging interpolation) to integrate the local marker points through spatial continuity. Through spatial interpolation of the entire data set, a detailed curvature change map can be constructed, showing the spatial distribution of curvature changes during the entire measurement cycle, with special emphasis on those areas with the largest curvature changes. The visual display of the curvature change map can help technicians intuitively understand the fluctuation characteristics of the pressure data and provide a basis for subsequent data analysis.

[0073] S203: extracting a continuous path with minimal curvature according to the curvature change graph, ensuring that each segment on the connection path achieves curvature minimization, capturing the optimal geodesic, and obtaining a parameter curvature graph;

[0074] Using the shortest path algorithm, such as the Dijkstra algorithm, a continuous path with the minimum curvature is found from the curvature change graph. This path represents the smoothest transition during the pressure change and can be regarded as the optimal geodesic. During the calculation process, each node is selected based on the minimum cumulative curvature from its starting point. Formulated as follows: If the node is defined To Node The path curvature is , then the path selection satisfies: in, is the number of nodes in the path. In this way, the curvature of the selected path is minimized in the entire data set, and the parameter curvature map finally generated clearly indicates the optimal path of the data, providing an accurate basis for further analysis and decision-making.

[0075] See also Figure 4 , the specific steps of S3 are:

[0076] S301: Based on the parameter curvature graph, identify the local minimum point of curvature in the entire graph, mark all potential minimum curvature regions, and obtain a potential minimum curvature point set;

[0077] First, the parameter curvature map is accurately scanned, and a search algorithm is used to identify all local minimum points. This process involves calculating the curvature in the neighborhood around each data point and comparing it with the curvature of the center point. If the curvature of the center point is lower than that of the surrounding points, it is considered a local minimum. Each local minimum point will record its position and curvature value to form a potential minimum curvature point set, which reflects the potential stable area in the pressure change and provides key data support for further analysis. S302: Based on the potential minimum curvature point set, the curvature comparison algorithm is applied to compare the curvature values ​​of adjacent data points point by point, minimize the curvature difference between each point and the adjacent points, connect to form a continuous path, and obtain a preliminary minimum curvature path;

[0078] Using an iterative algorithm, starting from the potential minimum curvature point set, the curvature values ​​of adjacent points are compared point by point. Specifically, by defining a curvature difference function in and Points and Point The goal is to find the connection with the smallest curvature difference among all adjacent point pairs, and gradually build a continuous path. This process is iterated until the path is as smooth as possible, that is, the curvature difference between each point and its adjacent points is minimized.

[0079] S303: Based on the preliminary minimum curvature path, the path is adjusted, and the position of each point on the path is adjusted by a local optimization adjustment algorithm to generate an optimized calibration path;

[0080] Local optimization adjustment algorithm, according to the formula:

[0081]

[0082] Calculate the new position of each point on the path ,in, is the position of the current point on the path, is the learning rate, controlling the adjustment step size, is the curvature function About Dot The gradient directly affects the moving direction and distance of the point. is the adjustment factor, depending on the point Direction, Yes The cosine value of the current direction and the optimized direction enhances the directional continuity of the path. is a distance-based attenuation factor, Yes The distance to the starting point of the path, So An exponential decay function with the square of as the exponent is used to adjust the movement strength of points far away from the starting point.

[0083] The execution process is as follows:

[0084] First, calculate the current point The curvature gradient , to determine the initial direction and step size of optimization, and adjust the step size by the learning rate control, and then through the cosine adjustment factor Introduce directional continuity between path points to maintain the overall smoothness of the path, distance attenuation factor The distance between the path point and the starting point is taken into account to adjust the influence of the distant point and reduce the adjustment intensity of the area far away from the starting point. Finally, the position of the point is updated based on the factors. , generate the optimized calibration path. Weight coefficient The exact value of is determined by testing the performance of different parameter configurations on the validation set, and methods such as cross-validation are used to ensure that the best weight configuration is found.

