Micro-pump station pressure fluctuation control method and system

Abstract

The application provides a micro-pump station pressure wide-range fluctuation control method and system, and belongs to the technical field of micro-pump station control. The method comprises the following steps: obtaining real-time operation data of a micro-pump station, performing pump station operation state analysis on the real-time operation data, constructing a micro-pump station operation state model, performing pump station system dynamic characteristic analysis based on the micro-pump station operation state model, and obtaining a micro-pump station pressure fluctuation influence factor analysis result. Through the construction of the micro-pump station operation state model and the pressure fluctuation influence factor analysis result, the operation characteristics and the pressure fluctuation reasons of the pump station can be comprehensively and deeply understood, and a solid foundation is provided for subsequent control strategy making.

Description

Technical Field

[0001] This invention proposes a method and system for controlling large-range pressure fluctuations in micro pump stations, belonging to the field of micro pump station control technology. Background Technology

[0002] Miniature pumping stations play a vital role in water supply and drainage systems, and their operational stability directly affects the normal operation of the entire system. In actual operation, miniature pumping stations often face the problem of large pressure fluctuations. These large pressure fluctuations not only affect the normal operation of the pumping station equipment and shorten its lifespan, but may also lead to a decline in water supply or drainage quality and even safety accidents. Existing pumping station pressure control methods are mostly based on simple algorithms such as traditional PID control, which are insufficient to cope with complex and changing operating conditions and various interference factors, and cannot effectively prevent large pressure fluctuations. Therefore, an innovative control method for miniature pumping stations to prevent large pressure fluctuations is urgently needed. Summary of the Invention

[0003] This invention provides a method and system for controlling large-scale pressure fluctuations in micro-pump stations, to solve the problems mentioned in the background section above:

[0004] This invention proposes a method for controlling large-range pressure fluctuations in micro-pump stations, the method comprising:

[0005] S1: Obtain real-time operation data of the micro pump station, analyze the pump station operation status of the real-time operation data, construct a micro pump station operation status model, and conduct dynamic characteristic analysis of the pump station system based on the micro pump station operation status model to obtain the analysis results of the factors affecting the pressure fluctuation of the micro pump station.

[0006] S2: Extract pressure fluctuation features from the inlet and outlet pressure data in the real-time operation data. Use a combination of Fourier transform and wavelet analysis to analyze the frequency components and time-domain characteristics of the pressure signal to obtain pressure fluctuation feature data of the micro-pump station. Based on the pressure fluctuation feature data, use a clustering algorithm to classify the pressure fluctuation patterns, identify normal fluctuation patterns, small-range fluctuation patterns, and large-range fluctuation patterns, and establish a feature library corresponding to different fluctuation patterns. For the large-range fluctuation pattern, further analyze its causes and patterns.

[0007] S3: Based on the analysis results of the factors affecting the pressure fluctuation of the micro pump station and the feature library corresponding to different fluctuation modes, construct a pressure prediction model for the micro pump station; use the pressure prediction model to predict the pressure of the micro pump station in real time and obtain pressure prediction data.

[0008] S4: Obtain the control target data of the micro pump station; based on the pressure prediction data and the control target data, conduct a pressure fluctuation risk assessment, calculate the probability of pressure exceeding the allowable range and the resulting losses, and obtain pressure fluctuation risk assessment data; based on the pressure fluctuation risk assessment data, and combined with fuzzy control theory, formulate a pressure fluctuation control strategy, and determine the control measures to be taken under different risk levels;

[0009] S5: Based on the pressure prediction data and the established pressure fluctuation control strategy, the pumps and valves of the micro pump station are controlled in real time. During the control process, the operating status and pressure changes of the pump station are monitored in real time, and the actual pressure data is compared with the predicted pressure data. The control strategy is dynamically adjusted according to the comparison results. At the same time, the control effect is evaluated. If the pressure fluctuation still exceeds the allowable range, the pressure prediction model and control strategy are further optimized to form a closed-loop control system.

[0010] The present invention proposes a micro pumping station pressure large-range fluctuation prevention control system, including a memory, a processor, and a computer program stored in the memory and capable of running on the memory. The processor executes the program to implement any of the micro pumping station pressure large-range fluctuation prevention control methods described above.

[0011] Beneficial effects of this invention:

[0012] By constructing a micro-pump station operation status model and analyzing the factors affecting pressure fluctuations, we can gain a comprehensive and in-depth understanding of the pump station's operating characteristics and the causes of pressure fluctuations, providing a solid foundation for the formulation of subsequent control strategies.

[0013] A method combining Fourier transform, wavelet analysis, and clustering algorithms is used to extract and classify pressure fluctuation characteristics, which can accurately identify different fluctuation patterns, especially large-scale fluctuation patterns, providing strong support for precise control.

[0014] By using the LSTM algorithm in deep learning to build a stress prediction model, it is possible to predict stress change trends in advance, achieve proactive control, and effectively avoid large-scale stress fluctuations.

[0015] By combining fuzzy control theory to formulate a pressure fluctuation control strategy, and dynamically adjusting and optimizing it based on real-time monitoring data, a closed-loop control system is formed, which improves the accuracy and stability of control and ensures the safe and efficient operation of the micro pump station. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method steps described in this invention;

[0017] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of S2. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0019] One embodiment of the present invention, such as Figure 1 As shown, a method for controlling large-range pressure fluctuations in a micro pumping station includes:

[0020] S1: Obtain real-time operation data of the micro pump station, analyze the pump station operation status of the real-time operation data, construct a micro pump station operation status model, and conduct dynamic characteristic analysis of the pump station system based on the micro pump station operation status model to obtain the analysis results of the factors affecting the pressure fluctuation of the micro pump station.

[0021] S2: Extract pressure fluctuation features from the inlet and outlet pressure data in the real-time operation data. Use a combination of Fourier transform and wavelet analysis to analyze the frequency components and time-domain characteristics of the pressure signal to obtain pressure fluctuation feature data of the micro-pump station. Based on the pressure fluctuation feature data, use a clustering algorithm to classify the pressure fluctuation patterns, identify normal fluctuation patterns, small-range fluctuation patterns, and large-range fluctuation patterns, and establish a feature library corresponding to different fluctuation patterns. For the large-range fluctuation pattern, further analyze its causes and patterns.

[0022] S3: Based on the analysis results of the factors affecting the pressure fluctuation of the micro pump station and the feature library corresponding to different fluctuation modes, construct a pressure prediction model for the micro pump station; use the pressure prediction model to predict the pressure of the micro pump station in real time and obtain pressure prediction data.

[0023] S4: Obtain the control target data of the micro pump station; based on the pressure prediction data and the control target data, conduct a pressure fluctuation risk assessment, calculate the probability of pressure exceeding the allowable range and the possible losses, and obtain pressure fluctuation risk assessment data; based on the pressure fluctuation risk assessment data, combined with fuzzy control theory, formulate a pressure fluctuation control strategy, and determine the control measures to be taken under different risk levels.

[0024] S5: Based on the pressure prediction data and the established pressure fluctuation control strategy, the pumps and valves of the micro pump station are controlled in real time. During the control process, the operating status and pressure changes of the pump station are monitored in real time, and the actual pressure data is compared with the predicted pressure data. The control strategy is dynamically adjusted according to the comparison results. At the same time, the control effect is evaluated. If the pressure fluctuation still exceeds the allowable range, the pressure prediction model and control strategy are further optimized to form a closed-loop control system.

[0025] The working principle of the above technical solution is as follows:

[0026] Key operational data of the micro-pump station, such as inlet and outlet water pressure, pump operating frequency, pipeline flow rate, and valve opening, are collected in real time by sensors. Based on the collected real-time operational data, an operational status model of the micro-pump station is constructed. This model can intuitively reflect the dynamic relationship between various parameters within the pump station, providing a foundation for subsequent pressure fluctuation analysis and prediction.

[0027] A combination of Fourier transform and wavelet analysis was used to extract pressure fluctuation features from inlet and outlet pressure data. Fourier transform was used to analyze the frequency components of the pressure signal, while wavelet analysis was used to extract time-domain features such as abrupt changes, fluctuation amplitude, and duration. Clustering algorithms were employed to classify the extracted pressure fluctuation features, identifying normal fluctuation patterns, small-range fluctuation patterns, and large-range fluctuation patterns, and a feature library corresponding to each fluctuation pattern was established. For large-range fluctuation patterns, the causes and patterns were further analyzed to provide a basis for precise control of large-range pressure fluctuations.

[0028] Based on the analysis of factors influencing pressure fluctuations in micro-pump stations and a feature library corresponding to different fluctuation patterns, a pressure prediction model for micro-pump stations is constructed using the Long Short-Term Memory (LSTM) algorithm in deep learning. The model is trained with a large amount of historical operating data to accurately predict pressure change trends over a future period. The trained pressure prediction model is then used to predict the pressure of the micro-pump station in real time, obtaining pressure prediction data. Based on the predicted data, it is possible to determine in advance whether large-scale pressure fluctuations are likely, providing a basis for subsequent control strategy formulation.

[0029] Acquire control target data for the micro-pump station, including the desired outlet pressure range, allowable pressure fluctuation range, and system response time requirements. Based on pressure prediction data and control target data, conduct a pressure fluctuation risk assessment, calculate the probability of pressure exceeding the allowable range and the potential losses, and obtain pressure fluctuation risk assessment data. Based on the pressure fluctuation risk assessment data and combined with fuzzy control theory, formulate a pressure fluctuation control strategy. Determine the control measures to be taken under different risk levels, such as adjusting the pump operating frequency and regulating valve opening, to achieve effective control of pressure fluctuations.

[0030] Based on pressure prediction data and a formulated pressure fluctuation control strategy, the pumps and valves of the micro-pump station are controlled in real time. During the control process, the operating status and pressure changes of the pump station are monitored in real time. The actual pressure data is compared with the predicted pressure data, and the control strategy is dynamically adjusted based on the comparison results. The control effect is evaluated, and if the pressure fluctuation still exceeds the allowable range, the pressure prediction model and control strategy are further optimized. Through continuous adjustment and optimization, a closed-loop control system is formed, continuously improving control performance and ensuring that the pressure fluctuation of the micro-pump station remains within the allowable range.

