Differential pressure sensor error compensation and prediction method for high static pressure environment
By collecting and processing differential pressure sensor data in high static pressure environments and using deep learning and time series models for error prediction, the error compensation and prediction problems of sensors in high static pressure environments are solved, accurate error compensation and future error warning are achieved, and the system's response speed and maintenance capabilities are improved.
Patent Information
- Application Number
- CN202510452906.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-09
- Publication Date
- 2025-07-25
AI Technical Summary
The existing differential pressure sensors lack multi-dimensional environmental parameter analysis in high-static pressure environments, resulting in limitations in univariate analysis and lack of combination of error compensation and prediction, which makes it impossible to achieve future active error warning, and insufficient system response speed and prevention and maintenance capabilities.
By collecting the differential pressure sensor readings and environmental data in high-static pressure environments, preprocessing and feature extraction, using deep learning algorithms to build an error prediction model, combining time series prediction models, error compensation and prediction are achieved, error prediction reports are generated, and future error warnings are performed.
It improves the measurement accuracy of the sensor in high static pressure environment, realizes accurate compensation and prediction of errors, and enhances the system's response speed and preventive and maintenance capabilities.
Smart Images
Figure CN120369186A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of differential pressure prediction, and more particularly to a method for error compensation and prediction of differential pressure sensors in a high static pressure environment. Background Art
[0002] In some industrial and scientific research applications, such as oil drilling, submersibles, deep-sea equipment, aerospace, etc., equipment needs to work in extremely high-pressure environments. In these environments, differential pressure sensors are used to measure the pressure difference of fluids or gases, and are key components to ensure the normal operation of the equipment. However, high static pressure environments often have an adverse impact on the performance of differential pressure sensors, resulting in inaccurate sensor readings; currently, the error processing of differential pressure sensors in high static pressure environments mainly relies on calibration procedures, and combines the temperature coefficient of the sensor to compensate and correct the error of the sensor. At the same time, a neural network model is used to learn the error pattern from historical data to predict future errors;
[0003] However, the above process still has the following disadvantages:
[0004] Firstly, existing differential pressure sensor error compensation and prediction methods mostly rely only on the original readings of the sensors to predict the errors of the sensors, lacking the construction of an error prediction model by collecting multi-dimensional environmental parameters in a high-pressure environment, resulting in more limited univariate analysis of the sensors;
[0005] Secondly, existing differential pressure sensor error compensation and prediction methods lack the combination of error compensation and error prediction to form a complete error compensation and prediction method, and cannot achieve active early warning of future errors, resulting in insufficient response speed and preventive maintenance ability of the system. Summary of the Invention
[0006] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for error compensation and prediction of differential pressure sensors in a high static pressure environment to solve the problems existing in the above background art.
[0007] The present invention provides the following technical solutions: A method for error compensation and prediction of differential pressure sensors in a high static pressure environment, comprising:
[0008] S1: Collecting the readings of differential pressure sensors in a high static pressure environment to obtain differential pressure data, and recording the corresponding environmental data at the same time;
[0009] S2: Preprocessing and feature extraction of the differential pressure data and the corresponding environmental data to extract the key feature parameters affecting the sensor;
[0010] S3: By dividing the extracted key feature parameters into a training set, a validation set, and a test set, based on the physical characteristics and working principle of the sensor, use a deep learning algorithm to construct an error prediction model for the sensor, and then use the trained error prediction model to predict the measurement error of the sensor to obtain an error prediction value;
[0011] S4: By comparing the error prediction value with the actual measurement result and calculating the compensated differential pressure value, it is used to compensate the measurement error of the sensor;
[0012] S5: By inputting the historical measurement values and the corresponding error values into the trained time series prediction model, an error prediction value is obtained to predict the sensor error in the future time period;
[0013] S6: By feeding back the error prediction results in the future time period and the compensated differential pressure value to the manager's terminal, and automatically generating an error prediction report and sending it to the manager's terminal.