[0085] See also Figure 5 , the specific steps of S4 are:

[0086] S401: Based on the optimized calibration path, the curvature analysis technology is used to scan the entire calibration path, identify and mark high curvature areas and low curvature areas, accurately divide each area, and generate an area marking map;

[0087] First, the optimized path is continuously calculated for curvature. The specific operation includes applying the curvature calculation formula to each path segment. in, and Represents the coordinates on the path. This formula can be used to quantitatively analyze the curvature strength of the entire path. According to the curvature value, the path is divided into high curvature areas and low curvature areas. Each area will be displayed on the area marking map with different colors or marks to ensure that the characteristics and boundaries of the area can be clearly identified.

[0088] S402: deploying dense control points in the high curvature area indicated by the regional marking map, reducing the density of control points in the low curvature area, optimizing the layout of each control point, and obtaining a preliminary control point layout plan;

[0089] According to the indication of the regional marker map, high curvature areas require more dense control points to carefully monitor possible changes, while low curvature areas can use fewer control points to reduce resource consumption. and , respectively represent the control point density of high curvature and low curvature areas, and the formula is as follows ,in, and are the lengths of the high curvature and low curvature regions, respectively. Through this strategy, resources are effectively allocated to improve monitoring efficiency and generate a preliminary control point layout plan.

[0090] S403: Based on the preliminary control point layout plan, automatically collect and analyze calibration data, adjust the position and density parameters of the control points according to the analysis results, adjust each control point through loop feedback, optimize the calibration performance, and obtain the adjusted control point configuration;

[0091] After the layout plan is implemented, automatic data collection is performed to monitor the data collected at each control point. By analyzing the data, the effectiveness of the control point layout is evaluated, such as the continuity and accuracy of data collection. Based on the analysis results, the position and density of the control points are dynamically adjusted to ensure that the control point coverage area is maximized and the data quality is optimized. The adjustment formula in this process is ,in, is the current location, is the curvature difference to be improved, This method is used to continuously optimize the configuration of control points until the best calibration performance is achieved, thus forming the final control point configuration.

[0092] See also Figure 6 , the specific steps of S5 are:

[0093] S501: Based on the adjusted control point configuration, in combination with the neural network model, configure the input layer of the neural network to receive the control point data and the real-time pressure reading, set the input parameters and start the real-time capture of the data stream, and generate the configuration data stream;

[0094] First, the structure of the neural network is established, including the design of the input layer to receive data from the control points and real-time pressure readings. The input layer receives the pressure data transmitted by each control point through multiple nodes, and uses the timestamp and position information as part of the network input. Define the input parameters for the neural network, such as the data sampling frequency, the normalized processing range of the input data, and the initialization weights. Enabling real-time capture of the data stream involves turning on a real-time data listener, which continuously monitors the data input and ensures that the data flows into the neural network for real-time processing. In this way, a configured data stream is obtained for further analysis by the network.

[0095] S502: according to the configuration data stream, activate the real-time data analysis function of the neural network, automatically process the input data, identify the control points with large deviations, calculate the required parameter adjustment values ​​using the gradient descent method, and generate a preliminary parameter adjustment plan;

[0096] Gradient descent method, according to the formula:

[0097]

[0098] Calculate parameter adjustment values ,in, is the current parameter value, is the learning rate, is the loss function About parameters The gradient of is the regularization coefficient, is the sample variance, is the gradient sign adjustment coefficient, is the sign of the gradient, is the square root of the scale of the gradient.

[0099] The execution process is as follows:

[0100] First, the parameters represents the current network weight, Controls the learning step size, is the sensitivity of the current loss to the weight, reflecting the adjustment direction and degree of each weight. Multiplying by the sample variance provides a penalty for outliers and enhances the generalization ability of the model. Adjust the gradient update to ensure stability at different gradient scales by determining the best performance on the validation set. and values, perform multiple experiments to optimize these values, and finally update the parameters with the adjusted gradient values , and obtain a preliminary parameter adjustment plan.

