An intelligent oil pump oil quality monitoring method based on multi-source data fusion

Abstract

The application discloses a kind of intelligent oil pump oil quality monitoring methods based on multi-source data fusion, belong to transformer field, this method includes multi-source data acquisition, data processing, data fusion, result display, data long-distance transmission steps: first through micro water, pressure, flow sensor acquisition oil pump pipeline multidimensional original data and transport to single-chip microcomputer control panel;Again, original data is filtered to obtain standardized data set;Subsequently, using weighted average method and fuzzy logic combined multi-source data fusion algorithm processes data set, constructs oil quality state evaluation model to determine whether oil quality is deteriorated;Then, the monitoring result is locally visualized;Finally, long-distance transmission of monitoring data and local remote data synchronization are realized.The application improves the accuracy of oil quality monitoring through multi-parameter collaborative monitoring and data fusion, realizes the local real-time display and remote real-time monitoring of oil quality state, and adapts to the operation requirements of transformer maintenance site.

Description

Technical Field

[0001] This invention relates to the field of transformer technology, and in particular to an intelligent oil pump oil quality monitoring method based on multi-source data fusion. Background Technology

[0002] As the core equipment for transformer oil injection and drainage, the performance of the oil pump directly affects the efficiency and quality of oil injection and drainage. Existing oil pumps have several problems: high-speed operation of gear pumps can easily generate acetylene and water vapor, threatening insulation performance; excessively high injection pressure may cause pipe rupture and oil spraying from the explosion-proof membrane; excessively high flow rates can easily trigger gas activation and equipment tripping accidents. Current control methods used by maintenance personnel, such as taking oil samples before and after injection, selecting oil pumps with different flow rates, and intermittently starting and stopping the oil pump, are ineffective in practice, increasing manpower and material costs without effectively ensuring the quality of equipment maintenance and safe and stable operation.

[0003] Staff cannot remotely monitor oil quality in real time, resulting in poor flexibility and convenience in on-site operation management. Furthermore, in existing sensor monitoring processes, signals from different sensors are prone to cross-interference, leading to a large amount of abnormal and invalid information in the raw data. The lack of targeted signal conditioning and data preprocessing further reduces the accuracy of oil quality monitoring. Moreover, oil quality monitoring is disconnected from the pressure and flow control of the oil pump, failing to provide data support for the safe operation of the pump. Currently, there is a lack of an intelligent oil pump oil quality monitoring method that combines multi-parameter sensing, data fusion analysis, and local / remote dual-end display, making it difficult to meet the needs of transformer oil injection and drainage operations for accurate, real-time, and intelligent oil quality monitoring. Summary of the Invention

[0004] The purpose of this invention is to address the problems of existing transformer oil pump oil quality monitoring, such as lag, low accuracy of single parameter monitoring, poor data processing, lack of remote real-time monitoring capabilities, and inability to accurately and efficiently control oil quality. Therefore, this invention proposes an intelligent oil pump oil quality monitoring method based on multi-source data fusion.

