Control method for intelligently regulating and controlling multi-working-condition ultrahigh-lift single-stage centrifugal pump

Abstract

The invention relates to the technical field of centrifugal pump control, and discloses an intelligent control multi-working-condition ultrahigh-lift single-stage centrifugal pump control method which comprises the steps that attention-guided dimension reduction fusion and multi-scale feature extraction are conducted on multi-source working condition data, and multi-scale operation features are obtained; time evolution and clustering processing are conducted on the multi-scale operation characteristics through a time clustering algorithm based on density, and the current working condition of the ultrahigh-lift single-stage centrifugal pump is obtained; and taking the current working condition, the performance parameters and the performance score of the ultrahigh-lift single-stage centrifugal pump as state vectors, and selecting an optimal control vector from a control space by adopting a reinforcement learning model. By fusing multi-source data and introducing time evolution modeling and reinforcement learning optimization strategies, intelligent sensing and self-adaptive control of the ultra-high-lift single-stage centrifugal pump under multiple working conditions and the whole life cycle are achieved, the problems that response to complex operation states is slow and adjustment lags in traditional rule driving control are solved, and the service life of the ultra-high-lift single-stage centrifugal pump is prolonged. The energy efficiency ratio, the operation stability and the equipment service life are improved.

Description

Technical Field

[0001] The present invention relates to the field of centrifugal pump control, and in particular to a control method for an intelligently regulated multi-operating-condition ultra-high-lift single-stage centrifugal pump. Background Art

[0002] The ultra-high-lift single-stage centrifugal pump is a typical fluid conveying device that transports liquid from a low-pressure area to a high-pressure area through the rotation of the impeller. With advantages such as simple structure, low cost, and easy installation and maintenance, it has been widely used in various industrial fields such as industrial cooling, urban water supply, agricultural irrigation, mine drainage, and the chemical industry. Ultra-high-lift single-stage centrifugal pumps are particularly popular in applications requiring continuous liquid delivery while maintaining stable pressure output due to their high reliability and low failure rate. However, traditional ultra-high-lift single-stage centrifugal pumps have limited head and struggle to provide the required operating pressure in high-drop or high-resistance conveying environments. Therefore, the ultra-high-lift single-stage centrifugal pump came into being. Through impeller structure optimization, speed increase, and hydraulic parameter adjustment, it achieves the goal of achieving higher head in a single-stage structure. It is widely used in special applications such as high-rise building water supply, mountain water transportation, and remote pipeline pressure replenishment.

[0003] As conveying conditions become increasingly complex, ultra-high-lift single-stage centrifugal pumps face frequent load fluctuations, multiple operating mode switching, changes in liquid physical properties, and equipment aging during operation. These factors not only place higher demands on the pump's structural strength and hydraulic performance, but also make traditional control methods based on fixed control logic or static empirical models difficult to adapt. Such control methods are unable to dynamically respond to complex operating conditions, making it difficult to achieve the optimal balance between operating efficiency, system stability, and equipment lifespan.

[0004] Current research has attempted to improve centrifugal pump control by incorporating artificial intelligence and model prediction methods. For example, patent CN113901710B proposes a centrifugal pump operation optimization control method. This method constructs an offline database to establish a mathematical model of the system, uses actual operating data to modify an improved BPNN prediction model, and uses a binary tree search algorithm to achieve rapid online control rate matching, thereby optimizing operating parameters such as head, efficiency, and flow rate. This method enables rapid adjustment and adaptive compensation based on empirical data, alleviating operational errors caused by model bias to a certain extent. However, this method relies on a pre-built database and offline model for control, lacking the ability to recognize and dynamically adapt to complex and variable operating conditions (such as load disturbances, changes in liquid properties, and frequent starts and stops). Furthermore, it primarily relies on static models. While a historical data correction mechanism is introduced, it fails to model performance degradation trends caused by equipment aging and operational phase transitions.