[0031] The effects of the above technical solution are as follows:

[0032] By analyzing pump station status, extracting pressure fluctuation features, and using deep learning to predict based on real-time operational data, pressure fluctuation trends can be accurately predicted, allowing for proactive control measures to prevent pressure fluctuations from exceeding permissible limits.

[0033] By implementing pressure fluctuation risk assessment and control strategies, the probability of abnormal pressure fluctuations in pumping stations has been effectively reduced, thereby reducing the risk of equipment damage and system failure.

[0034] Based on real-time predictive data and control strategies, the pump frequency and valve opening can be flexibly adjusted under different risk levels to ensure rapid response of the pumping station and stable operation of the system when facing pressure fluctuations.

[0035] By dynamically adjusting the control strategy and optimizing the pressure prediction model, the system can adapt to fluctuation patterns under different operating conditions, ensuring continuous optimization and achieving the best control effect.

[0036] By optimizing the pump station's operating mode to avoid unnecessary large fluctuations, the frequency of excessive adjustment of pump frequency and valve opening was reduced, thereby improving the pump station's energy utilization efficiency.

[0037] By conducting in-depth analysis and prediction of pump station pressure fluctuations, not only is an early warning mechanism provided for daily operation, but also data support is provided for long-term maintenance and management, reducing maintenance costs and management difficulty.

[0038] In one embodiment of the present invention, S1 includes:

[0039] S11. Install high-precision sensors at key locations in the micro pump station, and collect real-time operating data of the pump station in real time according to the set sampling frequency through the data acquisition system, and transmit the collected real-time operating data to the data processing center.

[0040] S12. Clean the collected real-time running data to remove outliers and noisy data; and normalize the data to unify data of different dimensions into the range of [0,1] or [-1,1].

[0041] S13. Using system identification methods, the preprocessed data is processed to construct a micro pump station operation status model;

[0042] S14. Based on the constructed pump station operation status model, conduct simulation experiments to simulate the operation of the pump station under different working conditions; through parameter sensitivity analysis, determine the key factors affecting pressure fluctuations and their mechanisms of action; and obtain the analysis results of the factors affecting pressure fluctuations in micro pump stations.

[0043] The working principle of the above technical solution is as follows:

[0044] High-precision sensors are installed at key locations in the micro-pump station (such as inlets, outlets, pumps, pipes, and valves). These sensors accurately detect various physical quantities during the pump station's operation, including pressure, flow rate, frequency, and opening degree. By strategically placing the sensors, comprehensive and accurate information about the pump station's operating status is ensured. The data acquisition system collects data measured by each sensor in real time according to a set sampling frequency (e.g., once per second), forming a real-time operational dataset. The selection of the sampling frequency must comprehensively consider data accuracy and system processing capabilities to ensure the real-time nature and validity of the data. The collected data is transmitted to the data processing center via wired or wireless communication, providing a foundation for subsequent data processing and analysis.

[0045] Because sensor measurements may be affected by various interference factors, the collected data may contain outliers and noisy data. Data cleaning improves data quality and reliability by identifying and removing these outliers. For example, for values ​​significantly outside the normal range, mean filtering or median filtering is used for correction. Mean filtering replaces outliers by calculating the average value of data within a certain window, while median filtering uses the median value of the data within the window as the replacement value. Both methods effectively reduce noise interference and make the data smoother. Pump station operating status involves data with various units, such as pressure in Pascals (Pa) and flow rate in cubic meters per second (m³ / s). 3 The frequency unit is Hertz (Hz), etc. To facilitate subsequent model building and data analysis, the data needs to be normalized to unify data with different dimensions into the range of [0,1] or [-1,1]. Normalization eliminates the influence of data dimensions, making different parameters comparable in the model, which helps improve the model's training effect and generalization ability.

[0046] System identification methods (such as least squares or maximum likelihood estimation) are used to process the preprocessed data. The purpose of system identification is to determine the mathematical model of the system based on the input and output data. Least squares estimates model parameters by minimizing the sum of squared errors, thus minimizing the error between the model output and the actual output; maximum likelihood estimation, based on probability and statistics, seeks model parameters that maximize the probability of the observed data. These methods can extract the dynamic characteristics of the system from large amounts of data, providing a basis for constructing an accurate operating state model. The constructed operating state model of the micro-pump station can take the form of a state-space model or a transfer function model. The state-space model can describe the internal state and dynamic behavior of the system and is suitable for multi-input multi-output systems; the transfer function model describes the dynamic characteristics of the system through the relationship between inputs and outputs, and its form is simple and intuitive. Both models can intuitively reflect the dynamic relationships between various parameters within the pump station, providing a mathematical foundation for subsequent simulation experiments and pressure fluctuation analysis.

[0047] Based on the constructed pump station operation state model, simulation experiments were conducted. By simulating the pump station's operation under different conditions, such as sudden pump start-up and shutdown, and sudden changes in inlet water pressure, the dynamic response of various parameters was observed. Simulation experiments can predict and assess the pump station's operation before actual operation, providing a reference for formulating reasonable operation strategies. Furthermore, simulation experiments can be repeated, facilitating comprehensive analysis of different operating conditions. Through parameter sensitivity analysis, the key factors affecting pressure fluctuations and their mechanisms of action were identified. For example, the impact of pump frequency changes on outlet water pressure and the impact of valve opening adjustments on pipeline flow were analyzed. Parameter sensitivity analysis can identify parameters with significant impact on pressure fluctuations, providing key areas of focus for subsequent pressure fluctuation control. By deeply studying the mechanisms of these key factors, more effective control strategies can be developed to reduce the amplitude of pressure fluctuations and improve the operational stability and reliability of the pump station.

[0048] The effects of the above technical solution are as follows:

[0049] By installing high-precision sensors at key locations and collecting data in real time, the operation data of the pumping station can be comprehensively and accurately reflected to reflect the real-time status of the pumping station, providing reliable data support for subsequent analysis and control.

[0050] By cleaning and filtering the collected data, outliers and noise were removed, ensuring the quality and validity of the data and thus avoiding the impact of data errors on model building and prediction results.

[0051] Data normalization enables data of different dimensions to be unified to the same scale range, thereby improving data comparability and providing a unified input for subsequent modeling and analysis.

[0052] By using system identification methods (such as least squares or maximum likelihood estimation) to model the preprocessed data, the dynamic relationships between various parameters in the pumping station can be accurately captured, providing a precise model for subsequent dynamic characteristic analysis and prediction.

[0053] By using state-space models or transfer function models, the representation of the pump station's operating state model is simplified, and the relationships between various parameters can be intuitively reflected, making the model easy to understand and analyze.

[0054] Simulation experiments based on the constructed operating state model can simulate various operating conditions and parameter changes, helping to deeply analyze the dynamic response and pressure fluctuation of the pump station under different operating conditions, thus providing sufficient data support and theoretical basis for subsequent fluctuation control.

[0055] By using parameter sensitivity analysis, we can identify the key factors affecting pressure fluctuations and understand their mechanisms of action. This provides a scientific basis for developing precise pressure fluctuation control strategies and avoids ineffective or excessive control interventions.

[0056] One embodiment of the present invention, such as Figure 2 As shown, S2 includes:

[0057] S21. Perform Fourier transform on the inlet and outlet pressure data in the real-time operation data to convert them from the time domain to the frequency domain and analyze the frequency components of the pressure signal.

[0058] S22. Wavelet analysis is used to decompose the pressure signal into multiple scales and extract time-domain features at different scales; thus obtaining pressure fluctuation characteristic data of micro-pump stations.

[0059] S23. Use the K-means clustering algorithm to perform cluster analysis on the pressure fluctuation feature data; calculate the distance from each pressure fluctuation feature data point to each cluster center, and assign it to the category corresponding to the nearest cluster center;

[0060] S24. Update the cluster centers. Repeat the above allocation and update process until the cluster centers no longer change or the set number of iterations is reached, thereby identifying different fluctuation patterns and establishing a feature library corresponding to different fluctuation patterns.

[0061] S25. For large-scale fluctuation patterns, further analyze the causes and patterns of their occurrence; and establish a causal relationship model for large-scale fluctuation patterns.

[0062] The working principle of the above technical solution is as follows:

[0063] The Fourier transform is a mathematical tool for converting time-domain signals into frequency-domain signals. Performing a Fourier transform on inlet and outlet pressure data converts the pressure signal from a time-varying representation to a frequency-varying representation. In the frequency domain, the various frequency components of the pressure signal can be clearly seen. By analyzing the frequency domain signal, the dominant frequency components and their amplitudes in the pressure signal can be determined. The dominant frequency components reflect the periodic variations that predominate in pressure fluctuations, while the amplitude represents the intensity of that frequency component. Identifying the frequency characteristics associated with pressure fluctuations helps in understanding the inherent patterns of pressure fluctuations; for example, certain frequency components may be related to specific pump station operating conditions or external disturbances.

[0064] Wavelet analysis possesses multi-scale analysis capabilities, allowing signals to be decomposed into different scales. Multi-scale decomposition of a pressure signal is equivalent to analyzing the signal at different frequency ranges and resolutions. Sub-band signals at each scale reflect the characteristics of the original signal at different scales, thus enabling a more comprehensive capture of the pressure signal's details. Within the sub-band signals at each scale, the modulus maxima properties of wavelet transform can be used to detect temporal features in the pressure signal, such as abrupt changes, fluctuation amplitudes, and durations. Abrupt changes typically correspond to abnormal changes in the pressure signal, such as a sharp rise or fall in pressure; fluctuation amplitudes reflect the severity of pressure changes; and duration indicates the length of time pressure fluctuations last. These temporal features provide crucial information for subsequent pressure fluctuation pattern recognition.