[0101] S503: Based on the preliminary parameter adjustment plan, the parameter settings of all control points are refined one by one, the position and response threshold of each point are optimized, the calibration data in the adjustment process is monitored in real time, the adjustment effect is analyzed, and the dynamically optimized calibration parameters are generated;

[0102] The parameter settings for each control point will be refined one by one based on the preliminary parameter adjustment plan. This includes optimizing the position and response threshold of each control point to ensure that each point can most accurately reflect the pressure changes. The real-time monitoring tool will continue to collect data during the adjustment process to evaluate the impact of each adjustment on the overall system performance. Through data analysis, the adjustment strategy is continuously optimized to ensure that the calibration parameters of each control point are in the optimal state. Ultimately, this series of operations will generate a set of dynamically optimized calibration parameters to improve the response efficiency and accuracy of the entire system.

[0103] See also Figure 7 , the specific steps of S6 are:

[0104] S601: Based on the dynamically optimized calibration parameters, the response characteristics and thresholds of each control point of the pressure transmitter are adjusted to ensure that all control points operate according to the latest operating parameters and generate a state record of the adjusted control points;

[0105] First, the technician inputs the dynamically optimized calibration parameters into the control system of the pressure transmitter. The parameters include but are not limited to the response threshold, sensitivity, and reaction time. The adjustment of each parameter is aimed at optimizing the performance of the control point so that it can more accurately reflect the actual pressure changes. Subsequently, the system will automatically adjust the settings of each control point, monitor the effects of these changes in real time, and record the adjusted status of each control point, including the adjustment time, the adjusted parameter value, and the performance data before and after the adjustment. This record will serve as the basic data for verifying the effect of the new parameter settings.

[0106] S602: Analyze the state records of the adjusted control points, calculate the deviation between the actual output and the expected output, quantify the error using statistical tools, evaluate the performance of each control point, and obtain the error analysis results;

[0107] Through data analysis software, the adjusted status records of each control point are deeply analyzed. Statistical tools such as mean square error (MSE) and standard deviation are used to quantify the error between the actual output and the expected output of each control point. The calculation formula may be: ,in, and Respectively represent The actual and expected outputs of each control point can be determined in this way, and the performance level and accuracy of each control point can be determined, resulting in a comprehensive error analysis.

[0108] S603: Based on the error analysis results, adjust the calibration parameters to reduce the deviation, continuously monitor the adjustment process and collect performance data, and organize the data to obtain the final calibration record;

[0109] Based on the error analysis results, the calibration parameters are further fine-tuned to reduce the deviation. Minor parameter adjustments are implemented during the optimization process to ensure that each adjustment is accurate. The adjustment process is continuously monitored, and the system automatically records the parameter value, adjustment time, and performance data after each adjustment. At the same time, the data is collected and organized by the system to form the final calibration record. This record reflects in detail all changes from the initial state to the final state, including the history of all adjustments and the specific impact on system performance. Through continuous monitoring and recording, it can be ensured that each control point can operate under optimal conditions, greatly improving the accuracy and reliability of the entire calibration process.

[0110] See also Figure 8 ,An instrument calibration intelligent system includes: a data acquisition and processing module collects and records the initial pressure readings from the pressure transmitter, removes the noise exceeding the threshold in a low-pass filter, and normalizes the data to obtain the standardized pressure data;

[0111] The curvature analysis module uses differential geometry theory to analyze the standardized pressure data, mark the high and low points of the curvature, determine the area with the maximum curvature change, extract the continuous path with the minimum curvature, and obtain the parametric curvature map;

[0112] The path optimization module identifies the local minimum point of curvature in the entire graph based on the parameter curvature graph, applies the curvature comparison algorithm to compare the curvature values ​​of adjacent data points point by point, connects them to form a continuous path, adjusts the position of each point on the path, and generates an optimized calibration path;

[0113] The layout optimization module identifies and marks high curvature areas and low curvature areas based on the optimized calibration path, deploys dense control points in high curvature areas, reduces the density of control points in low curvature areas, optimizes the layout of each control point, and obtains a preliminary control point layout plan;

[0114] The configuration data acquisition module adjusts the position and density parameters of the control points based on the preliminary control point layout plan, combines the neural network model, configures the input layer of the neural network to receive the control point data and real-time pressure readings, sets the input parameters and starts the real-time capture of the data stream, and generates the configuration data stream;

[0115] The calibration parameter optimization module uses neural networks to perform real-time data analysis based on the configuration data stream, identify control points with large deviations, automatically calculate the required parameter adjustment values, optimize the position and response threshold of each point, analyze the adjustment effect, and generate dynamically optimized calibration parameters;

[0116] The deviation analysis and data collation module adjusts the response characteristics and thresholds of the control points based on dynamically optimized calibration parameters, calculates the deviation between the actual output and the expected output, adjusts the calibration parameters to reduce the deviation, monitors the adjustment process and collects performance data for data collation to obtain the final calibration record.