[0005] In view of the problem, the application provides a kind of intelligent regulation and control multi-working condition super-high lift single-stage centrifugal pump control method, the operation optimization and regulation control of pump equipment in multiple working conditions are realized by artificial intelligence technology, and the overall performance and operating life of super-high lift single-stage centrifugal pump are significantly improved. SUMMARY

[0006] The application provides an intelligent regulation and control multi-working condition super-high lift single-stage centrifugal pump control method, which solves the problem of inaccurate working condition perception under complex conditions, and accurately outputs multiple performance parameters such as actual lift and hydraulic efficiency by using neural regression forest to fit and predict the operating characteristics under the current working condition, and solves the problems of difficult performance multi-index modeling and insufficient dynamic response capability by comprehensively quantifying through a nonlinear performance scoring function, and the current working condition, performance parameters and performance scores are used as state vectors to dynamically select the optimal control vector in the control space, realizing the joint regulation of frequency converter frequency, main shaft speed and valve opening, thereby solving the problems of control parameter adjustment lag, control target conflict and poor system stability.

[0007] To achieve the above purpose, the application provides an intelligent regulation and control multi-working condition super-high lift single-stage centrifugal pump control method, comprising the following steps: S1: periodically collecting multi-source working condition data during the operation of super-high lift single-stage centrifugal pump by using acquisition equipment, and performing attention-guided dimensionality reduction fusion on the multi-source working condition data to obtain a multi-source working condition fusion tensor; S2: multi-scale feature extraction is performed on the multi-source working condition fusion tensor to obtain multi-scale operating characteristics of the super-high lift single-stage centrifugal pump, and a density-based time clustering algorithm is used to perform time evolution processing on the multi-scale operating characteristics, and the time evolution processed multi-scale operating characteristics are clustered into clustering clusters; S3: the working condition corresponding to the clustering cluster to which the multi-scale operating characteristics belong is taken as the current working condition of the super-high lift single-stage centrifugal pump, and the performance parameters of the super-high lift single-stage centrifugal pump under the current working condition are output by using neural regression forest based on the current working condition and multi-scale operating characteristics, and the performance parameters are converted into performance scores by using a comprehensive performance scoring function; S4: the current working condition, performance parameters and performance scores of the super-high lift single-stage centrifugal pump are taken as state vectors, and the optimal control vector corresponding to the state vector is selected from the control space by using a reinforcement learning model, and the optimal control vector is used to adjust the frequency of the frequency converter, the speed of the main shaft and the opening of the valve of the super-high lift single-stage centrifugal pump.

[0008] As a further improvement method of the present invention: Optionally, the acquisition device includes a pressure sensor, an electromagnetic flowmeter, a current sensor, a power meter, a speed / frequency sensor, a temperature sensor, and an acceleration sensor; and the acquisition device is used to periodically acquire multi-source operating condition data during the operation of the ultra-high lift single-stage centrifugal pump, including: The multi-source operating condition data includes a variety of operating condition data, which are respectively the inlet instantaneous pressure data, outlet instantaneous pressure data, speed data, frequency data, temperature data, acceleration data, current data, power data of the ultra-high head single-stage centrifugal pump and the instantaneous flow data of the ultra-high head single-stage centrifugal pump; The operating condition data in the multi-source operating condition data are all sequence data with a length of N, where N represents the number of sampling moments in the process of collecting the multi-source operating condition data. The time interval between adjacent sampling moments is set to 100 milliseconds, and N is set to 50. The missing values ​​in the multi-source working condition data are interpolated and completed using a linear weighted interpolation method to obtain the completed multi-source working condition data; Adaptive weighted differential filtering is used to perform differential filtering on the data values ​​of all working condition data in the completed multi-source working condition data.