[0065] K-means clustering is a commonly used unsupervised learning algorithm for classifying data points into different categories. Before starting clustering, the number of clusters needs to be determined; in this embodiment, three categories are set: normal fluctuation pattern, small-range fluctuation pattern, and large-range fluctuation pattern. Then, cluster centers are randomly initialized; each cluster center is a representative point of its category. The distance from each pressure fluctuation feature data point to each cluster center is calculated; common distance metrics include Euclidean distance. Each data point is assigned to the category corresponding to the nearest cluster center, thus initially classifying all pressure fluctuation feature data into different categories.

[0066] After data point allocation, the mean of all data points in each category is recalculated, and this mean is used as the new cluster center. The purpose of this step is to make the cluster centers closer to the actual center location of the category. This process of data point allocation and cluster center updating is repeated until the cluster centers no longer change or the set number of iterations is reached. At this point, the clustering results can be considered converged, and each data point has been reasonably assigned to its corresponding category, thus identifying different fluctuation patterns. Simultaneously, the pressure fluctuation characteristic data in each category is stored to establish a feature library corresponding to different fluctuation patterns, providing a reference for subsequent pressure fluctuation analysis and control.

[0067] To address large-scale fluctuation patterns, this study identifies common factors causing these fluctuations by comparing data under different operating conditions. For example, it observes the characteristics and trends of pressure fluctuation data under various conditions such as sudden drops in inlet pressure and pump failure, analyzing the correlation between these factors and large-scale fluctuations. Based on the data comparison analysis, a causal model for large-scale fluctuation patterns is established. This model describes the causal relationship between factors causing large-scale fluctuations and pressure fluctuations; for example, a sudden drop in inlet pressure may lead to a sharp decrease in pressure, and pump failure may cause pressure instability. This causal model allows for a deeper understanding of the mechanisms underlying large-scale fluctuation patterns, providing a basis for developing targeted control strategies.

[0068] The effects of the above technical solution are as follows:

[0069] By converting the inlet and outlet pressure data from the time domain to the frequency domain using Fourier transform, the frequency components of the pressure signal can be analyzed more accurately, the frequency characteristics related to pressure fluctuations can be identified, and the accuracy of fluctuation analysis can be improved.

[0070] By employing wavelet analysis to decompose pressure signals at multiple scales, time-domain features at different scales can be effectively extracted, abrupt changes and fluctuation amplitudes in pressure signals can be detected, noise interference with analysis results can be reduced, and the extraction effect of fluctuation features can be improved.

[0071] By using the K-means clustering algorithm to perform cluster analysis on pressure fluctuation characteristic data, different fluctuation patterns, such as normal fluctuations, small-range fluctuations, and large-range fluctuations, can be quickly and effectively identified, helping the system to better identify and classify different fluctuation types.

[0072] By iteratively updating the cluster centers, the stability of the clustering results is ensured, and a reliable feature library of fluctuation patterns can be obtained through multiple iterations, providing accurate data support for subsequent fluctuation control and analysis.

[0073] The automated clustering analysis process reduces the need for manual intervention, enables automatic identification of fluctuation patterns, and allows for intelligent decision-making based on data, thereby improving the system's automation level.

[0074] Further analysis of large-scale fluctuation patterns can identify their causes and patterns, helping to identify common factors that may lead to large-scale fluctuations, such as a sudden drop in inlet water pressure or water pump failure, thus providing important preventive measures to prevent large-scale fluctuations.

[0075] By establishing a causal relationship model for large-scale fluctuation patterns, we can more accurately understand the key factors that lead to large-scale fluctuations, providing a theoretical basis for formulating more scientific and effective fluctuation control strategies and optimizing the operation and management of pumping stations.

[0076] In one embodiment of the present invention, step S21 includes:

[0077] The inlet and outlet pressure data are preprocessed by Fourier transform. The preprocessed inlet and outlet pressure data are then transformed from the time domain to the frequency domain to obtain the pressure signal in the frequency domain.

[0078] In the frequency domain, the frequency components of the pressure signal are analyzed, and the amplitude of the frequency components is calculated; wherein the amplitude of the frequency components is obtained by the following formula:

[0079]

[0080] in,

[0081] Where N represents the sampling points of the pressure signal; k represents the frequency index; X k Represents the frequency domain pressure signal sequence and the DFT result; Re(X) k ) represents X k The real part; Re(X) k ) represents X k The imaginary part;

[0082] Based on the frequency component analysis results, frequency characteristics related to pressure fluctuations are identified, and the obtained frequency component data and frequency characteristics related to pressure fluctuations are organized and stored.

[0083] The working principle of the above technical solution is as follows:

[0084] Inlet and outlet pressure data may come from sensors with different ranges or accuracies, resulting in variations in their numerical ranges and units. Data standardization involves scaling these data according to certain rules to achieve a uniform scale. For example, Z-score standardization involves subtracting the mean from the data and then dividing by the standard deviation, resulting in a mean of 0 and a standard deviation of 1. Data standardization eliminates the influence of different dimensions, allowing Fourier transforms to more accurately analyze the frequency characteristics of the data and avoiding analytical biases caused by inconsistent data scales. During data acquisition, various noise interferences are unavoidable, such as sensor noise and electromagnetic interference. These noises can mask the true characteristics of the pressure signal, affecting the results of the Fourier transform. Various noise removal methods exist, commonly including low-pass filtering, high-pass filtering, and band-pass filtering. Low-pass filtering removes high-frequency noise while retaining low-frequency pressure signal components; high-pass filtering, conversely, removes low-frequency noise; and band-pass filtering selectively retains signals within a specific frequency range. By removing noise, the signal-to-noise ratio of the pressure signal can be improved, enabling the Fourier transform to more clearly identify the frequency components in the pressure signal.

[0085] Fourier transform operations are performed on the preprocessed influent and effluent pressure data. The basic principle of Fourier transform is to decompose a time-domain signal into a linear combination of sine and cosine functions of different frequencies. In the discrete case, the Fast Fourier Transform (FFT) algorithm is typically used, which can efficiently calculate the frequency domain representation of the discrete time series. Through Fourier transform, the pressure signal, which originally varies with time, is converted into a frequency-domain signal that varies with frequency, allowing us to analyze the characteristics of the pressure signal from a frequency perspective. After Fourier transform, the pressure signal in the frequency domain is obtained. The frequency domain signal, with frequency as the abscissa and amplitude or phase as the ordinate, clearly shows the distribution of the pressure signal at different frequencies. The frequency domain signal provides an intuitive representation for analyzing the frequency components of the pressure signal, facilitating subsequent frequency analysis and feature extraction.

[0086] In the frequency domain, the dominant frequency components of a pressure signal can be identified by observing the spectrum. These dominant frequency components are those with large amplitudes in the spectrum, representing the predominant periodic variations in the pressure signal. For example, if a significant peak appears near a certain frequency point in the spectrum, then that frequency point is a dominant frequency component of the pressure signal. For each identified dominant frequency component, its amplitude needs to be calculated. The amplitude reflects the strength of that frequency component in the pressure signal; a larger amplitude indicates a greater influence of that frequency component on pressure fluctuations. By calculating the amplitude of frequency components, the contribution of different frequency components to the pressure signal can be quantified, providing a basis for subsequent frequency feature extraction.

[0087] Based on the frequency component analysis results, frequency characteristics related to pressure fluctuations are identified. For example, by comparing the spectrum diagrams under normal and abnormal operating conditions, it can be found that the amplitude of certain frequency ranges or specific frequency points increases significantly or new peaks appear under abnormal operating conditions. These frequency ranges or specific frequency points may be related to pressure fluctuations. Identifying these frequency characteristics helps us understand the generation mechanism of pressure fluctuations; for example, some frequency characteristics may be related to the pump's operating frequency, the pipeline's resonant frequency, etc. The analyzed frequency component data, including the main frequency components and their amplitudes, as well as the frequency characteristics related to pressure fluctuations, are organized and stored. This data is an important basis for subsequent pressure fluctuation analysis, pattern recognition, and control strategy formulation. By organizing and storing this data, it can be easily queried, analyzed, and utilized, providing support for the stable operation of micro pump stations.

[0088] The effects of the above technical solution are as follows:

[0089] By standardizing and removing noise from the inlet and outlet pressure data, the accuracy of the Fourier transform was ensured, thereby improving the reliability of the frequency domain analysis results.

[0090] By performing preprocessing operations such as noise removal, interference components in the original data are effectively reduced, the clarity of the frequency domain signal after Fourier transform is enhanced, and frequency component analysis becomes more accurate.

[0091] By analyzing the frequency components of the pressure signal in the frequency domain, the main frequency components can be identified and their amplitudes calculated, which further improves the understanding of the causes of pressure fluctuations and provides richer evidence for fluctuation control.

[0092] By identifying frequency characteristics related to pressure fluctuations based on frequency component analysis results, the main frequency bands or specific frequency points causing pressure fluctuations can be accurately identified, ensuring precise location of the fluctuation source and reducing misjudgments.

[0093] By organizing and storing the frequency component data, pressure fluctuation characteristics can be archived for a long time, facilitating subsequent querying and analysis, thus improving the operability and traceability of the data.

[0094] By automating processing and frequency component analysis, the reliance on manual analysis is reduced, enabling the system to efficiently and automatically identify and classify pressure fluctuations during long-term operation, thus optimizing management efficiency.

[0095] The organized and stored frequency component data provides accurate foundational data for subsequent fluctuation pattern analysis, prediction, and control strategies, ensuring the scientific validity and feasibility of fluctuation control methods.

[0096] The Fourier transform provides an efficient way to convert time-domain signals into frequency-domain representations, which can help identify periodic variations and frequency characteristics in signals, and has significant advantages, especially in analyzing periodic fluctuations or noise.

[0097] By calculating the amplitude of frequency components, the intensity of different frequencies in a signal can be accurately captured, and which frequency components dominate the entire signal can be identified. This is particularly useful for inlet and outlet water pressure signals, revealing frequency fluctuations that affect system operation.