[0005] To achieve the above objectives, the present invention employs the following technology: an intelligent oil pump oil quality monitoring method based on multi-source data fusion, comprising the following steps: S1. Multi-source data acquisition: The micro water sensor, pressure sensor, and flow sensor are fixedly deployed at the oil injection port, the vibration-free zone in the middle section, and the oil drain port of the transformer oil pump pipeline, respectively. The raw data of micro water content in oil, real-time pipeline pressure, and real-time oil injection and discharge flow are collected synchronously at a sampling frequency of 10Hz, and the raw data are transmitted to the single-chip microcomputer control board for centralized temporary storage in real time. S2. Differentiated Data Processing: The data processing module retrieves the raw data from the microcontroller control board and performs scenario-based differentiated noise reduction filtering: low-pass filtering is used for the analog signal of the micro water sensor, mean filtering is used for the pulse signal of the pressure sensor, and threshold filtering is used for the digital signal of the flow sensor. After noise reduction, the multi-source data is standardized and normalized to obtain a standardized monitoring dataset. S3. Dynamic correction of sensor parameter confidence: Real-time data on oil temperature and vibration frequency during oil pump operation are collected by oil temperature sensor and vibration sensor, and the confidence weight of parameters of micro water, pressure and flow sensor is dynamically adjusted based on oil pump oil temperature and vibration frequency operating data. S4. Multi-source data fusion and oil quality status assessment: A dynamic fusion algorithm combining weighted average method and fuzzy logic is adopted to fuse multi-source parameters after confidence correction. The fused data is input into the oil quality status assessment model, and the oil quality is determined to be qualified, slightly deteriorated or severely deteriorated by the oil quality status assessment model. S5. Oil quality deterioration trend prediction and graded early warning: Time series analysis is performed on the fused data to construct an oil quality deterioration trend prediction model. The graded early warning is triggered by combining the current oil quality status and the predicted trend. If it is determined to be slightly deteriorated, a first-level early warning is triggered. If it is determined to be severely deteriorated, a second-level early warning is triggered. S6. Oil quality abnormality linkage control: When the second-level warning is triggered, the microcontroller control board automatically outputs a control signal to the oil pump drive module to perform flow reduction, power reduction or shutdown operations. S7. Monitoring results are displayed simultaneously on multiple terminals: the oil quality status, deterioration trend, early warning information and control instructions are displayed in real time on the local adaptive display screen and simultaneously transmitted to the remote terminal; S8. Monitoring data feedback optimization: Store full-process data of abnormal events and iteratively optimize the parameters of oil quality assessment model, trend prediction model and confidence correction model through machine learning.

[0006] As a further description of the above technical solution: In step S1, the micro water sensor, pressure sensor and flow sensor all adopt the 485 communication protocol, and the installation distance between the sensors is 40cm to avoid signal cross-interference.

[0007] As a further description of the above technical solution: In step S2, the analog signal of the micro water sensor is filtered with a cutoff frequency of 1kHz to eliminate electromagnetic interference, the pulse signal of the pressure sensor is filtered with a mean value of 5 with a window size to eliminate vibration fluctuations, and the digital signal of the flow sensor is filtered with ±10% threshold to remove abnormal jump data, so as to ensure the effectiveness of single-source monitoring data.

[0008] As a further description of the above technical solution: In step S3, when the oil temperature exceeds the range of 40-60℃, the confidence weight of the micro water sensor is increased; when the vibration frequency is >50Hz, the confidence weight of the pressure sensor is increased.

[0009] As a further description of the above technical solution: In step S5, an oil quality deterioration trend prediction model is constructed using the exponential smoothing method to predict the changes in oil quality status within the next 5-10 minutes. The graded early warning mechanism is as follows: a first-level early warning is triggered when there is slight deterioration and no deterioration trend, and a second-level early warning is triggered when there is slight deterioration and a deterioration trend or severe deterioration.

[0010] As a further description of the above technical solution: In step S6, the control signal is generated specifically according to the type of oil quality abnormality: when the flow rate is too fast, the injection and discharge oil flow rate is reduced; when the pressure is too high, the pressure is released and the oil pump power is reduced; when the oil quality is severely deteriorated, the machine is shut down directly.

[0011] As a further description of the above technical solution: In step S7, the transmission to the remote terminal is implemented based on the 485 communication protocol. The transmitted data packet encapsulates the oil quality fusion analysis results, the original sensor data, and the oil pump operating status parameters to ensure the integrity and real-time performance of the data transmission.

[0012] As a further description of the above technical solution: In step S7, the local display screen is an outdoor adaptive brightness display screen, which can automatically adjust the brightness according to the ambient light.