[0009] Optionally, attention-guided dimensionality reduction fusion is performed on the multi-source operating condition data after differential filtering, including: Convert the multi-source working condition data after differential filtering into matrix form: ; in, for In matrix form with M rows and M columns, Located in the matrix The Row m column, Indicates the The working condition data is The differential filtering processing results of the data values ​​at the sampling moments are obtained. M represents the number of categories of the working condition data in the multi-source working condition data. The first to M-th working condition data are the inlet instantaneous pressure data, outlet instantaneous pressure data, speed data, frequency data, temperature data, acceleration data, current data, power data and instantaneous flow data of the ultra-high head single-stage centrifugal pump, respectively. M is 9. Indicates the sequence data length of the operating condition data in the multi-source operating condition data after differential filtering; The Rows are represented as Indicates that the ultra-high head single-stage centrifugal pump is a vector composed of data values of M kinds of working condition data at each sampling time point; calculating time attention of each sampling time point in the matrix form of multi-source working condition data, wherein the time attention of the sampling time point is calculated in the following manner: ; wherein T represents transposition, the time attention of the i-th sampling time point, which is a scalar between 0 and 1; wherein e represents an exponential function with a natural constant as a base, Q represents a trainable query vector, represents a trainable linear transformation matrix; convert the matrix into a time attention weighted covariance matrix ; perform eigenvalue decomposition on the time attention weighted covariance matrix , select eigenvectors corresponding to the largest K eigenvalues to form a principal component matrix, and project the matrix to a low-dimensional space as a multi-source working condition fusion tensor, wherein the multi-source working condition fusion tensor is in the form of an N-1 row and K column matrix, and the i-th row of the multi-source working condition fusion tensor is a multi-source dimensionality reduction projection vector of the i-th sampling time point. the i-th sampling time point.

[0010] Optionally, multi-scale feature extraction is performed on the multi-source working condition fusion tensor, including: calculating local similarity and global similarity of multi-source dimensionality reduction projection vectors of any two sampling time points in the multi-source working condition fusion tensor, taking the local similarity and the global similarity as matrix elements, and respectively constructing a local adjacency graph and a global adjacency graph of N-1 sampling time points, wherein the local adjacency graph and the global adjacency graph are both in the form of an N-1 row and N-1 column matrix; taking the multi-source working condition fusion tensor, the local adjacency graph and the global adjacency graph as inputs of a graph neural network; the graph neural network includes a plurality of graph convolution layers nested in a hierarchical structure; a residual channel enhancement module is introduced between each graph convolution layer to enhance the stability and non-linear expression ability of feature transmission; the residual channel enhancement module includes an SE channel attention mechanism and a residual connection path; after the graph neural network is processed by multiple layers of graph convolution, embedded features of the last layer of graph convolution are output as multi-scale running features F representing the running state of the super-high lift single-stage centrifugal pump.

[0011] ​​Optionally, a density-based time clustering algorithm is used to perform time evolution processing on the multi-scale operation features, and the multi-scale operation features after time evolution processing are clustered into clustering clusters, including: A density-based time clustering algorithm is used to generate time evolution factors of the multi-scale operation features F ; The time evolution factors are used to perform time evolution processing on the multi-scale operation features F, to obtain multi-scale operation features after time evolution processing , denotes a control coefficient, which is set to 0.1; The multi-scale operation features after time evolution processing are calculated The Euclidean distance between the multi-scale operation features after time evolution processing and the core points of different clustering clusters is calculated, and the multi-scale operation features after time evolution processing are clustered into the clustering cluster with the closest Euclidean distance, and the working condition label of the clustered clustering cluster is taken as the current working condition of the super-high lift single-stage centrifugal pump.

[0012] Optionally, based on the current working condition of the super-high lift single-stage centrifugal pump and the multi-scale operation features, a neural regression forest is used to output performance parameters of the super-high lift single-stage centrifugal pump under the current working condition, including: The neural regression forest includes multiple parallel regression tree structures, and the split function of each regression tree is parameterized modeled by a neural network, so that it has stronger nonlinear fitting capability; the current working condition of the super-high lift single-stage centrifugal pump and the multi-scale operation features are jointly constructed into a high-dimensional feature vector, and the high-dimensional feature vector is taken as the input of the regression tree in the neural regression forest; each regression tree performs nonlinear regression mapping on the high-dimensional feature vector through an embedded shallow neural network, to obtain the regression output of each regression tree, wherein the regression output is the mapping result of the high-dimensional feature vector to the performance parameter dimension; by fusing the regression outputs of multiple regression trees, the performance parameters of the super-high lift single-stage centrifugal pump under the current working condition are obtained, and the performance parameters include actual lift, hydraulic efficiency, theoretical flow deviation, shaft power, and operating resistance change rate.