[0098] By analyzing the amplitude of frequency components, it is easier to identify frequency characteristics related to system pressure fluctuations. This allows for the early detection of potential anomalies or faults, such as equipment vibration or system instability.

[0099] Organizing and storing the analysis results of frequency components not only helps with subsequent data review and analysis, but also serves as historical data on the system's health status, helping to analyze long-term trends and potential problems in pressure fluctuation patterns.

[0100] By processing and analyzing pressure data through Fourier transform, the complex calculations and processing of time-domain signals can be effectively reduced, improving processing efficiency and making real-time monitoring and analysis more accurate and efficient.

[0101] In one embodiment of the present invention, step S22 includes:

[0102] S221. Using wavelet basis functions, perform multi-scale wavelet decomposition on the pressure signal; by progressively decomposing the signal to different scales, obtain a series of sub-band signals with different frequency ranges; and each sub-band signal reflects the characteristics of the original signal at different scales.

[0103] S222. In the sub-band signals at each scale, the modulus maxima characteristic of wavelet transform is used to detect abrupt changes in the pressure signal; for each sub-band signal at each scale, its fluctuation amplitude is calculated; wherein, the fluctuation amplitude is calculated using the following formula:

[0104]

[0105] Where p(t) represents the original pressure signal; j represents the wavelet decomposition scale; W j p(t) represents the wavelet coefficients at scale j; |W j p(t)| represents the modulus of the wavelet coefficients; λ represents the time derivative of the wavelet coefficients. j α represents the scale-dependent attenuation coefficient, with units of s / Pa; j P0 represents the local variance influence factor; P0 represents the reference pressure, such as the mean pressure. This indicates the local variance centered at t with a scale j smaller than t; and,

[0106]

[0107] Where Δt represents the sampling time interval; F represents the local window half-width. This represents a local average value.

[0108] S223. Obtain the time of occurrence of each mutation point and the duration of its corresponding fluctuation amplitude; by analyzing the duration, obtain the stability and trend of pressure fluctuations;

[0109] S224. Integrate the detected mutation point information, the calculated fluctuation amplitude data, and the duration analysis results to form micro-pump station pressure fluctuation characteristic data.

[0110] The working principle of the above technical solution is as follows:

[0111] Wavelet basis functions are the core of wavelet analysis. They possess localization properties, enabling simultaneous analysis of signals in both time and frequency domains. Using wavelet basis functions to perform multi-scale wavelet decomposition on pressure signals is equivalent to observing the signal through a set of "windows" at different scales. Different wavelet basis functions have different shapes and characteristics, suitable for different types of signal analysis. In practical applications, appropriate wavelet basis functions, such as Daubechies wavelets and Symlets wavelets, can be selected based on the characteristics of the pressure signal. By progressively decomposing the signal to different scales, a series of sub-band signals with different frequency ranges are obtained. During the decomposition process, the signal is decomposed into approximate signals (low-frequency components) and detail signals (high-frequency components). The approximate signal contains the main trend and low-frequency components of the signal, while the detail signal contains high-frequency details and abrupt changes. As the decomposition scale increases, the frequency ranges of the approximate and detail signals gradually narrow, thus achieving multi-resolution analysis of the signal. Each sub-band signal reflects the characteristics of the original signal at different scales. For example, at a coarse scale, the sub-band signal mainly reflects the overall trend of the signal; at a fine scale, the sub-band signal can capture the local details and abrupt changes of the signal.

[0112] Wavelet transform possesses a modulus maxima characteristic; at abrupt changes in the signal, the modulus of the wavelet transform will reach a maxima. Utilizing this characteristic, abrupt changes in pressure signals can be detected by searching for these modulus maxima in sub-band signals at various scales. These abrupt changes typically correspond to abnormal changes in the pressure signal, such as a sharp rise or fall in pressure. These abrupt changes may reflect abnormal conditions during pump station operation, such as pump failure or sudden valve opening and closing. By detecting these abrupt changes, anomalies in the pressure signal can be identified in a timely manner, providing crucial information for subsequent analysis and control. For each sub-band signal at each scale, its fluctuation amplitude is calculated. The fluctuation amplitude can be measured by calculating the standard deviation, range, or other statistical measures of the sub-band signal. The standard deviation reflects the dispersion of the signal data relative to the mean, while the range is the difference between the maximum and minimum values ​​of the signal. Fluctuation amplitude information at different scales helps to comprehensively understand the fluctuation characteristics of the pressure signal. For example, at a fine scale, a large fluctuation amplitude may indicate the presence of high-frequency, drastic fluctuations in the signal; at a coarse scale, a large fluctuation amplitude may indicate the presence of low-frequency, large-amplitude changes in the signal. By analyzing the fluctuation amplitude at different scales, the fluctuation of pressure signals can be grasped more accurately.

[0113] After detecting abrupt changes, the time of occurrence of each abrupt change is acquired. This can be achieved by recording the time position corresponding to the maximum value of the wavelet transform modulus. The timing information of abrupt changes is crucial for analyzing the triggering moment and sequence of events in pressure fluctuations; for example, it can help determine whether a pump failure or valve adjustment caused the pressure surge. Simultaneously, the duration of the fluctuation amplitude corresponding to each abrupt change is acquired. The duration of the fluctuation amplitude reflects the length of time the pressure fluctuation lasts from start to finish. By analyzing the duration, the stability and trend of pressure fluctuations can be obtained. For example, some pressure fluctuations may be transient, such as instantaneous pressure changes caused by external disturbances; while others may last longer, such as continuous pressure fluctuations caused by a decline in pump performance. Duration analysis helps determine the nature and severity of pressure fluctuations, providing a basis for taking appropriate control measures.

[0114] The detected abrupt change information (including its occurrence time), calculated fluctuation amplitude data, and duration analysis results are integrated. This information describes the fluctuation characteristics of the pressure signal from different perspectives, and their integration comprehensively reflects the dynamic characteristics of pressure fluctuations. For example, the occurrence time and fluctuation amplitude of the abrupt change point reveal the triggering time and intensity of the pressure fluctuation; the duration reveals the stability and trend of the pressure fluctuation. The integrated information forms the pressure fluctuation characteristic data for the micro-pump station. This characteristic data is a crucial foundation for subsequent pressure fluctuation pattern recognition, risk assessment, and control strategy development. Through the analysis and processing of this characteristic data, a deeper understanding of the inherent laws governing pressure fluctuations can be achieved, ensuring the stable operation of the micro-pump station.

[0115] The effects of the above technical solution are as follows:

[0116] By performing multi-scale analysis of pressure signals through wavelet decomposition, information can be extracted from sub-band signals in different frequency ranges, thus providing a comprehensive and detailed analysis of signal fluctuations and enhancing the multi-dimensional identification of fluctuation characteristics.

[0117] By leveraging the modulus maxima characteristic of wavelet transform, abrupt changes in pressure signals can be accurately detected, ensuring timely identification of abnormal changes, such as sharp increases or decreases in pressure, thus enhancing the ability to respond to abnormal fluctuations.

[0118] By calculating the fluctuation amplitude of each sub-band signal, the intensity of signal fluctuations at different scales can be quantitatively measured, providing more accurate fluctuation amplitude data and contributing to a deeper understanding of pressure signal fluctuations.

[0119] By using wavelet decomposition, the bias of traditional methods towards a certain frequency range or scale is avoided, thus enabling a more comprehensive and accurate capture of the fluctuation characteristics of pressure signals at different scales and reducing the possibility of missing local fluctuation features.

[0120] By analyzing the duration of abrupt change points, we can assess the stability and trend of pressure fluctuations, helping to distinguish between short-term and long-term fluctuations, thus providing more accurate information for the prevention and control of pressure fluctuations.

[0121] By integrating information on mutation points, fluctuation amplitude data, and duration analysis results into pressure fluctuation characteristic data, consistent and systematic data support can be provided for subsequent control strategies and optimization schemes, thereby improving the efficiency and accuracy of decision-making.

[0122] By automating wavelet decomposition and fluctuation amplitude analysis, the need for manual intervention and calculation is reduced, enabling the system to perform pressure fluctuation analysis and feature extraction efficiently and automatically, thus improving the ease of operation and accuracy.

[0123] Multi-scale decomposition of wavelet basis functions can decompose a signal into multiple sub-band signals with different frequency ranges, each reflecting the variation characteristics of the original signal at different time scales. This method can effectively handle non-stationary signals and reveal detailed information about the signal at different scales.

[0124] By utilizing the modulus maxima property of wavelet transform, abrupt changes in signals can be accurately detected. These abrupt changes typically represent abnormal conditions in the system, such as drastic pressure fluctuations or equipment failures. Early detection of these abrupt changes is crucial for fault warning and system optimization.

[0125] By calculating the wavelet coefficient modulus and fluctuation amplitude at each scale, the fluctuation intensity of the signal at different scales can be quantified. Combined with factors such as attenuation coefficient and local variance, the sensitivity and accuracy of signal characteristics are further improved.

[0126] By obtaining the duration of abrupt change points and combining it with fluctuation amplitude data, the stability and trend of pressure fluctuations can be analyzed. This helps identify regular fluctuations or instabilities during system operation, providing decision support for subsequent system adjustments and optimizations.

[0127] By integrating information on abrupt change points, fluctuation amplitude, and duration, comprehensive characteristic data of pressure fluctuations in micro-pump stations can be generated. This data not only reflects the dynamic characteristics of pressure fluctuations but also serves as a quantitative indicator of system health, providing crucial reference for subsequent performance evaluation, fault diagnosis, and maintenance.

[0128] In one embodiment of the present invention, S224 includes:

[0129] Information on abrupt changes detected in sub-band signals at various scales is organized and stored in a unified data format. The calculated fluctuation amplitude data is then normalized.