[0014] Preferably, in S1, under a high static pressure environment, the differential pressure sensor and the environmental parameter sensor are installed at positions reflecting the environmental conditions and differential pressure changes, and the same data acquisition frequency is set for the differential pressure sensor and the environmental parameter sensor according to the environmental change rate, for collecting differential pressure data and the corresponding environmental data, and synchronously recording the differential pressure data and the corresponding environmental data. The environmental parameter sensor includes a temperature sensor, a humidity sensor, a pressure sensor, and a vibration sensor, and the environmental parameters include temperature, humidity, atmospheric pressure, and vibration.
[0015] Preferably, the preprocessing operations on the differential pressure data and the corresponding environmental parameters in S2 include data cleaning, data conversion, data standardization, and data integration, and then key feature parameters affecting the sensor are extracted from the preprocessed differential pressure data and the corresponding environmental data. The extracted key feature parameters include temperature fluctuation value, pressure change rate, humidity change value, and vibration change value.
[0016] Preferably, in S3, a neural network is used as the deep learning architecture to construct an error prediction model for the sensor according to the physical characteristics and working principle of the sensor. Based on dividing the extracted key feature parameters into a training set, a validation set, and a test set for model training, the error prediction model of the sensor is trained using the training set, the parameters and structure of the model are adjusted using the validation set, the performance of the model is evaluated using the test set, and the measurement error of the sensor is predicted by extracting the same key feature parameters and inputting them into the trained error prediction model, and an error prediction value is output. The specific prediction formula is Wherein, The output value of the representation model, i.e., the error prediction value, X represents the input key feature parameter matrix, W1 represents the weight from the input layer to the hidden layer, b1 represents the bias term from the input layer to the hidden layer, W2 represents the weight from the hidden layer to the output layer, b2 represents the bias term from the hidden layer to the output layer, and f represents the activation function.
[0017] Preferably, in S4, the actual differential pressure measurement value M of the differential pressure sensor is obtained, and the predicted error prediction value is used to compensate the actual measurement value, and the compensated differential pressure value is calculated as where M′ represents the compensated differential pressure measurement value closer to the true value.
[0018] Preferably, in S5, the historical measurement values and corresponding error values of the sensor are collected, preprocessed, and an autoregressive model is selected. The historical measurement values and corresponding error values are divided into a training set, a validation set, and a test set for training and evaluating the autoregressive model. Then, the historical error values are used as the features input to the autoregressive model, and the error prediction values are output to predict the sensor error in a future time period;
[0019] The specific analysis method of the error prediction value is as follows:
[0020] Step S511: Collect the historical measurement values M1, M2, …, M t and the corresponding true values m1, m2, …, m t at the corresponding time points;
[0021] Step S512: Conduct error analysis on the measurement value at each time point and the corresponding true value at the time point, and calculate the error value at each historical time point as E t = M t - m t , where E t represents the error value at time point t, M t represents the measurement value at time point t, and m t represents the corresponding true value at time point t;
[0022] Step S513: By using the error value E t at each historical time point as the feature input to the autoregressive model, and outputting the error prediction value as The specific formula is where represents the model error prediction value at time point t, φ i represents the autoregressive coefficient, p represents the order of the model, E t-i represents the historical error at time point t - i, and ε tDenote the error term at time point t, and output the estimated error value as the sensor error within the predicted future time period.
[0023] Preferably, in S6, the estimated error value and the compensated differential pressure value are sent to the manager's terminal, and then an error prediction report is automatically generated according to the error prediction analysis process, and an error estimation threshold is set. If it is detected that the estimated error value exceeds the error estimation threshold, an early warning prompt is immediately triggered.
[0024] Technical effects and advantages of the present invention:
[0025] The present invention collects differential pressure sensor readings in a high static pressure environment to obtain differential pressure data, records the corresponding environmental data at the same time, preprocesses the collected differential pressure data and the corresponding environmental data, extracts the key characteristic parameters affecting the sensor from them, constructs an error prediction model of the sensor by using a deep learning algorithm, then uses the trained error prediction model to predict the measurement error of the sensor to obtain an error prediction value, compares the error prediction value with the actual measurement result, and calculates the compensated differential pressure value for error compensation of the measurement error of the sensor. By inputting the historical measurement values and the corresponding error values into the trained time series prediction model, an estimated error value is obtained to predict the sensor error within the future time period, and then the error prediction result within the future time period and the compensated differential pressure value are fed back to the manager's terminal, and an error prediction report is automatically generated and sent to the manager's terminal. By collecting multi-dimensional environmental parameters in a high-pressure environment to construct an error prediction model, the problem of predicting the error of the sensor relying on the original readings of the sensor is solved, and the limitation of traditional univariate analysis is avoided. By combining error compensation and error prediction to form a complete error compensation and prediction, active early warning of future errors is realized, the response speed and preventive maintenance ability of the system are improved, it is beneficial to more accurately predict and compensate errors, and further improve the accuracy of the sensor. Description of the Drawings
[0026] Figure 1 It is a method step diagram of the present invention.