[0013] Optionally, a comprehensive performance score function is used to convert the performance parameters into performance scores, including: The performance parameters are normalized; The normalized performance parameters are taken as the input of the comprehensive performance score function, to obtain the performance scores corresponding to the performance parameters: ; wherein, denotes a normalized performance parameter vector, and the normalized performance parameter vector includes the normalized actual lift ​, hydraulic efficiency , theoretical flow deviation , shaft power , and operating resistance change rate , , wherein the performance parameter vector after normalization is represented by , the corresponding performance score is represented by , the score coefficients are represented by , and the score coefficients are equal and are 0.2.

[0014] Optionally, an optimal control vector corresponding to a state vector is selected from a control space by using a reinforcement learning model, including: a state vector as an input of the reinforcement learning model is constructed: [current working condition of the super-high-lift single-stage centrifugal pump, performance parameter, performance score]; a control space C of the super-high-lift single-stage centrifugal pump is constructed, and the control space C includes three continuous control dimensions: a frequency conversion frequency of a frequency converter in the super-high-lift single-stage centrifugal pump ; a main shaft rotating speed ; a valve opening degree of an outlet of the super-high-lift single-stage centrifugal pump ; , wherein is an output range of the frequency conversion frequency in the control space C, is a minimum output value of the frequency conversion frequency, is a maximum output value of the frequency conversion frequency, is an output range of the main shaft rotating speed in the control space C, represents a minimum output value of the main shaft rotating speed, represents a maximum output value of the main shaft rotating speed, is an output range of the valve opening degree in the control space C. a policy network in the reinforcement learning model is constructed by using a policy gradient type reinforcement learning method, the policy network takes the state vector as an input, and selects an optimal frequency conversion frequency, an optimal main shaft rotating speed and an optimal valve opening degree corresponding to the state vector from the three continuous control dimensions in the control space C to form an optimal control vector, and the optimal control vector is used to adjust the frequency conversion frequency, the main shaft rotating speed and the valve opening degree of the super-high-lift single-stage centrifugal pump.

[0015] Compared with the prior art, the present application provides an intelligent control method for a multi-working-condition super-high-lift single-stage centrifugal pump, and the present application has the following beneficial effects: Firstly, the application significantly enhances the expression ability of complex working condition characteristics by combining timing attention weighting and covariance principal component decomposition. The attention-guided dimensionality reduction fusion method not only considers the statistical synergistic relationship between features, but also dynamically identifies abnormal time points (such as sudden vibration, sharp pressure difference change, etc.) with key diagnostic value in time series data, and gives them higher participation weight, so as to better retain representative change information in the dimensionality reduction process. Compared with the traditional PCA, the attention-guided dimensionality reduction fusion effectively solves the problem of important details being submerged caused by global averaging, improves the sensitivity to non-stationary and sudden working conditions, and is especially suitable for high-speed rotating and strong nonlinear response of super-high lift centrifugal pump system. Specifically, the dimensionality reduction result has higher discriminability and compressibility, providing more compact and dynamic semantic input features for subsequent state recognition, strategy optimization and other modules, greatly improving the response accuracy and adaptive ability of the entire control system.