[0130] For each mutation point, the fluctuation amplitude and duration are encoded; based on the length of the duration, it is divided into different time intervals, and each interval is assigned a corresponding encoding value.

[0131] Using a database table format, the organized mutation point information is used as the primary key, the normalized fluctuation amplitude data and the encoded duration data are used as fields to establish data relationships; based on the time location of the mutation point, the corresponding fluctuation amplitude and duration information are integrated together to form a complete feature dataset for each mutation point.

[0132] The abrupt change point feature data corresponding to sub-band signals at each scale are fused to obtain comprehensive fluctuation amplitude and duration features.

[0133] The working principle of the above technical solution is as follows:

[0134] The abrupt changes detected in subband signals at various scales contain rich information about pressure fluctuations. Organizing this information and recording its temporal location helps clarify the specific moment of the abnormal pressure change, facilitating subsequent analysis of the sequence and causal relationships of events. Recording the corresponding subband signal scale is also crucial, as different scales reflect the characteristics of the original signal at different resolutions, helping us understand the performance of pressure fluctuations across different frequency ranges. Recording the amplitude changes at the abrupt changes—the magnitude of the sharp rise or fall in pressure—directly reflects the intensity of the abnormal pressure change and is essential for assessing the severity of pressure fluctuations. Storing the data in a unified format, such as a table, makes the data more standardized and structured. Tables facilitate data viewing, management, and querying. Different types of information (time location, subband signal scale, amplitude change) are stored in different columns, with each row representing information about a single abrupt change. This allows for convenient extraction and processing of this data during subsequent data processing and analysis.

[0135] The amplitude of subband signals at different scales can differ by orders of magnitude. For example, fine-scale subband signals may capture small high-frequency fluctuations with relatively small amplitudes, while coarse-scale subband signals may reflect large low-frequency changes with relatively large amplitudes. This difference in magnitude can make subsequent data integration and comparison difficult, as directly comparing data of different orders of magnitude may lead to inaccurate or unreasonable results. Normalizing the calculated amplitude data can map amplitude data at different scales to the same numerical range, typically the [0,1] interval. Normalization eliminates the influence of data magnitude, making amplitude data at different scales comparable and facilitating subsequent data integration, feature extraction, and data analysis.

[0136] Encoding the duration of fluctuations corresponding to each mutation point transforms continuous duration data into discrete coded values. Continuous duration data can be complex to analyze and process, while discrete coded values ​​facilitate feature extraction and data analysis. Based on the duration's length, it is divided into different time intervals, and each interval is assigned a corresponding coded value. For example, the duration can be divided into three intervals: short (0-10 seconds), medium (10-60 seconds), and long (greater than 60 seconds), coded as 1, 2, and 3 respectively. In this way, the duration of each mutation point can be represented by a simple coded value, facilitating subsequent statistical analysis and pattern recognition.

[0137] A database table format is used, with the organized mutation point information as the primary key, normalized fluctuation amplitude data, and coded duration data as fields, establishing data relationships. The primary key uniquely identifies each mutation point, allowing different types of data to be linked to form a complete data record. This facilitates the retrieval of corresponding fluctuation amplitude and duration data when querying and analyzing information about a specific mutation point. Based on the temporal location of the mutation point, the corresponding fluctuation amplitude and duration information are integrated to form a complete feature dataset for each mutation point. This complete feature dataset contains multifaceted information about the mutation point, such as temporal location, sub-band signal scale, amplitude variation, fluctuation amplitude, and duration. This information comprehensively reflects the characteristics of pressure fluctuations, providing comprehensive data support for subsequent pressure fluctuation pattern recognition and risk assessment.

[0138] Subband signals at different scales may reflect different aspects of pressure signals. Fine-scale subband signals can capture high-frequency details and local abrupt changes in pressure signals, while coarse-scale subband signals can reflect low-frequency trends and overall changes. Therefore, the abrupt change feature data corresponding to subband signals at each scale has unique value. By fusing the abrupt change feature data corresponding to subband signals at each scale—for example, for each abrupt change, weighted averaging of the normalized fluctuation amplitude and encoded duration corresponding to each subband signal at each scale—a comprehensive fluctuation amplitude and duration feature can be obtained. The weighted average can assign different weights based on the importance of different scales; for example, if the coarse-scale signal is considered to better reflect the overall trend of pressure fluctuations, a larger weight can be assigned to the coarse-scale signal. Through fusion processing, the characteristics of subband signals at different scales can be comprehensively considered, resulting in a more comprehensive and accurate description of pressure fluctuation characteristics, which helps improve the accuracy of subsequent pattern recognition and risk assessment.

[0139] The effects of the above technical solution are as follows:

[0140] By unifying and organizing information such as the time location of mutation points, subband signal scale, and amplitude changes, and storing it in a standardized table format, the systematization and standardization of the data are ensured, facilitating subsequent analysis and comparison.

[0141] By normalizing the fluctuation amplitude of subband signals at different scales, interference from magnitude differences is eliminated, making the fluctuation amplitude data more comparable and integrated at different scales, thus improving the accuracy and reliability of the data.

[0142] By dividing the duration of fluctuations into different intervals and encoding them, the originally continuous data is transformed into discrete encoded values, which simplifies the subsequent feature extraction and data analysis process and facilitates efficient processing and application.

[0143] By using a database table format to associate and store mutation point information, normalized fluctuation amplitude data, and coded duration data, the data integration and query process is greatly simplified, improving data management efficiency and convenience.

[0144] By weighted averaging of the abrupt change point feature data of subband signals at different scales, the fluctuation amplitude and duration features of each scale can be integrated to form a comprehensive feature dataset, making the analysis of pressure fluctuations more comprehensive and accurate.

[0145] Through automated coding, normalization, and data integration, the need for manual intervention is significantly reduced, the efficiency and accuracy of data processing are improved, and the automation level of the entire system is enhanced.

[0146] The structured database storage method makes it easier to expand, update and analyze information on mutation points and fluctuation characteristics, providing a flexible operating space and facilitating subsequent analysis and research on more complex fluctuation patterns.

[0147] In one embodiment of the present invention, S3 includes:

[0148] S31. Obtain historical operating data of the micro pump station and perform the same preprocessing operation on the historical data as on the real-time operating data;

[0149] S32. Construct a stress prediction model based on the LSTM algorithm; train the stress prediction model using preprocessed historical data;

[0150] S33. Use the trained pressure prediction model to predict the pressure of the micro pump station in real time; input the current real-time operating data and output the pressure prediction data for a period of time in the future;

[0151] S34. Based on the predicted data, set a pressure fluctuation threshold to determine in advance whether the pressure may fluctuate significantly.

[0152] The working principle of the above technical solution is as follows:

[0153] Acquiring historical operating data of micro-pump stations, including real-time operating data (such as pump speed, flow rate, valve opening, etc.) and corresponding pressure data, is fundamental to building an effective pressure prediction model. Historical data contains patterns and regularities in pressure changes during pump station operation. By learning from and analyzing this data, the model can capture various factors influencing pressure changes and their interrelationships, thus providing a basis for future pressure predictions. Performing the same preprocessing operations on historical data as on real-time operating data aims to ensure data consistency and quality. Preprocessing operations may include data cleaning (removing outliers, filling in missing values, etc.), data standardization (making data of different dimensions comparable), and noise removal (improving the signal-to-noise ratio). Through preprocessing, interfering factors in the data can be eliminated, enabling the model to learn more accurately the useful information in the data and improve prediction accuracy.

[0154] A stress prediction model based on the LSTM (Long Short-Term Memory) algorithm is constructed. LSTM is a special type of recurrent neural network (RNN) that effectively solves the gradient vanishing and gradient exploding problems existing in traditional RNNs, and is particularly suitable for processing and predicting time-varying events with long time intervals. When constructing the model, the number of neurons in the input layer, hidden layer, and output layer needs to be determined. The number of neurons in the input layer is determined based on the size of the selected historical data time window and the number of features. For example, if data from the past 10 time points is selected as input, and each time point has 5 features, then the number of neurons in the input layer is 10 × 5 = 50. The number of neurons in the hidden layer can be adjusted through experiments and experience. Generally, the more neurons in the hidden layer, the stronger the model's fitting ability, but it may also lead to overfitting; too few neurons may prevent the model from fully learning the complex patterns in the data. The number of neurons in the output layer is usually determined based on the prediction target. In this scheme, the number of neurons in the output layer is 1, used to output the predicted stress value at a future time point.

[0155] When training a stress prediction model using preprocessed historical data, the historical data is divided into training, validation, and test sets. The training set is used to learn the model parameters, the validation set is used to evaluate the model's performance during training to prevent overfitting, and the test set is used to evaluate the model's generalization ability. Cross-validation methods are used for model training, such as k-fold cross-validation. The dataset is divided into k subsets, and k-1 subsets are used as the training set each time, with the remaining subset as the validation set. This process is repeated k times, and the average performance is used as the model's evaluation result. During training, the model's weights and biases are continuously adjusted to minimize the model's prediction error. Backpropagation and gradient descent are typically used to update the model's parameters. The model's performance is evaluated using the validation set. Training stops when the model's performance on the validation set no longer improves to avoid overfitting. After training and optimization, the model can learn the patterns of stress changes in historical data, thus accurately predicting future stress.

[0156] When using a trained pressure prediction model to predict the pressure of a micro-pump station in real time, the current real-time operating data is input. This real-time operating data undergoes the same preprocessing operations as historical data before being used as input to the model. Based on learned historical patterns and regularities, the model processes and analyzes the input real-time data, outputting pressure prediction data for a future period. Real-time prediction can promptly reflect the pressure change trend under the current operating status of the pump station, providing decision support for operators.