[0117] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent method for instrument calibration, characterized in that: The following steps are involved: Collect initial pressure data from the pressure transmitter, perform signal denoising and data standardization processing, and obtain standardized pressure data; Analyzing the standardized pressure data by differential geometry theory, performing parameter space curvature measurement, and calculating geodesic distribution to generate a parameter curvature map; Based on the parameter curvature map, perform a calibration path minimum curvature search, capture a path with minimum curvature, and perform path adjustment to obtain an optimized calibration path; According to the optimized calibration path, the control points of the pressure transmitter are spatially reconfigured, with high-density layout in the high-curvature area and low-density layout in the low-curvature area, and the control points are adjusted according to the feedback of the calibration effect to obtain the adjusted control point configuration; In combination with the neural network model, dynamically adjusting the parameters of the adjusted control point configuration, optimizing the real-time calibration process, and adjusting the calibration parameters according to the output error to generate dynamically optimized calibration parameters; The dynamically optimized calibration parameters are used to complete the final calibration of the pressure transmitter, and error analysis and performance verification are performed.

2. The intelligent instrument calibration method according to claim 1, characterized in that: The standardized pressure data is specifically a pure signal with noise removed and a data value that has been normalized. The parameter curvature map includes color coding of curvature values ​​and curvature contour lines. The optimized calibration path includes minimizing path length, optimizing path curvature, and improving path smoothness. The adjusted control point configuration includes optimizing control point spacing, classifying control point sensitivity, and adjusting control point response time. The dynamically optimized calibration parameters include pressure thresholds, temperature compensation coefficients, and pressure response time adjusted with real-time feedback.

3. The intelligent instrument calibration method according to claim 1, characterized in that: The standardized pressure data is analyzed by differential geometry theory, parameter space curvature measurement is performed, and geodesic distribution is calculated to generate a parameter curvature map in the following steps: The standardized pressure data are analyzed using differential geometry theory, local maxima and minima are identified by calculating the second-order derivatives of the data points, high points and low points of the curvature are marked, and local curvature marks are generated; Based on the local curvature mark, a spatial analysis technique is applied to map the curvature change of the overall parameter space, determine the area with the maximum curvature change, and generate a curvature change map; According to the curvature variation graph, a continuous path with minimized curvature is extracted to ensure that each segment on the connection path achieves curvature minimization, capture the optimal geodesic, and obtain a parameter curvature graph.

4. The intelligent method for instrument calibration according to claim 1, characterized in that: Based on the parameter curvature map, the minimum curvature search of the calibration path is performed, the path with the minimum curvature is captured, and the path is adjusted. The steps of obtaining the optimized calibration path are as follows: Based on the parameter curvature map, identifying the local minimum point of curvature in the entire map, marking all potential minimum curvature regions, and obtaining a potential minimum curvature point set; According to the potential minimum curvature point set, a curvature comparison algorithm is applied to compare the curvature values ​​of adjacent data points point by point, minimizing the curvature difference between each point and adjacent points, connecting to form a continuous path, and obtaining a preliminary minimum curvature path; Based on the preliminary minimum curvature path, path adjustment is performed, and the position of each point on the path is adjusted by a local optimization adjustment algorithm to generate an optimized calibration path.

5. The intelligent method for instrument calibration according to claim 4, characterized in that: The local optimization adjustment algorithm is based on the formula: Calculate the new position of each point on the path ,in, is the position of the current point on the path, is the learning rate, is the curvature function About Dot The gradient of is the adjustment factor, Yes The cosine value of the current direction and the optimized direction, is a distance-based attenuation factor, Yes The distance to the starting point of the path, So The square of is an exponential decay function.

