Electric submersible pump working condition pressure sensor data preprocessing method

Abstract

The invention discloses an electric submersible pump working condition pressure sensor data preprocessing method. The method comprises the following steps: S1, receiving data of a working condition pressure sensor; s2, filtering the received initial data, removing high-frequency noise and interference, and retaining key fluctuation and change information; s3, performing normalization processing on the filtered data, and scaling the data to a specific range to ensure data consistency and comparability; s4, sampling is carried out on the data after linear normalization, the data dimension and storage requirements are reduced, and meanwhile key features of the data are reserved; and S5, outputting data. The method has the beneficial effects that the data quality and stability are improved by preprocessing the data of the pressure sensor; interference of noise and abnormal values is effectively reduced, working condition monitoring accuracy is improved, and prediction and diagnosis capabilities are enhanced; reliable basic data are provided for subsequent analysis and judgment of working conditions, and a solution provided for monitoring and management of an electric submersible pump system is more reliable and efficient.

Description

Technical Field

[0001] The invention relates to a method for monitoring the working condition of a submersible electric pump, in particular to a method for preprocessing data of a working condition pressure sensor of a submersible electric pump, and belongs to the technical field of oil and gas development. Background Art

[0002] In the field of submersible pump condition monitoring, accurately acquiring and analyzing submersible pump condition data is the key to improving production efficiency, extending equipment life and reducing failures. Data preprocessing is an important part of pressure sensor data processing, which involves filtering, normalizing and sampling the original data. This improves data quality, reduces the impact of noise and outliers, and provides a reliable data basis for subsequent condition analysis and judgment.

[0003] At present, there are still some deficiencies in the preprocessing of submersible pump working pressure sensor data. First, noise and interference are common problems. Due to environmental factors and the influence of the equipment itself, there are often high-frequency noise and interference in the pressure sensor data. Secondly, the processing of outliers is also one of the challenges. Due to sensor failure, sampling errors or other factors, data mutations occur, resulting in data outliers. In addition, the pressure sensor data ranges of different submersible pumps are different, which makes direct comparison and analysis of data difficult. Summary of the invention

[0004] In order to overcome the above-mentioned deficiencies in the existing preprocessing of data of the working pressure sensor of the submersible electric pump, the present invention provides a method for preprocessing data of the working pressure sensor of the submersible electric pump.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for preprocessing data of a submersible electric pump working pressure sensor, the steps of which are as follows:

[0006] S1. Receive data: receive initial data from the working pressure sensor;

[0007] S2, filtering: filtering the received initial data to remove high-frequency noise and interference and retain key fluctuation and change information;

[0008] S3, normalization processing: normalize the filtered data and scale the data to a specific range to ensure data consistency and comparability;

[0009] S4, sampling processing: sampling from the linearly normalized data to reduce data dimension and storage requirements while retaining key data features;

[0010] S5. Data output: The pre-processed data is output to the subsequent pump operating condition analysis and prediction model for further processing and analysis.

[0011] Furthermore, in step S2, a moving average filtering processing method is adopted according to the working characteristics of the submersible electric pump working condition.

[0012] Furthermore, in step 3, a linear normalization method is used for processing.

[0013] Furthermore, in step 4, an equal-interval sampling processing method is adopted.

[0014] In step 2, the data samples collected by the pressure sensor are processed by moving average filtering to obtain smooth data.

[0015]

[0016] Where: filtered_value[i] is the value of the collected data sample after filtering,

[0017] x[i] is the original data sample collected by the pressure sensor,

[0018] M is the window length of the filter,

[0019] The window length M of the filter in formula (1) is:

[0020] M=K·T

[0021] Where: M is the window length of the filter,

[0022] K is the coefficient,

[0023] T is the time scale,

[0024] The value range of K is 0.5 to 2; the value range of window length M is 5 to 100.