[0016] At the same time, the application accurately predicts the actual lift, hydraulic efficiency, theoretical flow deviation, shaft power and operating resistance change rate of the super-high lift single-stage centrifugal pump through the neural regression forest model, can realize the comprehensive characterization of the multi-dimensional performance of the super-high lift single-stage centrifugal pump, and the comprehensive performance score function can quantitatively evaluate multiple heterogeneous performance indicators, enhance the comparability and sensitivity of the decision. Further, the reinforcement learning model is used to dynamically decide the optimal control strategy based on the state vector, which can realize the adaptive adjustment of the frequency, main shaft speed and valve opening under varying working conditions, thereby improving the energy efficiency ratio, stability and life of the super-high lift single-stage centrifugal pump. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of an intelligent control multi-working condition super-high lift single-stage centrifugal pump control method provided by an embodiment of the application is shown. Figure 2 A structural diagram of a super-high lift single-stage centrifugal pump provided by an embodiment of the application is shown. DETAILED DESCRIPTION

[0018] The implementation, functional characteristics and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the application, and do not limit the application.

[0019] The embodiment of the present application provides a kind of intelligent control multi-working condition superhigh lift single-stage centrifugal pump control method.The execution subject of the kind of intelligent control multi-working condition superhigh lift single-stage centrifugal pump control method includes but is not limited to at least one of the electronic equipment that can be configured to execute the method provided by the embodiment of the present application, such as server, terminal and the like.In other words, the kind of intelligent control multi-working condition superhigh lift single-stage centrifugal pump control method can be executed by the software or hardware installed in terminal equipment or server equipment, and the software can be blockchain platform.The server includes but is not limited to: single server, server cluster, cloud server or cloud server cluster and the like.

[0020] Referring to Figure 1 And Figure 2 , embodiment 1 of the present application is: S1: periodically collect multi-source working condition data in the running process of superhigh lift single-stage centrifugal pump by using acquisition equipment, and perform attention-guided dimensionality reduction fusion on the multi-source working condition data to obtain multi-source working condition fusion tensor.