[0157] Based on forecast data, pressure fluctuation thresholds are set to predict in advance whether large-scale pressure fluctuations are likely. The setting of these thresholds is typically based on statistical analysis of historical data and practical operational experience. For example, a reasonable threshold range is determined by analyzing the normal pressure fluctuation range in historical data. When the predicted pressure exceeds the normal fluctuation range by a certain percentage, it is considered that a large-scale fluctuation is likely. This percentage can be adjusted according to actual conditions to ensure the accuracy and timeliness of the judgment. Predicting the possibility of large-scale pressure fluctuations in advance is of great significance. Once a large-scale fluctuation is determined, operators can take timely measures, such as adjusting pump operating parameters and checking equipment status, to avoid pressure fluctuations affecting the normal operation of the pumping station and ensure its safe and stable operation.

[0158] The effects of the above technical solution are as follows:

[0159] By constructing a pressure prediction model based on the LSTM algorithm and training the model with historical data, the pressure change patterns of micro pumping stations can be effectively captured, thereby achieving high-precision pressure prediction and improving the reliability and stability of the prediction.

[0160] By using cross-validation to train the stress prediction model and adjusting the model's weights and biases during training, overfitting was effectively avoided, the model's generalization ability in different scenarios was enhanced, and the model was ensured to adapt to various stress fluctuation patterns.

[0161] By using a trained LSTM model for real-time pressure prediction, future pressure changes can be predicted quickly and accurately during the operation of a micro pumping station, improving the response speed and processing efficiency of real-time monitoring.

[0162] By setting a pressure fluctuation threshold, the system can anticipate potential large-scale fluctuations when the predicted pressure exceeds the normal fluctuation range, thereby providing maintenance personnel with timely early warning information and reducing the risk of equipment damage and system failure.

[0163] Automated stress prediction and fluctuation warning functions reduce the need for manual monitoring and intervention, lower the risk of human error, reduce operating and maintenance costs, and improve management efficiency.

[0164] By preprocessing historical and real-time data and analyzing them using the efficient LSTM algorithm, accurate prediction results are provided, enhancing the precision and reliability of the decision support system and helping managers make more scientific operational decisions.

[0165] The LSTM-based pressure prediction model can handle different historical and real-time data, has strong adaptability, can be flexibly adjusted according to changes in system operating status, and can be extended to other types of pump stations or equipment, providing broader application value.

[0166] In one embodiment of the present invention, step S4 includes:

[0167] S41. Based on the actual operating needs and design requirements of the preset micro pump station, obtain control target data, and calculate the probability of pressure exceeding the allowable range based on pressure prediction data and control target data;

[0168] S42. Assess the potential losses caused by pressure exceeding the allowable range and obtain pressure fluctuation risk assessment data;

[0169] S43. Establish a fuzzy control rule base; formulate a pressure fluctuation control strategy based on the fuzzy control rule base.

[0170] The working principle of the above technical solution is as follows:

[0171] Obtaining control target data based on the pre-set actual operating needs and design requirements of the micro-pump station forms the foundation for the entire pressure fluctuation control strategy. The desired outlet pressure range clarifies the target pressure range that the pump station needs to reach during operation, ensuring that the outlet water meets actual usage requirements. The allowable pressure fluctuation range specifies the permissible fluctuation amplitude during normal operation; excessive pressure fluctuations may damage the equipment or affect the quality of the outlet water. The system response time requirement limits the time interval from detecting a pressure anomaly to taking control measures, ensuring that the system can respond to pressure changes promptly. For example, if the desired outlet pressure range is [0.2, 0.4] MPa, the allowable pressure fluctuation range is ±0.05 MPa, and the system response time requirement is no more than 10 seconds, these specific data provide clear standards for subsequent pressure analysis and control.

[0172] Based on pressure prediction data and control target data, the probability of pressure exceeding the allowable range is calculated. Using methods such as Monte Carlo simulation, a large number of pressure prediction samples are generated to simulate pressure changes under various possible conditions. Then, the proportion of samples exceeding the allowable range out of the total sample size is counted to obtain the probability of pressure exceeding the allowable range. This method takes into account the uncertainty and randomness of pressure changes, providing a more accurate basis for risk assessment. For example, if 200 out of 1000 generated pressure prediction samples exceed the allowable range, then the probability of pressure exceeding the allowable range is 20%. By calculating this probability, the likelihood of pressure exceeding the allowable range can be understood in advance, providing a reference for subsequent risk assessment and control strategy formulation.

[0173] Assessing the potential losses caused by pressure exceeding permissible limits includes equipment damage costs, downtime losses, repair costs, and potential social impacts. Equipment damage costs involve the expense of replacing or repairing damaged equipment; downtime losses refer to economic losses caused by production stoppages and water supply interruptions due to abnormal pressure stopping the pump station; repair costs include the expenses for equipment maintenance and upkeep; social impacts may include the impact on the domestic water supply of surrounding residents and on industrial production. By assessing these losses, pressure fluctuation risk assessment data is obtained, enabling a comprehensive understanding of the potential consequences of pressure fluctuations and providing a basis for risk level classification.

[0174] Based on pressure fluctuation risk assessment data and combined with fuzzy control theory, this approach uses the pressure fluctuation risk assessment data as fuzzy input variables. Fuzzy control theory can handle information with uncertainty and fuzziness, classifying risk levels into three levels: low, medium, and high, and mapping the risk assessment data to these three fuzzy sets. For example, when the risk assessment data is low, the corresponding risk level is low; when the risk assessment data is at a medium level, the corresponding risk level is medium; and when the risk assessment data is high, the corresponding risk level is high. In this way, continuous risk assessment data is transformed into discrete fuzzy levels, facilitating subsequent fuzzy inference and control strategy formulation.

[0175] Adjustments to pump operating frequency and valve opening degree are used as fuzzy output variables. Adjusting the pump operating frequency directly affects the pump's flow rate and pressure, while adjusting the valve opening degree changes the pipeline resistance, thus affecting pressure distribution. By determining these fuzzy output variables, the pressure of the pumping station can be regulated and controlled. A fuzzy control rule base is established, and corresponding control rules are formulated based on the risk level. For example, when the risk level is high, rules for larger pump frequency and valve opening degree adjustments are formulated to quickly reduce the risk caused by pressure fluctuations; when the risk level is medium, rules for moderate adjustments are formulated; and when the risk level is low, rules for smaller adjustments are formulated or the current operating state is maintained. The fuzzy control rule base is based on expert experience and actual operating conditions, reflecting the control measures that should be taken under different risk levels.

[0176] Based on the fuzzy control rule base, a pressure fluctuation control strategy is formulated, and control measures are determined for different risk levels. Once pressure fluctuations are detected in real time and the risk level is assessed, the appropriate pump operating frequency adjustment and valve opening adjustment are selected according to the rules in the fuzzy control rule base to control the pump station in real time, ensuring the pressure returns to the allowable range and guaranteeing stable operation of the pump station.

[0177] The effects of the above technical solution are as follows:

[0178] By combining pressure prediction data and control target data, the probability of pressure exceeding the allowable range can be calculated, enabling accurate assessment of the risk of pressure fluctuations. Based on the calculation results, effective control strategies can be formulated, thereby improving the accuracy and reliability of pressure fluctuation control.

[0179] By assessing the potential losses caused by pressure exceeding the allowable range and combining this with fuzzy control theory for risk assessment, pressure fluctuation problems at different risk levels can be accurately identified, providing an effective early warning mechanism for the system and enhancing its risk prediction capabilities.

[0180] By setting fuzzy output variables such as pump frequency adjustment and valve opening adjustment, and formulating specific control strategies based on the fuzzy control rule base, the system can quickly respond and take appropriate control measures under different risk levels, thereby improving the system's emergency response capability and control effect.

[0181] By implementing precise pressure fluctuation control strategies and adjusting the working status of water pumps and valves in a timely manner, the risk of equipment overload operation is reduced, thereby effectively reducing the possibility of equipment failure and downtime and ensuring the stable operation of the micro pump station.

[0182] By combining fuzzy control theory for risk assessment and stress control, energy waste caused by over-control is reduced, while maintenance and downtime losses due to equipment failure and shutdown are also reduced, thereby improving the system's economy and long-term sustainability.

[0183] By combining fuzzy control theory with pressure fluctuation risk assessment data, a more systematic and scientific decision support is provided, helping managers make more accurate and effective control decisions.

[0184] This control method can automatically adjust the control strategy according to different risk levels, has strong adaptability, can cope with pressure fluctuation problems in various complex operating environments, and improves the adaptability and flexibility of the system.

[0185] In one embodiment of the present invention, step S5 includes:

[0186] S51. Based on pressure prediction data and the established pressure fluctuation control strategy, send control commands to the pumps and valves of the micro pump station; during the control process, continuously monitor the operating status and pressure changes of the pump station in real time.

[0187] S52. Collect actual pressure data in real time through sensors, compare it with predicted pressure data, calculate control error, and dynamically adjust control strategy based on comparison results;

[0188] S53. Periodically evaluate the control effect. If the pressure fluctuation still exceeds the allowable range, further optimize the pressure prediction model and control strategy to form a closed-loop control system.

[0189] The working principle of the above technical solution is as follows:

[0190] Based on pressure prediction data and a pre-defined pressure fluctuation control strategy, control commands are sent to the pumps and valves of the micro-pump station. This process is typically implemented using a programmable logic controller (PLC) or a distributed control system (DCS). PLCs offer high reliability and flexible programming, while DCSs are suitable for large-scale, distributed control systems. These control systems allow for precise real-time adjustment of pump operating frequency and valve opening. For example, if pressure prediction indicates a potential pressure increase in the near future, the system sends commands to reduce pump operating frequency or decrease valve opening to reduce the amount of water entering the pipeline, thereby lowering the pressure; conversely, if a pressure decrease is predicted, the system increases pump operating frequency or increases valve opening.

[0191] During the control process, the operating status and pressure changes of the pumping station are continuously monitored in real time. Real-time monitoring can promptly acquire various operating parameters of the pumping station, such as pump speed, current, flow rate, valve opening, and actual pressure in the pipeline. By monitoring these operating statuses and pressure changes, any abnormalities in the control process can be detected in a timely manner, such as pump malfunctions or valve jamming, providing a basis for subsequent control adjustments and ensuring that the pumping station operates as expected.