[0025] In step 3, the values ​​after smoothing filtering are linearly normalized, and the processing process is expressed as follows:

[0026]

[0027] Where: normalized_value[i] is the normalized value,

[0028] filtered_value[i] is the value of the collected data sample after filtering.

[0029] min_value is the minimum value of the collected data samples.

[0030] max_value is the maximum value of the collected data samples.

[0031] After calculation by formula (2), the normalized_value[i] data will be mapped to the range of 0 to 1.

[0032] In step 4, the normalized data is subjected to equal-interval sampling, and the specific method is expressed as follows:

[0033] sampled_value[i]=normalized_value[i·step] (3)

[0034] Where: sampled_value[i] is the sampled value,

[0035] i is the index of the data sample, ranging from 0 to N-1, where N is the length of the input data and step is the sampling step.

[0036] In step 2,

[0037] The coefficient K is selected based on experience and demand;

[0038] The time scale T depends on the working characteristics and signal changes of the downhole pressure sensor;

[0039] The window length M is adjusted according to actual tests.

[0040] The sampling step length step in the formula (3) is selected according to the data sampling rate and application requirements, and the value range is 20 to 100.

[0041] In step S4, key feature data is important information for analyzing and predicting the working condition of the submersible electric pump, which is extracted from the original data by a preprocessing method. The key features include: basic pressure change trend, key fluctuation and change information, normalized value range, and data points sampled at equal intervals.

[0042] The beneficial effects of the present invention are as follows: by preprocessing the pressure sensor data, the data quality and stability are improved; the interference of noise and abnormal values ​​is effectively reduced, the accuracy of working condition monitoring is improved, and the prediction and diagnosis capabilities are enhanced; reliable basic data is provided for subsequent analysis and judgment of working conditions, and the solution provided for the monitoring and management of submersible electric pump systems is more reliable and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the data preprocessing method of the submersible electric pump working condition pressure sensor of the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. However, those skilled in the art should be aware that the present invention is not limited to the specific embodiments listed, and all embodiments should be included in the protection scope of the present invention as long as they conform to the spirit of the present invention.

[0045] See attached Figure 1 The present invention provides a method for preprocessing data of a submersible electric pump working pressure sensor, and the steps are as follows:

[0046] S1, receiving data;

[0047] S2, filtering processing;

[0048] S3, normalization processing;

[0049] S4, sampling process;

[0050] S5. Data output.

[0051] The data preprocessing method of the submersible electric pump working condition pressure sensor of the present invention is as follows:

[0052] S1. Receive data

[0053] Receive operating pressure sensor data.

[0054] S2. Filtering

[0055] The received initial data of the pressure sensor is filtered to remove high-frequency noise and interference, and retain key fluctuation and change information; preferably, a moving average filtering method is used according to the working characteristics of the submersible electric pump working condition.

[0056] The data samples collected by the pressure sensor are processed by moving average filtering to obtain smooth data, as follows:

[0057]

[0058] Where: filtered_value[i] is the value of the collected data sample after filtering,

[0059] x[i] is the original data sample collected by the pressure sensor,

[0060] M is the window length of the filter.

[0061] The window length M of the filter in formula (1) is:

[0062] M=K·T

[0063] Where:

[0064] M is the window length of the filter. The window length M is adjusted according to actual tests. The value range of the window length M is 5 to 100 to meet application requirements.

[0065] K is a coefficient. When calculating, the coefficient K is selected based on experience and needs. It can be adjusted according to the specific application situation. The value range of K is 0.5 to 2.

[0066] T is the time scale, which depends on the working characteristics and signal changes of the downhole pressure sensor.

[0067] S3. Normalization processing: normalize the filtered data and scale the data to a specific range to ensure data consistency and comparability; preferably, a linear normalization method is used for processing.

[0068] The values ​​after smoothing and filtering in step 2 are linearly normalized. The processing process is expressed as follows:

[0069]

[0070] Where: normalized_value[i] is the normalized value,

[0071] filtered_value[i] is the value of the collected data sample after filtering.