[0021] The acquisition equipment includes pressure sensor, electromagnetic flowmeter, current sensor, power meter, rotational speed / frequency sensor, temperature sensor and acceleration sensor; the multi-source working condition data in the running process of superhigh lift single-stage centrifugal pump is periodically collected by using the acquisition equipment, which includes: The multi-source working condition data includes various working condition data, which are in turn respectively the inlet instantaneous pressure data, outlet instantaneous pressure data, rotational speed data, frequency data, temperature data, acceleration data, current data, power data and instantaneous flow data of superhigh lift single-stage centrifugal pump; In the embodiment of the present application, the structure of the superhigh lift single-stage centrifugal pump includes: high specific speed closed impeller, double volute, main shaft and bearing assembly, multi-stage sealed pump body and variable frequency drive motor system; referring to Figure 2 As shown in the figure, it is the structure schematic diagram of superhigh lift single-stage centrifugal pump provided by an embodiment of the present application; Specifically, the high specific speed closed impeller adopts long blade closed structure, has larger impeller diameter, realizes larger fluid tangential velocity to obtain higher lift, and adopts corrosion-resistant or high-strength alloy material to withstand the stress generated by high-speed rotation; double volute is used for balancing radial force, reducing flow velocity, recovering kinetic energy and improving outlet pressure; the main shaft adopts heat-treated stainless steel or high-strength steel material to adapt to high speed and large load; the multi-stage sealed pump body adopts the combination of mechanical seal, labyrinth seal and balance pipe seal to prevent high-pressure side liquid leakage; the variable frequency drive motor system is matched with frequency converter, and is used for adjusting the variable frequency frequency, main shaft speed and valve opening degree of superhigh lift single-stage centrifugal pump; It should be noted that the pressure sensor is deployed on the outlet side and the inlet side of the ultra-high lift single-stage centrifugal pump to collect the inlet instantaneous pressure data and the outlet instantaneous pressure data of the ultra-high lift single-stage centrifugal pump, and the speed / frequency sensor is used to sense the feedback results of the frequency converter in the ultra-high lift single-stage centrifugal pump to obtain the speed data and the frequency data. The temperature sensor and the acceleration sensor are deployed in the main shaft area of ​​the ultra-high lift single-stage centrifugal pump to sense the main shaft temperature and the acceleration of the main shaft in the horizontal direction to obtain the temperature data and the acceleration data. The electromagnetic flowmeter is deployed inside the pump body of the ultra-high lift single-stage centrifugal pump to collect the instantaneous flow rate flowing through the pump body as the instantaneous flow rate data flowing through the ultra-high lift single-stage centrifugal pump. The current sensor and the power meter are used to collect the current data and power data of the ultra-high lift single-stage centrifugal pump. Specifically, instantaneous pressure data is used to evaluate changes in head and impedance, instantaneous flow data is used to characterize the load of ultra-high head single-stage centrifugal pumps, speed data and frequency data are used to characterize the motion state of the inverter in ultra-high head single-stage centrifugal pumps, current data and power data are used to characterize the energy consumption status of ultra-high head single-stage centrifugal pumps, and temperature data and acceleration data are used to characterize the structural stability of ultra-high head single-stage centrifugal pumps. The operating condition data in the multi-source operating condition data are all sequence data with a length of N, where N represents the number of sampling moments in the process of collecting the multi-source operating condition data. The time interval between adjacent sampling moments is set to 100 milliseconds, and N is set to 50. The missing values ​​in the multi-source working condition data are interpolated and completed using a linear weighted interpolation method to obtain the completed multi-source working condition data; Adaptive weighted differential filtering is used to perform differential filtering on the data values ​​of all working condition data in the completed multi-source working condition data. Specifically, the differential filtering process of the adaptive weighted differential filtering method is as follows: Extracting working condition data from the completed multi-source working condition data ,in, are the data values ​​collected from the 1st to the Nth sampling moments in the working condition data x, respectively. ; Calculate the local standard deviation of each data value in the working condition data x. The local standard deviation of is: ; in, Represents data values The local standard deviation, L represents the length of the local time window, set L to 5, ; The local standard deviation of the data value is converted into an adaptive weighting coefficient. The formula for converting to adaptive weighting coefficient is: ; in, Indicates the n-1th data value in the working condition data x The local standard deviation of Represents the local standard deviation The converted adaptive weighting coefficient; Adaptive weighted differential filtering is performed on the data values ​​using adaptive weighting coefficients, where the data values The adaptive weighted difference filtering formula is: ; in, Represents data values Adaptive weighted difference filtering result; Compared with the traditional first-order difference method with fixed coefficients, the adaptive weighted difference filtering method adaptively allocates difference weights according to the local change trend of the input data. Specifically, when the fluctuation is small (low-frequency drift is significant), the difference effect is enhanced. When the fluctuation is large (mutation exists), the weight is smoothly adjusted to reduce the sensitivity of the difference to high-frequency noise, thereby enhancing the response ability to small changes, retaining key structural features while suppressing high-frequency noise interference, and can effectively alleviate the information distortion problem caused by the original difference when processing non-stationary and noise-sensitive data.