[0192] Real-time pressure data is collected by sensors, which accurately convert the pressure signal in the pipeline into an electrical signal and transmit it to the control system. The collected actual pressure data is compared with the predicted pressure data. This comparison process reveals the difference between the actual operating conditions and the predictions. For example, if the predicted pressure is 0.3 MPa at a certain moment, but the actual collected pressure is 0.35 MPa, this indicates a deviation between the actual and predicted pressures. The control error is calculated based on the comparison results, reflecting the magnitude of the difference between the actual and predicted pressures. The control strategy is dynamically adjusted according to the magnitude and direction of the control error. When the deviation between the actual and predicted pressures is large, it indicates that the current control measures may not be effective enough, requiring adjustments to the magnitude and direction of pump frequency and valve opening adjustments. For example, if the actual pressure is much higher than the predicted pressure, it may be necessary to increase the reduction in pump frequency or valve opening; if the actual pressure is much lower than the predicted pressure, it may be necessary to increase the increase in pump frequency or valve opening. Through this dynamic adjustment, control accuracy can be improved, bringing the actual pressure closer to the desired pressure range.

[0193] Regularly evaluate the control effectiveness and set evaluation indicators, such as pressure fluctuation range and system stability. Pressure fluctuation range reflects the magnitude of pressure change over a certain period, while system stability reflects the pump station's ability to resist external interference and maintain normal operation. Analyzing these evaluation indicators reveals the actual effectiveness of the current control strategy. For example, if the pressure fluctuation range remains large, it indicates that the control strategy's suppression of pressure fluctuations is not ideal; poor system stability may lead to frequent pressure fluctuations or equipment failures. If pressure fluctuations still exceed the allowable range, further optimize the pressure prediction model and control strategy. For the pressure prediction model, increase training data to enable the model to learn more pressure change patterns under different operating conditions; adjust model parameters to optimize model performance and improve prediction accuracy. For the control strategy, refine fuzzy control rules and make more detailed classifications and adjustments to control measures under different risk levels based on actual conditions. By continuously optimizing the pressure prediction model and control strategy, a closed-loop control system is formed, allowing the system to continuously adjust and improve itself according to actual conditions, continuously improving control performance, ensuring the pressure of the micro-pump station remains stable within the allowable range, and guaranteeing the safe and stable operation of the pump station.

[0194] The effects of the above technical solution are as follows:

[0195] By sending control commands to pumps and valves in real time based on pressure prediction data and control strategies, and by precisely adjusting the pump operating frequency and valve opening through a PLC or DCS system, it is possible to effectively respond quickly to actual pressure fluctuations, significantly improving control accuracy and real-time performance.

[0196] By continuously monitoring the pump station's operating status and pressure changes in real time, and comparing predicted and actual pressure data, the control strategy can be dynamically adjusted to ensure that the system can flexibly adjust control measures according to the actual situation, thereby enhancing the system's adaptability to complex pressure fluctuations.

[0197] By comparing the actual pressure with the predicted pressure, calculating the control error, and dynamically adjusting the pump frequency and valve opening adjustment range and direction, the control deviation caused by the prediction error can be effectively corrected, further improving the control accuracy and stability of the system.

[0198] By regularly evaluating the control effectiveness and setting assessment indicators such as pressure fluctuation range and system stability, deficiencies in the system can be identified in a timely manner and targeted optimizations can be made, enhancing the system's optimization and self-improvement capabilities and ensuring continuous improvement in control effectiveness.

[0199] By continuously optimizing control strategies, improving pressure prediction models, and adjusting fuzzy control rules, it is possible to effectively reduce excessive pressure fluctuations and equipment failures caused by inaccurate control, thereby reducing the risk of equipment damage and system downtime.

[0200] By continuously forming a closed-loop control system, the pressure fluctuations can be precisely controlled and optimized, maintaining the stable operation of the micro pump station system, significantly improving the system's reliability and stability, and reducing the impact of external interference on the system.

[0201] By regularly evaluating and optimizing control effectiveness, and adjusting predictive models and control strategies based on actual operational data, the scientific rigor and accuracy of the decision-making process are ensured, helping managers make control decisions that are more in line with actual conditions.

[0202] In one embodiment of the present invention, S52 includes:

[0203] Pressure sensors placed at key locations in the micro pump station are used to collect actual pressure data in real time at a set sampling frequency; and the collected raw data is preliminarily processed.

[0204] A timestamp matching algorithm is used to precisely align the actual pressure data and the predicted pressure data on the time axis. The synchronized and aligned actual pressure data and predicted pressure data are compared point by point to calculate the control error at each time point.

[0205] Perform trend analysis on the calculated control error, and dynamically adjust the control strategy based on the results of the error trend analysis;

[0206] After dynamically adjusting the control strategy, the changes in actual pressure data are continuously monitored to verify the adjustment effect.

[0207] The working principle of the above technical solution is as follows:

[0208] Pressure sensors are strategically placed at key locations in the micro-pump station. These sensors act as the "eyes" of the pump station's pressure, sensing real-time pressure changes within the pipeline. Actual pressure data is collected in real-time at a set sampling frequency. The selection of the sampling frequency needs to comprehensively consider factors such as the rate of pressure change and the response speed of the control system. A suitable sampling frequency ensures that the collected data accurately reflects the dynamic changes in pressure, avoiding the loss of important information due to an excessively low sampling frequency, and preventing excessive data processing burden due to an excessively high sampling frequency. The collected raw data often contains various noises and interferences, such as electromagnetic interference and sensor measurement errors. Therefore, preliminary processing of the raw data is necessary, such as filtering and denoising. Filtering can be achieved by designing appropriate filters, such as low-pass, high-pass, or band-pass filters, to remove high-frequency noise or low-frequency interference from the data. Denoising can be achieved using mathematical methods, such as wavelet denoising and median filtering, to further improve data quality. The data after preliminary processing is more accurate and reliable, providing a solid foundation for subsequent data comparison and analysis.

[0209] A timestamp matching algorithm is employed to precisely align actual pressure data with predicted pressure data on the timeline. Since there may be slight time differences between actual data acquisition and predicted data generation, direct comparison without alignment would lead to inaccurate error calculations. The timestamp matching algorithm matches actual and predicted pressure data at the same point in time based on the data's timestamps, ensuring accurate comparison. The synchronized actual and predicted pressure data are then compared point-by-point, calculating the control error at each time point. The control error is represented by the difference between the actual and predicted pressure values, directly reflecting the deviation between actual and predicted conditions. For example, at a given time point, if the actual pressure is 0.32 MPa and the predicted pressure is 0.3 MPa, the control error at that time point is 0.02 MPa. By calculating the control error at each time point, a comprehensive understanding of the differences between actual and predicted pressures over the entire time range can be obtained.

[0210] Trend analysis is performed on the calculated control error to observe how the error changes over time. Error curves can be plotted to visually observe the trend of error changes, such as increases, decreases, or fluctuations; error statistics, such as mean and variance, can also be calculated. The mean reflects the average level of the error, while the variance reflects the dispersion of the error. Error trend analysis can determine whether the error exhibits a significant upward or downward trend, and whether periodic fluctuations exist. If the error continues to rise, it indicates that the deviation between the actual pressure and the predicted pressure is increasing, and the control system may have performance degradation issues. If the error exhibits periodic fluctuations, it may be related to factors such as the pump station's operating cycle or external disturbances. Error trend analysis helps to understand the dynamic performance and stability of the control system, providing a basis for subsequent control strategy adjustments.

[0211] Based on the error trend analysis, the pump frequency adjustment is dynamically adjusted. When the actual pressure is consistently lower than the predicted pressure, it indicates that the current pump output pressure is insufficient, and the pump frequency adjustment needs to be appropriately increased to increase the pump speed and thus increase the outlet pressure. Conversely, when the actual pressure is consistently higher than the predicted pressure, it indicates that the pump output pressure is too high, and the pump frequency adjustment needs to be reduced to decrease the pump speed and decrease the outlet pressure. In this way, the actual pressure can be made closer to the predicted pressure, improving the accuracy of control.

[0212] Adjusting valve opening is also a crucial means of pressure control. Based on the magnitude and trend of the error, the valve opening is flexibly adjusted to quickly respond to pressure changes. For example, when the error is large and trending upwards, it indicates a rapid pressure increase, requiring a rapid increase in valve opening to expand the flow area in the pipeline and reduce pressure. When the error is small and stable, it means the pressure is close to the desired value, and the valve opening can be kept constant or finely adjusted to maintain pressure stability. Combining intelligent control algorithms such as fuzzy control and neural network control allows for automatic adjustment of control parameters based on the error and its rate of change, achieving more refined control. Fuzzy control can handle information with uncertainty and fuzziness, automatically adjusting control quantities according to the magnitude of the error and its rate of change using fuzzy rules. Neural network control, on the other hand, can automatically optimize control parameters by learning from a large amount of historical data, improving the control's adaptability and robustness. Introducing intelligent control algorithms makes the control system more intelligent and efficient, better able to cope with complex pressure changes.

[0213] After dynamically adjusting the control strategy, the changes in actual pressure data are continuously monitored to verify the adjustment effect. The effectiveness of the adjustment strategy is evaluated by comparing indicators such as control error and pressure fluctuation range before and after the adjustment. For example, if the control error significantly decreases and the pressure fluctuation range narrows after adjustment, the adjustment strategy is effective; conversely, if the adjustment effect is unsatisfactory, further analysis of the reasons is needed. If the adjustment effect is unsatisfactory, in-depth analysis of the reasons is required, which may include inaccurate pressure prediction models, unreasonable control strategies, sensor malfunctions, etc. Based on the analysis results, the control strategy is readjusted, such as further optimizing the pressure prediction model, adjusting the parameters of the intelligent control algorithm, and replacing or calibrating sensors. Through continuous adjustment and optimization, until satisfactory control effects are achieved, a closed-loop control system is formed, continuously improving the accuracy and stability of pump station pressure control.