[0013] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: By fixing and partitioning micro-water, pressure, and flow sensors at key locations in the oil pump pipeline, signal cross-interference is avoided, ensuring the accuracy of raw data acquisition. Differentiated noise reduction filtering is applied to the signal characteristics of different sensors. At the same time, the confidence weight of sensor parameters is dynamically adjusted based on oil pump temperature and vibration frequency operating data, solving the problem of single-source data deviation caused by changes in operating conditions. This makes multi-source data fusion more consistent with the actual operating state of the oil pump, significantly improving the accuracy of oil quality condition determination and effectively avoiding misjudgment and missed judgment caused by single parameter monitoring or fixed weight fusion. By performing time series analysis on the fused data, a model for predicting oil quality deterioration trends is constructed. Combined with the current oil quality status, a graded early warning is triggered. Mild deterioration triggers a level one warning to alert staff, while severe deterioration triggers a level two warning to initiate emergency response. This allows staff to grasp the oil quality deterioration trend in advance, reserving sufficient time for safe handling of transformer oil injection and drainage operations, and reducing the safety risks caused by oil quality deterioration from the source. When a Level 2 severe deterioration warning is triggered, the microcontroller control board can automatically output a control signal to the oil pump drive module. Based on the actual situation, it can perform targeted operations such as reducing flow, reducing power, or shutting down the machine without manual intervention. This quickly stops the continuous deterioration of the oil quality and effectively avoids accidents such as oil pump pipe bursts, gas activation, and equipment tripping caused by oil deterioration. It ensures the safe and stable operation of transformer oil injection and drainage, while reducing the workload of on-site personnel and the error rate of manual control. The local display screen is adapted to different lighting conditions for transformer oil injection and drainage operations, and can refresh monitoring data in real time. On-site staff can intuitively and quickly grasp the key parameters of oil quality and oil pump operation. At the same time, the data processing and display process is fully automated, reducing the workload of on-site staff. Attached Figure Description

[0014] Figure 1 A flowchart according to an embodiment of the present invention is shown. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Reference Figure 1 This embodiment provides a smart oil pump oil quality monitoring method based on multi-source data fusion, which includes the following steps: S1. Multi-source data acquisition: Micro-water sensor, pressure sensor, and flow sensor are fixedly deployed at the oil injection port, vibration-free section, and oil drain port of the transformer oil pump pipeline, respectively. They synchronously collect raw data on the micro-water content in the oil, real-time pipeline pressure, and real-time oil injection / drainage flow rate at a sampling frequency of 10Hz. The raw data is then transmitted in real time to the microcontroller control board for centralized storage. During deployment, the installation spacing between sensors is 40cm to avoid signal cross-interference. The micro-water sensor collects data on the micro-water content in the oil, the pressure sensor collects real-time pipeline pressure data, and the flow sensor collects real-time oil injection / drainage flow rate data. The installation position and signal conditioning circuit of the flow sensor are optimized to improve the stability of flow data acquisition. Each sensor transmits the collected raw data to the microcontroller control board. The micro water sensor, pressure sensor, and flow sensor all use the 485 communication protocol. The microcontroller control board is an STM32F103C8T6 microcontroller control board, which completes the centralized reception and temporary storage of multi-source raw data, providing a foundation for subsequent data processing.

[0017] S2. Differential Data Processing: The data processing module retrieves raw data from the microcontroller control board and performs scenario-specific noise reduction filtering: low-pass filtering with a cutoff frequency of 1kHz is used to eliminate electromagnetic interference for the analog signal of the micro water sensor; mean filtering with a window size of 5 is used to eliminate vibration fluctuations for the pulse signal of the pressure sensor; and ±10% threshold filtering is used to remove abnormal jump data for the digital signal of the flow sensor. Subsequently, the multi-source data is normalized to obtain a standardized monitoring dataset.