[0072] min_value is the minimum value of the collected data samples.

[0073] max_value is the maximum value of the collected data samples.

[0074] After calculation by formula (2), the normalized_value[i] data will be mapped to the range of 0 to 1.

[0075] The normalized data normalized_value[i] can eliminate the influence of the absolute size of the original data on the result and provide a more unified data representation.

[0076] S4, sampling processing: sampling from the linearly normalized data to reduce data dimension and storage requirements while retaining key data features;

[0077] Preferably, an equal-interval sampling processing method is adopted.

[0078] The data after normalization in step 3 is sampled at equal intervals, and the specific method is expressed as follows:

[0079] sampled_value[i]=normalized_value[i·step] (3)

[0080] Where: sampled_value[i] is the sampled value,

[0081] i is the index of the data sample, ranging from 0 to N-1, where N is the length of the input data.

[0082] step is the sampling step size.

[0083] The sampling step length step is selected according to the sampling rate of the data and the application requirements, preferably, the value range is 20 to 100.

[0084] S5. Data output: The pre-processed data is output to the subsequent pump operating condition analysis and prediction model for further processing and analysis.

[0085] Furthermore, in the filtering process of step S2, in order to ensure the accuracy and reliability of the operating condition data, high-frequency noise and interference are removed, and key fluctuation and change information is retained, wherein:

[0086] The high-frequency noise is a fast, short-term data fluctuation caused by internal and external interference factors. High-frequency noise can distort the real signal and lead to errors in data analysis.

[0087] The key fluctuations are data reflecting real equipment status or operation changes, such as the start / stop of a pump, or sudden changes in oil well pressure. These fluctuations provide key information for operating condition analysis.

[0088] The change information reflects the long-term trend or pattern of the data and is related to the long-term behavior or health status of the device.

[0089] In order to effectively separate these components from the raw data, filtering processing is applied to eliminate noise and highlight key fluctuations and changes. Through this pre-processing method, the accuracy and effectiveness of subsequent analysis can be ensured, making the operating condition analysis and prediction more reliable.

[0090] High-frequency noise is related to the motor speed, number of blades, design and purpose of the submersible electric pump. For example, the blade channel frequency in a submersible centrifugal pump, a pump has 6 blades and rotates at a speed of 50Hz, it will generate a blade channel frequency at 300Hz, at this time the frequency of high-frequency noise can be set to 300Hz. It should be noted that this is only a rough estimate. In order to more accurately determine the range of high-frequency noise, it is necessary to specifically analyze the operating characteristics of the pump, the measured data and possible external interference.

[0091] Furthermore, in the sampling process of step S4, the key features of the data are information retained during the data preprocessing process, which is representative and has decisive significance for subsequent analysis and prediction. The key features include: basic pressure change trend, key fluctuation and change information, normalized numerical range, and data points sampled at equal intervals.

[0092] The basic pressure variation trend is data obtained after filtering and represents the actual working condition of the submersible electric pump. This trend reflects the basic operating characteristics of the submersible electric pump at different time points or working conditions.

[0093] The key fluctuation and change information is the information retained during the filtering process, and is related to certain specific operations or states of the submersible pump. For example, when the submersible pump is started, shut down or encounters an abnormality, these key fluctuations may become obvious indicators.

[0094] The normalized numerical range is obtained by converting data into a uniform range through normalization processing, thereby ensuring the consistency and comparability of data from different data sources or different time periods.

[0095] The equally spaced sampling of data points ensures the simplicity and manageability of the data, but at the same time this sampling method also ensures that the above-mentioned key features are retained in the sampled data.

[0096] The data preprocessing algorithm of the submersible electric pump working condition pressure sensor of the present invention is realized by using C language, and is realized by combining a PIC single chip microcomputer and an AD7714 digital-to-analog converter.