[0214] The effects of the above technical solution are as follows:

[0215] By using a timestamp matching algorithm to accurately align actual pressure data and predicted pressure data, and comparing them point by point on the timeline to calculate control error, the control error at each time point can be accurately calculated, thereby improving the accuracy of system control.

[0216] By conducting trend analysis on control errors and observing their changing trends and periodic characteristics, we can gain a more comprehensive understanding of the dynamic performance and stability of the control system, providing a scientific basis for further optimizing control strategies.

[0217] Based on the results of error trend analysis, dynamically adjusting the pump frequency and valve opening can flexibly respond to system changes, improve the system's adaptability to pressure fluctuations, and ensure the precision and accuracy of pressure control.

[0218] By introducing intelligent control algorithms (such as fuzzy control and neural network control), the control parameters are automatically adjusted according to the error and the rate of change of the error, so that the system maintains higher stability during dynamic adjustment, reduces the complexity of manual adjustment, and enhances the stable operation of the system.

[0219] By flexibly adjusting valve opening and pump frequency, the system can quickly respond to changes in actual pressure, avoiding disruption to the normal operation of the pumping station due to excessive pressure fluctuations, and improving the system's response speed and efficiency to pressure changes.

[0220] After dynamically adjusting the control strategy, by continuously monitoring the actual pressure data and evaluating the control effect before and after the adjustment, the control error can be effectively reduced and the pressure fluctuation range can be decreased, thereby improving the control accuracy and operational stability of the pumping station.

[0221] By evaluating the adjusted control effect, the effectiveness of the adjustment strategy can be quickly verified, and the reasons for unsatisfactory control effects can be identified in a timely manner, thereby optimizing the control strategy and improving the efficiency of control strategy adjustment and optimization.

[0222] By combining intelligent control algorithms, the system can automatically adjust according to changes in error without relying on manual intervention, thereby enhancing the system's intelligence level and autonomous adjustment capability, and thus improving the overall intelligence of the system operation.

[0223] According to one embodiment of the present invention, a micro pumping station pressure fluctuation prevention control system includes a memory, a processor, and a computer program stored in the memory and executable on the memory. The processor executes the program to implement any of the micro pumping station pressure fluctuation prevention control methods described above.

[0224] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for controlling large-range pressure fluctuations in a micro-pump station, characterized in that, The method includes: S1: Acquire real-time operating data of the micro pumping station, analyze the pumping station's operating status based on the real-time operating data, construct a micro pumping station operating status model, and based on the micro pumping station operating status model, perform dynamic characteristic analysis of the pumping station system to obtain the analysis results of factors affecting pressure fluctuations in the micro pumping station; the dynamic characteristic analysis of the pumping station system specifically includes: Based on the constructed pump station operation status model, simulation experiments were conducted to simulate the operation of the pump station under different working conditions; through parameter sensitivity analysis, the key factors affecting pressure fluctuations and their mechanisms of action were identified; and the analysis results of the factors affecting pressure fluctuations in micro pump stations were obtained. S2: Extract pressure fluctuation features from the inlet and outlet pressure data in the real-time operation data. Use a combination of Fourier transform and wavelet analysis to analyze the frequency components and time-domain characteristics of the pressure signal to obtain pressure fluctuation feature data of the micro-pump station. Based on the pressure fluctuation feature data, use a clustering algorithm to classify the pressure fluctuation patterns, identify normal fluctuation patterns, small-range fluctuation patterns, and large-range fluctuation patterns, and establish a feature library corresponding to different fluctuation patterns. For the large-range fluctuation pattern, further analyze its causes and patterns, and establish a causal relationship model for the large-range fluctuation pattern. S3: Based on the analysis results of the factors affecting the pressure fluctuation of the micro pump station and the feature library corresponding to different fluctuation modes, construct a pressure prediction model for the micro pump station; use the pressure prediction model to predict the pressure of the micro pump station in real time and obtain pressure prediction data; set a pressure fluctuation threshold according to the prediction data to determine in advance whether the pressure will fluctuate in a large range. S4: Obtain the control target data of the micro pump station; based on the pressure prediction data and the control target data, conduct a pressure fluctuation risk assessment, calculate the probability of pressure exceeding the allowable range and the resulting losses, and obtain pressure fluctuation risk assessment data; based on the pressure fluctuation risk assessment data, and combined with fuzzy control theory, formulate a pressure fluctuation control strategy, and determine the control measures to be taken under different risk levels; S5: Based on the pressure prediction data and the established pressure fluctuation control strategy, the pumps and valves of the micro pump station are controlled in real time. During the control process, the operating status and pressure changes of the pump station are monitored in real time, and the actual pressure data is compared with the predicted pressure data. The control strategy is dynamically adjusted according to the comparison results. At the same time, the control effect is evaluated. If the pressure fluctuation still exceeds the allowable range, the pressure prediction model and control strategy are further optimized to form a closed-loop control system.

2. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 1, characterized in that, S1 includes: S11. Install high-precision sensors at key locations in the micro pump station, and collect real-time operating data of the pump station in real time according to the set sampling frequency through the data acquisition system, and transmit the collected real-time operating data to the data processing center. S12. Clean the collected real-time running data to remove outliers and noisy data; and normalize the data to unify data of different dimensions into the range of [0,1] or [-1,1]. S13. Using a system identification method, the preprocessed data is processed to construct a micro pump station operation status model.

3. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 1, characterized in that, The S2 includes: S21. Perform Fourier transform on the inlet and outlet pressure data in the real-time operation data to convert them from the time domain to the frequency domain and analyze the frequency components of the pressure signal. S22. Wavelet analysis is used to decompose the pressure signal into multiple scales and extract time-domain features at different scales; thus obtaining pressure fluctuation characteristic data of micro-pump stations. S23. Use the K-means clustering algorithm to perform cluster analysis on the pressure fluctuation feature data; calculate the distance from each pressure fluctuation feature data point to each cluster center, and assign it to the category corresponding to the nearest cluster center; S24. Update the cluster centers. Repeat the above allocation and update process until the cluster centers no longer change or the set number of iterations is reached, thereby identifying different fluctuation patterns and establishing a feature library corresponding to different fluctuation patterns.

4. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 3, characterized in that, S21 includes: The inlet and outlet pressure data are preprocessed by Fourier transform. The preprocessed inlet and outlet pressure data are then transformed from the time domain to the frequency domain to obtain the pressure signal in the frequency domain. In the frequency domain, the frequency components of the pressure signal are analyzed, and the amplitude of the frequency components is calculated. Based on the frequency component analysis results, frequency characteristics related to pressure fluctuations are identified, and the obtained frequency component data and frequency characteristics related to pressure fluctuations are organized and stored.

5. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 3, characterized in that, S22 includes: S221. Using wavelet basis functions, perform multi-scale wavelet decomposition on the pressure signal; by progressively decomposing the signal to different scales, obtain a series of sub-band signals with different frequency ranges; and each sub-band signal reflects the characteristics of the original signal at different scales. S222. In sub-band signals at various scales, the modulus maxima characteristic of wavelet transform is used to detect abrupt changes in the pressure signal; for each sub-band signal at each scale, its fluctuation amplitude is calculated. S223. Obtain the time of occurrence of each mutation point and the duration of its corresponding fluctuation amplitude; by analyzing the duration, obtain the stability and trend of pressure fluctuations; S224. Integrate the detected mutation point information, the calculated fluctuation amplitude data, and the duration analysis results to form micro-pump station pressure fluctuation characteristic data.

6. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 5, characterized in that, S224 includes: Information on abrupt changes detected in sub-band signals at various scales is organized and stored in a unified data format. The calculated fluctuation amplitude data is then normalized. For each mutation point, the fluctuation amplitude and duration are encoded; based on the length of the duration, it is divided into different time intervals, and each interval is assigned a corresponding encoding value. Using a database table format, the organized mutation point information is used as the primary key, the normalized fluctuation amplitude data and the encoded duration data are used as fields to establish data relationships; based on the time location of the mutation point, the corresponding fluctuation amplitude and duration information are integrated together to form a complete feature dataset for each mutation point. The abrupt change point feature data corresponding to sub-band signals at each scale are fused to obtain comprehensive fluctuation amplitude and duration features.

7. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 1, characterized in that, The S3 includes: S31. Obtain historical operating data of the micro pump station and perform the same preprocessing operation on the historical data as on the real-time operating data; S32. Construct a stress prediction model based on the LSTM algorithm; train the stress prediction model using preprocessed historical data; S33. Use the trained pressure prediction model to predict the pressure of the micro pump station in real time; input the current real-time operating data and output the pressure prediction data for a period of time in the future.

8. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 1, characterized in that, The S4 includes: S41. Based on the actual operating needs and design requirements of the preset micro pump station, obtain control target data, and calculate the probability of pressure exceeding the allowable range based on pressure prediction data and control target data; S42. Assess the losses caused by pressure exceeding the allowable range and obtain pressure fluctuation risk assessment data; S43. Establish a fuzzy control rule base; formulate a pressure fluctuation control strategy based on the fuzzy control rule base.

9. The method for controlling large-range pressure fluctuations in a micro-pump station according to claim 1, characterized in that, The S5 includes: S51. Based on pressure prediction data and the established pressure fluctuation control strategy, send control commands to the pumps and valves of the micro pump station; during the control process, continuously monitor the operating status and pressure changes of the pump station in real time. S52. Collect actual pressure data in real time through sensors, compare it with predicted pressure data, calculate control error, and dynamically adjust control strategy based on comparison results; S53. Periodically evaluate the control effect. If the pressure fluctuation still exceeds the allowable range, further optimize the pressure prediction model and control strategy to form a closed-loop control system.

10. A micro pump station pressure fluctuation prevention control system, characterized in that, The system includes a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement a micro pump station pressure fluctuation control method as described in any one of claims 1-9.

Images