6. The intelligent method for instrument calibration according to claim 1, characterized in that: According to the optimized calibration path, the control points of the pressure transmitter are spatially reconfigured, high-density layout is performed in the high curvature area, low-density layout is performed in the low curvature area, and the control points are adjusted according to the feedback of the calibration effect. The steps of obtaining the adjusted control point configuration are as follows: Based on the optimized calibration path, the curvature analysis technology is used to scan the entire calibration path, identify and mark high curvature areas and low curvature areas, accurately divide each area, and generate a regional marking map; Deploy dense control points in the high curvature area indicated by the regional marking map, reduce the density of control points in the low curvature area, optimize the layout of each control point, and obtain a preliminary control point layout plan; Based on the preliminary control point layout plan, calibration data is automatically collected and analyzed, and according to the analysis results, the position and density parameters of the control points are adjusted. Each control point is adjusted through loop feedback to optimize the calibration performance and obtain the adjusted control point configuration.

7. The intelligent method for instrument calibration according to claim 1, characterized in that: Combined with the neural network model, the adjusted control point configuration is dynamically adjusted to optimize the real-time calibration process, and the calibration parameters are adjusted according to the output error. The steps of generating dynamically optimized calibration parameters are: Based on the adjusted control point configuration, in combination with the neural network model, the input layer of the neural network is configured to receive the control point data and the real-time pressure reading, the input parameters are set and the real-time capture of the data stream is started to generate the configuration data stream; According to the configuration data stream, the real-time data analysis function of the neural network is activated to automatically process the input data, identify the control points with large deviations, calculate the required parameter adjustment values ​​using the gradient descent method, and generate a preliminary parameter adjustment plan; Based on the preliminary parameter adjustment plan, the parameter settings of all control points are refined one by one, the position and response threshold of each point are optimized, the calibration data during the adjustment process is monitored in real time, the adjustment effect is analyzed, and a preliminary parameter adjustment plan is generated. Dynamically optimized calibration parameters.

8. The intelligent method for instrument calibration according to claim 7, characterized in that: The gradient descent method, according to the formula: Calculate parameter adjustment values ,in, is the current parameter value, is the learning rate, is the loss function About parameters The gradient of is the regularization coefficient, is the sample variance, is the gradient sign adjustment coefficient, is the sign of the gradient, is the square root of the scale of the gradient.

9. An intelligent instrument calibration system, characterized in that: According to any one of claims 1 to 8, the intelligent instrument calibration method comprises: The data acquisition and processing module collects and records the initial pressure readings from the pressure transmitter, removes the noise exceeding the threshold in the low-pass filter, and normalizes the data to obtain standardized pressure data; The curvature analysis module uses differential geometry theory to analyze the standardized pressure data, mark the high points and low points of the curvature, determine the area with the maximum curvature change, extract the continuous path with the minimum curvature, and obtain a parameter curvature map; The path optimization module identifies the local minimum point of curvature in the entire graph according to the parameter curvature graph, applies a curvature comparison algorithm to compare the curvature values ​​of adjacent data points point by point, connects them to form a continuous path, adjusts the position of each point on the path, and generates an optimized calibration path; The layout optimization module identifies and marks high curvature areas and low curvature areas based on the optimized calibration path, deploys dense control points in high curvature areas, reduces the density of control points in low curvature areas, optimizes the layout of each control point, and obtains a preliminary control point layout plan; The configuration data acquisition module adjusts the position and density parameters of the control points based on the preliminary control point layout plan, combines the neural network model, configures the input layer of the neural network to receive control point data and real-time pressure readings, sets input parameters and starts real-time capture of data streams, and generates configuration data streams; The calibration parameter optimization module uses a neural network to perform real-time data analysis based on the configuration data stream, identifies control points with large deviations, automatically calculates the required parameter adjustment values, optimizes the position and response threshold of each point, analyzes the adjustment effect, and generates dynamically optimized calibration parameters; the deviation analysis and data collation module adjusts the response characteristics and thresholds of the control points based on the dynamically optimized calibration parameters, calculates the deviation between the actual output and the expected output, adjusts the calibration parameters to reduce the deviation, monitors the adjustment process and collects performance data for data collation to obtain the final calibration record.