[0022] Perform attention-guided dimensionality reduction fusion on multi-source operating condition data after differential filtering, including: Convert the multi-source working condition data after differential filtering into matrix form: ; in, for In matrix form with M rows and M columns, Located in the matrix The Row m column, Indicates the The working condition data is The differential filtering processing results of the data values ​​at the sampling moments are obtained. M represents the number of categories of the working condition data in the multi-source working condition data. The first to M-th working condition data are the inlet instantaneous pressure data, outlet instantaneous pressure data, speed data, frequency data, temperature data, acceleration data, current data, power data and instantaneous flow data of the ultra-high head single-stage centrifugal pump, respectively. M is 9. Indicates the sequence data length of the operating condition data in the multi-source operating condition data after differential filtering; Rows are represented as Indicates that the ultra-high head single-stage centrifugal pump is A vector composed of the data values ​​of M types of working condition data at sampling moments; Indicates the sequence data length of the operating condition data in the multi-source operating condition data after differential filtering; Calculate the time attention at each sampling moment in the multi-source working condition data in matrix form. The time attention at the sampling moment is calculated as follows: ; Where T represents transpose, Indicates the The temporal attention at each sampling moment is a scalar between 0 and 1; represents an exponential function with a natural constant as the base, Q represents a trainable query vector, Represents a trainable linear transformation matrix; is a scalar between 0 and 1; in the embodiment of the present invention, a linear transformation matrix is ​​used to transform Projecting from dimension M to the latent space, and using the query vector to generate the importance of the sampling moment, the trainable query vector is used to automatically identify key moments during the operation of ultra-high-lift single-stage centrifugal pumps, such as abnormal vibration, energy consumption peaks, and sudden changes in operating load. These key moments are assigned higher weights to ensure their maximum influence in subsequent feature extraction. The matrix Convert to temporal attention weighted covariance matrix : ; ; in, Representation matrix Specifically, the temporal attention weighted covariance matrix is ​​used to amplify the contribution of key moments in the statistical space, focusing on the fluctuation characteristics that are actually of diagnostic value, making abnormal and drastically changing time segments more decisive for the composition of the main direction of the feature. Weighted covariance matrix of temporal attention Perform eigendecomposition, select the eigenvectors corresponding to the largest K eigenvalues ​​to form the principal component matrix, and use the principal component matrix to transform the matrix Projected into low-dimensional space as a multi-source working condition fusion tensor, the multi-source working condition fusion tensor is in the form of a matrix with N-1 rows and K columns. Behavior Specifically, by compressing the original high-dimensional, multi-source working condition features into a set of smaller but more discriminative dynamic feature sequences, it provides high-quality input for subsequent clustering and control.

[0023] S2: Multi-scale feature extraction is performed on the multi-source operating condition fusion tensor to obtain the multi-scale operating characteristics of the ultra-high head single-stage centrifugal pump. The multi-scale operating characteristics are processed with time evolution using a density-based time clustering algorithm, and the multi-scale operating characteristics after time evolution are clustered into clusters.

[0024] Perform multi-scale feature extraction on multi-source working condition fusion tensors, including: Calculate the local similarity and global similarity of the multi-source dimensionality reduction projection vectors at any two sampling moments in the multi-source working condition fusion tensor, use the local similarity and global similarity as matrix elements, and construct the local adjacency graph and global adjacency graph of N-1 sampling moments respectively. The local adjacency graph and the global adjacency graph are both in the form of a matrix with N-1 rows and N-1 columns. As an embodiment of the present invention, the The multi-source dimension reduction projection vector at the sampling moment and the multi-source dimensionality reduction projection vector at the j-th sampling moment The local similarity and global similarity are calculated as follows: ; ; in, Represents multi-source dimensionality reduction projection vector Multi-source dimensionality reduction projection vector The local similarity of Represents multi-source dimensionality reduction projection vector Multi-source dimensionality reduction projection vector The global similarity of Represents the similarity scale coefficient, setting is 3, Indicates the local sampling time threshold, set is 5, L1 norm; represents the global similarity fusion coefficient (for example, setting is 0.5), Represents multi-source dimensionality reduction projection vector Multi-source dimensionality reduction projection vector The cosine similarity of is a trainable K-row and K-column projection matrix; The representation of the local adjacency graph is: , represents the local adjacency graph, and the representation of the global adjacency graph is: , represents the global adjacency graph; Specifically, by introducing local similarity and global similarity, the feature extraction capability of the working condition evolution pattern under different time scales is significantly enhanced, wherein the local adjacency graph accurately captures the sensitive changes of the super-high lift single-stage centrifugal pump in a short time, and effectively identifies transient anomalies; the global adjacency graph integrates the static geometric structure and dynamic projection information of the super-high lift single-stage centrifugal pump, and can learn the nonlinear dependence relationship between long-distance time steps. Specifically, the cosine similarity measures the direction consistency of the multi-source dimension reduction projection vector, emphasizes the trend commonality, and improves the perception ability of complex dynamic behaviors such as periodic instability and hysteresis effect.