[0027] Figure 2 It is a system structure block diagram of this embodiment. Detailed Embodiments
[0028] The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. In addition, the forms of the various structures described in the following embodiments are merely illustrative. A method for error compensation and prediction of a differential pressure sensor in a high static pressure environment according to the present invention is not limited to the various structures described in the following embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.
[0029] As Figure 1 shown, this embodiment provides a method for error compensation and prediction of a differential pressure sensor in a high static pressure environment, including:
[0030] S1: Collect the differential pressure sensor readings in a high static pressure environment to obtain differential pressure data, and record the corresponding environmental data at the same time.
[0031] In this embodiment, the S1 is based on installing the differential pressure sensor and the environmental parameter sensor at a position reflecting the environmental conditions and differential pressure changes in a high static pressure environment, and setting the same data acquisition frequency for the differential pressure sensor and the environmental parameter sensor according to the environmental change rate to collect differential pressure data and the corresponding environmental data, and synchronously record the differential pressure data and the corresponding environmental data. The environmental parameter sensor includes a temperature sensor, a humidity sensor, a pressure sensor, and a vibration sensor, and the environmental parameters include temperature, humidity, atmospheric pressure, and vibration.
[0032] S2: Preprocess and extract features from the differential pressure data and the corresponding environmental data to extract the key feature parameters affecting the sensor.
[0033] In this embodiment, the preprocessing operations of the S2 on the differential pressure data and the corresponding environmental parameters include data cleaning, data conversion, data normalization, and data integration, and then extract the key feature parameters affecting the sensor from the preprocessed differential pressure data and the corresponding environmental data. The extracted key feature parameters include temperature fluctuation value, pressure change rate, humidity change value, and vibration change value.
[0034] S3: Divide the extracted key feature parameters into a training set, a validation set, and a test set. Based on the physical characteristics and working principles of the sensor, use a deep learning algorithm to construct an error prediction model of the sensor, and then use the trained error prediction model to predict the measurement error of the sensor to obtain an error prediction value.
[0035] In this embodiment, S3 constructs an error prediction model of the sensor by using a neural network as a deep learning architecture according to the physical characteristics and working principle of the sensor. Based on dividing the extracted key feature parameters into a training set, a validation set, and a test set for model training, the error prediction model of the sensor is trained using the training set, the parameters and structure of the model are adjusted using the validation set, and the performance of the model is evaluated using the test set. By extracting the same key feature parameters and inputting them into the trained error prediction model, the measurement error of the sensor is predicted, and the error prediction value is output. The specific prediction formula is Where represents the output value of the model, that is, the error prediction value, X represents the input key feature parameter matrix, W1 represents the weight from the input layer to the hidden layer, b1 represents the bias term from the input layer to the hidden layer, W2 represents the weight from the hidden layer to the output layer, b2 represents the bias term from the hidden layer to the output layer, and f represents the activation function.
[0036] S4: By comparing the error prediction value with the actual measurement result and calculating the compensated differential pressure value, the measurement error of the sensor is compensated for error.
[0037] In this embodiment, S4 obtains the actual differential pressure measurement value M of the differential pressure sensor and uses the predicted error prediction value to compensate the actual measurement value, and calculates the compensated differential pressure value as Where M′ represents the compensated differential pressure measurement value closer to the true value.
[0038] S5: By inputting the historical measurement value and the corresponding error value into the trained time series prediction model, an error prediction value is obtained to predict the sensor error in the future time period.