[0018] S3. Dynamic correction of sensor parameter confidence: The oil temperature and vibration frequency data of the oil pump are collected in real time by oil temperature sensor and vibration sensor. Based on the oil pump oil temperature and vibration frequency operating condition data, the confidence weight of parameters of micro water, pressure and flow sensor are dynamically adjusted. Specifically, when the oil temperature exceeds the range of 40-60℃, the confidence weight of micro water sensor is increased; when the vibration frequency is >50Hz, the confidence weight of pressure sensor is increased.

[0019] S4. Multi-source data fusion and oil quality status assessment: A dynamic fusion algorithm combining weighted average method and fuzzy logic is adopted to fuse multi-source parameters after confidence correction. The fused data is input into the oil quality status assessment model. Based on preset thresholds (≤50ppm for qualified, 50-80ppm for mild deterioration, and >80ppm for severe deterioration), and with complementary verification of pressure (0.2-0.6MPa) and flow rate (1-3m³ / h) parameters, the oil quality status is accurately determined. The oil quality status assessment model determines whether the oil quality is qualified, mildly deteriorated, or severely deteriorated.

[0020] S5. Oil Quality Deterioration Trend Prediction and Graded Early Warning: Time series analysis was performed on the fused data, and an oil quality deterioration trend prediction model was constructed using the exponential smoothing method to predict changes in oil quality status within the next 5-10 minutes. The graded early warning mechanism is as follows: a first-level early warning is triggered when there is slight deterioration and no deterioration trend, and a second-level early warning is triggered when there is slight deterioration and a deterioration trend or severe deterioration.

[0021] S6. Oil quality anomaly linkage control: When a level 2 warning is triggered, the microcontroller control board automatically outputs a control signal to the oil pump drive module to perform flow reduction, power reduction, or shutdown operations. The control signal is generated specifically according to the type of oil quality abnormality: when the flow rate is too fast, reduce the oil injection / discharge flow rate; when the pressure is too high, release pressure and reduce the oil pump power; and when the oil quality is severely deteriorated, shut down the machine directly.

[0022] S7. Monitoring results are displayed simultaneously on multiple devices: The obtained oil quality fusion analysis results, specific values ​​of trace water content in the oil, real-time pipeline pressure data and flow auxiliary parameters are sent to the LCD1602 display screen through the data transmission link. The display screen is adapted to different lighting operation scenarios such as strong light and weak light at the transformer oil injection and drainage site, and supports real-time data refresh to realize local visualization display of monitoring data. On-site staff can intuitively and quickly grasp the oil quality and core operating parameters of the oil pump through the display screen. If the screen displays an abnormal oil quality warning signal at the same time, the staff can take measures such as stopping the machine and changing the oil on-site as soon as possible to avoid the operational safety hazards caused by oil deterioration. The results of oil quality fusion analysis, raw sensor data and oil pump operating status parameters are transmitted over long distances to achieve remote real-time monitoring of oil quality status, and the remote monitoring data is updated synchronously with the data on the local display screen. Long-distance transmission is achieved based on the 485 communication protocol. The transmitted data packets encapsulate oil quality fusion analysis results, raw sensor data, abnormal early warning signals, and oil pump operating status parameters to ensure the integrity and real-time performance of data transmission.

[0023] S8. Optimization of monitoring data feedback: It stores the entire process data of abnormal events and iteratively optimizes the parameters of the oil quality assessment model, trend prediction model, and confidence correction model through machine learning.