[0097] The data preprocessing method of the submersible electric pump working condition pressure sensor of the present invention has the following advantages, characteristics and positive effects:

[0098] 1. Improve data quality: By preprocessing the pressure sensor data of the submersible pump working condition, noise and outliers can be effectively filtered out, the data quality and accuracy can be improved, the misjudgment and misleading results can be reduced, and the reliability of subsequent analysis and decision-making can be improved.

[0099] 2. Improve the accuracy of working condition monitoring: By preprocessing the pressure sensor data, the working state of the submersible electric pump can be identified more accurately. By using appropriate filtering algorithms, normalization and sampling processing, the working condition characteristics and changing trends can be better captured, thereby improving the accuracy and reliability of working condition monitoring.

[0100] 3. Improve prediction and diagnosis capabilities: Preprocessed data can better adapt to subsequent prediction and diagnosis models; by reducing noise and data anomaly interference, preprocessed data can more accurately reflect the actual working conditions of submersible electric pumps, thereby improving the accuracy and reliability of prediction and diagnosis models.

[0101] It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the claims of this patent.

Claims

1. A method for preprocessing data of a submersible electric pump working pressure sensor, the steps of which are as follows: S1. Receive data: receive initial data from the working pressure sensor; S2, filtering: filtering the received initial data to remove high-frequency noise and interference and retain key fluctuation and change information; S3, normalization processing: normalize the filtered data and scale the data to a specific range to ensure data consistency and comparability; S4, sampling processing: sampling from the linearly normalized data to reduce data dimension and storage requirements while retaining key data features; S5. Data output: The pre-processed data is output to the subsequent pump operating condition analysis and prediction model for further processing and analysis.

2. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 1, characterized in that: In the step S2, a moving average filtering method is used according to the working characteristics of the submersible electric pump working condition.

3. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 2, characterized in that: In step 3, a linear normalization method is used for processing.

4. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 3 is characterized by: In step 4, an equal-interval sampling method is adopted.

5. The method for preprocessing data of the submersible pump operating pressure sensor according to claim 4 is characterized in that: In step 2, the data samples collected by the pressure sensor are processed by moving average filtering to obtain smooth data. Where: filtered_value[i] is the value of the collected data sample after filtering, x[i] is the original data sample collected by the pressure sensor, M is the window length of the filter, The window length M of the filter in formula (1) is: M=K·T Where: M is the window length of the filter, K is the coefficient, T is the time scale, The value range of K is 0.5 to 2; the value range of window length M is 5 to 100.

6. The method for preprocessing data of the submersible pump operating pressure sensor according to claim 5 is characterized in that: In step 3, the values ​​after smoothing filtering are linearly normalized, and the processing process is expressed as follows: Where: normalized_value[i] is the normalized value, filtered_value[i] is the value of the collected data sample after filtering. min_value is the minimum value of the collected data samples. max_value is the maximum value of the collected data samples. After calculation by formula (2), the normalized_value[i] data will be mapped to the range of 0 to 1.

7. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 1, characterized in that: In step 4, the normalized data is subjected to equal-interval sampling, and the specific method is expressed as follows: sampled_value[i]=normalized_value[i·step] (3) Where: sampled_value[i] is the sampled value, i is the index of the data sample, ranging from 0 to N-1, where N is the length of the input data. step is the sampling step size.

8. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 7, characterized in that: In the step 2, The coefficient K is selected based on experience and demand; The time scale T depends on the working characteristics and signal changes of the downhole pressure sensor; The window length M is adjusted according to actual tests.

9. The method for preprocessing data of a submersible electric pump operating pressure sensor according to claim 1, characterized in that: The sampling step length step in the formula (3) is selected according to the data sampling rate and application requirements, and the value range is 20 to 100.

10. The method for preprocessing data of a submersible pump operating pressure sensor according to claim 1, characterized in that: In step S4, key feature data is important information for analyzing and predicting the working condition of the submersible electric pump, which is extracted from the original data by a preprocessing method. The key features include: basic pressure change trend, key fluctuation and change information, normalized value range, and data points sampled at equal intervals.

Images