[0039] In this embodiment, S5 collects the historical measurement values and the corresponding error values of the sensor, performs preprocessing, selects an autoregressive model, divides the historical measurement values and the corresponding error values into a training set, a validation set, and a test set for training and evaluating the autoregressive model, and then uses the historical error value as the feature input into the autoregressive model and outputs the error prediction value to predict the sensor error in the future time period;
[0040] The specific analysis method of the error prediction value is as follows:
[0041] Step S511: Collect the historical measurement values M1, M2,..., M t and the corresponding true values m1, m2,..., m t at the corresponding time points;
[0042] Step S512: Conduct error analysis on the measured values at each time point and the corresponding true values at the time points, and calculate the error value E for each historical time point t = M t - m t , where E t represents the error value at time point t, M t represents the measured value at time point t, and m t represents the true value at the time point corresponding to time point t;
[0043] Step S513: By taking the error value E at each historical time point t as the feature input to the autoregressive model and outputting the error prediction value as The specific formula is where represents the model error prediction value at time point t, φ i represents the autoregressive coefficient, p represents the order of the model, E t-i represents the historical error at time point t - i, and ε t represents the error term at time point t, and the error prediction value is output as the sensor error within the predicted future time period.
[0044] S6: Feed back the error prediction results within the future time period and the compensated differential pressure value to the manager's terminal, and automatically generate an error prediction report and send it to the manager's terminal.
[0045] In this embodiment, S6 sends the error prediction value and the compensated differential pressure value to the manager's terminal, then automatically generates an error prediction report according to the error prediction analysis process, and sets an error prediction threshold. If it is detected that the error prediction value exceeds the error prediction threshold, an early warning prompt is immediately triggered.
[0046] As Figure 2 shown, this embodiment provides an implementation system corresponding to an article anti-counterfeiting method based on local feature visual information, including a data collection module, a feature extraction module, an error prediction module, an error compensation module, an error prediction module, and a result feedback module. The data collection module is connected to the feature extraction module, the feature extraction module is connected to the error prediction module, the error prediction module is connected to the error compensation module, the error compensation module is connected to the result feedback module, and the error prediction module is connected to the result feedback module.
[0047] The data collection module is used to collect differential pressure sensor readings in a high static pressure environment to obtain differential pressure data and record the corresponding environmental data at the same time;
[0048] The feature extraction module is used to preprocess and extract features from the differential pressure data and the corresponding environmental data, and extract the key feature parameters that affect the sensor;
[0049] The error prediction module divides the extracted key feature parameters into a training set, a validation set, and a test set. Based on the physical characteristics and working principle of the sensor, a deep learning algorithm is used to construct an error prediction model of the sensor. Then, the trained error prediction model is used to predict the measurement error of the sensor to obtain an error prediction value;
[0050] The error compensation module compares the error prediction value with the actual measurement result, and calculates the compensated differential pressure value for error compensation of the measurement error of the sensor;
[0051] The error estimation module inputs the historical measurement values and the corresponding error values into the trained time series prediction model to obtain an error estimation value, and predicts the sensor error in a future time period;
[0052] The result feedback module feeds back the error prediction result in the future time period and the compensated differential pressure value to the management terminal, and automatically generates an error prediction report and sends it to the management terminal.
[0053] Finally: The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
[0054] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in this application, and all should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.
Claims
1. A method for error compensation and prediction of a differential pressure sensor in a high static pressure environment, characterized in that, Including: S1: Used to collect the differential pressure sensor readings in a high static pressure environment to obtain differential pressure data, and record the corresponding environmental data at the same time; S2: Used to preprocess and extract features from the differential pressure data and the corresponding environmental data, and extract the key feature parameters that affect the sensor; S3: By dividing the extracted key feature parameters into a training set, a validation set and a test set, based on the physical characteristics and working principles of the sensor, using a deep learning algorithm to construct an error prediction model of the sensor, and then using the trained error prediction model to predict the measurement error of the sensor to obtain an error prediction value; S4: By comparing the error prediction value with the actual measurement result and calculating the compensated differential pressure value, used to compensate the measurement error of the sensor; S5: By inputting the historical measurement values and the corresponding error values into the trained time series prediction model to obtain an error prediction value, and predicting the sensor error in the future time period; S6: By feeding back the error prediction results in the future time period and the compensated differential pressure value to the manager's terminal, and automatically generating an error prediction report and sending it to the manager's terminal.