[0024] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for monitoring oil quality in intelligent oil pumps based on multi-source data fusion, characterized in that, Includes the following steps: S1. Multi-source data acquisition: The micro water sensor, pressure sensor, and flow sensor are fixedly deployed at the oil injection port, the vibration-free zone in the middle section, and the oil drain port of the transformer oil pump pipeline, respectively. The raw data of micro water content in oil, real-time pipeline pressure, and real-time oil injection and discharge flow are collected synchronously at a sampling frequency of 10Hz, and the raw data are transmitted to the single-chip microcomputer control board for centralized temporary storage in real time. S2. Differentiated Data Processing: The data processing module retrieves the raw data from the microcontroller control board and performs scenario-based differentiated noise reduction filtering: low-pass filtering is used for the analog signal of the micro water sensor, mean filtering is used for the pulse signal of the pressure sensor, and threshold filtering is used for the digital signal of the flow sensor. After noise reduction, the multi-source data is standardized and normalized to obtain a standardized monitoring dataset. S3. Dynamic correction of sensor parameter confidence: Real-time data on oil temperature and vibration frequency during oil pump operation are collected by oil temperature sensor and vibration sensor, and the confidence weight of parameters of micro water, pressure and flow sensor is dynamically adjusted based on oil pump oil temperature and vibration frequency operating data. S4. Multi-source data fusion and oil quality status assessment: A dynamic fusion algorithm combining weighted average method and fuzzy logic is adopted to fuse multi-source parameters after confidence correction. The fused data is input into the oil quality status assessment model, and the oil quality is determined to be qualified, slightly deteriorated or severely deteriorated by the oil quality status assessment model. S5. Oil quality deterioration trend prediction and graded early warning: Time series analysis is performed on the fused data to construct an oil quality deterioration trend prediction model. The graded early warning is triggered by combining the current oil quality status and the predicted trend. If it is determined to be slightly deteriorated, a first-level early warning is triggered. If it is determined to be severely deteriorated, a second-level early warning is triggered. S6. Oil quality abnormality linkage control: When the second-level warning is triggered, the microcontroller control board automatically outputs a control signal to the oil pump drive module to perform flow reduction, power reduction or shutdown operations. S7. Monitoring results are displayed simultaneously on multiple terminals: the oil quality status, deterioration trend, early warning information and control instructions are displayed in real time on the local adaptive display screen and simultaneously transmitted to the remote terminal; S8. Monitoring data feedback optimization: Store full-process data of abnormal events and iteratively optimize the parameters of oil quality assessment model, trend prediction model and confidence correction model through machine learning.

2. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S1, the micro water sensor, pressure sensor, and flow sensor all use the 485 communication protocol, and the installation spacing between the sensors is 40cm to avoid signal cross-interference.

3. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S2, a low-pass filter with a cutoff frequency of 1kHz is used to eliminate electromagnetic interference for the analog signal of the micro water sensor, a mean filter with a window size of 5 is used to eliminate vibration fluctuations for the pulse signal of the pressure sensor, and a ±10% threshold filter is used to remove abnormal jump data for the digital signal of the flow sensor to ensure the validity of single-source monitoring data.

4. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S3, when the oil temperature exceeds the 40-60℃ range, the confidence weight of the micro-water sensor is increased; when the vibration frequency is greater than 50Hz, the confidence weight of the pressure sensor is increased.

5. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S5, an oil quality deterioration trend prediction model is constructed using the exponential smoothing method to predict changes in oil quality status within the next 5-10 minutes. The graded early warning mechanism is as follows: a first-level early warning is triggered when there is slight deterioration and no deterioration trend, and a second-level early warning is triggered when there is slight deterioration and a deterioration trend or severe deterioration.

6. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S6, the control signal is generated specifically according to the type of oil quality abnormality: when the flow rate is too fast, the injection and discharge oil flow rate is reduced; when the pressure is too high, the pressure is released and the oil pump power is reduced; when the oil quality is severely deteriorated, the machine is shut down directly.

7. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 1, characterized in that, In step S7, the data is transmitted to the remote terminal based on the 485 communication protocol. The transmitted data packet encapsulates the oil quality fusion analysis results, the original sensor data, and the oil pump operating status parameters to ensure the integrity and real-time performance of the data transmission.

8. The intelligent oil pump oil quality monitoring method based on multi-source data fusion according to claim 7, characterized in that, In step S7, the local display screen is an outdoor adaptive brightness display screen, which can automatically adjust its brightness according to the ambient light.

Images