2. A differential pressure sensor error compensation and prediction method for a high static pressure environment according to claim 1, wherein The S1 is based on installing the differential pressure sensor and the environmental parameter sensor at positions that reflect the environmental conditions and differential pressure changes in a high static pressure environment, and setting the same data acquisition frequency for the differential pressure sensor and the environmental parameter sensor according to the environmental change rate, for collecting differential pressure data and the corresponding environmental data, and synchronously recording the differential pressure data and the corresponding environmental data. The environmental parameter sensor includes a temperature sensor, a humidity sensor, a pressure sensor and a vibration sensor, and the environmental parameters include temperature, humidity, atmospheric pressure and vibration.
3. A differential pressure sensor error compensation and prediction method for a high static pressure environment according to claim 1, characterized in that The preprocessing operations of the S2 on the differential pressure data and the corresponding environmental parameters include data cleaning, data conversion, data standardization and data integration, and then extracting the key feature parameters that affect the sensor from the preprocessed differential pressure data and the corresponding environmental data. The extracted key feature parameters include temperature fluctuation value, pressure change rate, humidity change value and vibration change value.
4. A differential pressure sensor error compensation and prediction method for a high static pressure environment according to claim 1, characterized in that The S3 constructs an error prediction model of the sensor by using a neural network as a deep learning architecture according to the physical characteristics and working principle of the sensor. Based on dividing the extracted key feature parameters into a training set, a validation set, and a test set for model training, the error prediction model of the sensor is trained using the training set, the parameters and structure of the model are adjusted using the validation set, and the performance of the model is evaluated using the test set. By extracting the same key feature parameters and inputting them into the trained error prediction model, the measurement error of the sensor is predicted, and an error prediction value is output. The specific prediction formula is Among them, represents the output value of the model, that is, the error prediction value. X represents the input key feature parameter matrix. W1 represents the weight from the input layer to the hidden layer. b1 represents the bias term from the input layer to the hidden layer. W2 represents the weight from the hidden layer to the output layer. b2 represents the bias term from the hidden layer to the output layer. f represents the activation function.
5. A differential pressure sensor error compensation and prediction method for a high static pressure environment according to claim 1, characterized in that The S4 compensates the actual measured value by obtaining the actual differential pressure measurement value M of the differential pressure sensor and using the predicted error prediction value and calculates the compensated differential pressure value as where M' represents the compensated differential pressure measurement value that is closer to the true value.
6. A differential pressure sensor error compensation and prediction method for a high static pressure environment according to claim 1, characterized in that, The S5 collects the historical measurement values and the corresponding error values of the sensor, and performs preprocessing, selects an autoregressive model, divides the historical measurement values and the corresponding error values into a training set, a validation set and a test set, for training and evaluating the autoregressive model, and then uses the historical error values as the features input into the autoregressive model and outputs an error prediction value to predict the sensor error in the future time period; The specific analysis method of the error prediction value is: Step S511: Collect the historical measurement values M1, M2, …, M t and the corresponding true values m1, m2, …, m t ; Step S512: Conduct error analysis on the measured values at each time point and the corresponding true values at the time points, and calculate the error value E for each historical time point t = M t - m t , where E t represents the error value at time point t, M t represents the measured value at time point t, and m t represents the true value at the time point corresponding to time point t; Step S513: By taking the error value E at each historical time point t as the feature input to the autoregressive model and outputting the error prediction value as The specific formula is where represents the model error prediction value at time point t, φ i represents the autoregressive coefficient, p represents the order of the model, E t-i represents the historical error at time point t-i, ε t represents the error term at time point t, and the error prediction value is output as the sensor error in the future time period.
7. A method for error compensation and prediction of a differential pressure sensor in a high static pressure environment according to claim 1, characterized in that, The S6 sends the error prediction value and the compensated differential pressure value to the manager's terminal, then automatically generates an error prediction report according to the error prediction analysis process, and sets an error prediction threshold. If it is detected that the error prediction value exceeds the error prediction threshold, an alarm prompt is immediately triggered.