[0025] The multi-source working condition fusion tensor, the local adjacency graph and the global adjacency graph are taken as inputs of the graph neural network. The graph neural network comprises a plurality of graph convolution layers nested in a hierarchical structure. A residual channel enhancement module is introduced between each graph convolution layer to enhance the stability and nonlinear expression capability of feature transmission. The residual channel enhancement module comprises an SE channel attention mechanism and a residual connection path. After the graph neural network is processed by multiple layers of graph convolution, the embedding features of the output of the last layer of graph convolution are output as multi-scale running features F representing the running state of the super-high lift single-stage centrifugal pump.

[0026] The graph convolution processing procedure based on the local adjacency graph and the global adjacency graph is as follows: ; ; ; Among them, represents the embedding features output by the y+1th graph convolution layer in the graph neural network, represents the embedding features output by the yth graph convolution layer in the graph neural network, represents the local adjacency graph, represents the degree matrix of the local adjacency graph , represents the global adjacency graph, represents the degree matrix of the global adjacency graph , represents the unit matrix, represents the normalization result of the local adjacency graph , represents the normalization result of the global adjacency graph ; represents an activation function, and the activation function is set to ReLU function; is a trainable local adjacency convolution matrix, represents the trainable global adjacency convolution matrix; The graph convolution operation of the dual-graph fusion fully integrates local and global information, thereby improving the robust modeling capability of the pump group for non-stationary sequences under multiple working conditions.

[0027] A density-based time clustering algorithm is used to process the time evolution of multi-scale operation features, and the multi-scale operation features after time evolution are clustered into clusters, including: The time evolution factor of the multi-scale operation feature F is generated using a density-based time clustering algorithm: ; in, represents the time evolution factor of the multi-scale operation feature F, represents an exponential function with a natural constant as the base, Indicates the current number of acquisitions of multi-source working condition data. Indicates the preset maximum number of collection times, time indicates the total running time of the ultra-high head single-stage centrifugal pump, and TIME indicates the maximum running time of the ultra-high head single-stage centrifugal pump. are time evolution coefficients, set They are 0.4, 0.6 respectively; Using time evolution factors Perform time evolution processing on the multi-scale operation feature F to obtain the multi-scale operation feature after time evolution processing , Indicates the control coefficient, set is 0.1; Calculate the multi-scale operation characteristics after time evolution processing The Euclidean distance between the core points of different clusters, the multi-scale operation characteristics after time evolution processing Cluster into the cluster with the closest Euclidean distance.

[0028] S3: The operating condition corresponding to the cluster to which the multi-scale operating characteristics belong is taken as the current operating condition of the ultra-high head single-stage centrifugal pump. Based on the current operating condition and the multi-scale operating characteristics, the neural regression forest is used to output the performance parameters of the ultra-high head single-stage centrifugal pump under the current operating condition, and the comprehensive performance scoring function is used to convert the performance parameters into a performance score.

[0029] The operating condition labels of the clustered clusters are used as the current operating conditions of the ultra-high head single-stage centrifugal pump; It should be noted that by pre-collecting multi-scale operating features under various working conditions and clustering them using the DBSCAN algorithm, multiple clusters and the core points of the clusters are obtained. The proportion of working condition labels under each cluster is counted, and the working condition label with the highest proportion is selected as the working condition label of the cluster.

[0030] Based on the current working condition and multi-scale operation characteristics of the super-high lift single-stage centrifugal pump, the neural regression forest outputs the performance parameters of the super-high lift single-stage centrifugal pump under the current working condition, including: The neural regression forest includes a plurality of parallel regression tree structures, and the split function of each regression tree is parameterized and modeled by a neural network, so that it has stronger nonlinear fitting capability; the current working condition and multi-scale operation characteristics of the super-high lift single-stage centrifugal pump are jointly constructed into a high-dimensional feature vector, and the high-dimensional feature vector is taken as the input of the regression tree in the neural regression forest; each regression tree performs nonlinear regression mapping on the high-dimensional feature vector through the embedded shallow neural network to obtain the regression output of each regression tree, wherein the regression output is the mapping result of the high-dimensional feature vector to the performance parameter dimension; the performance parameters of the super-high lift single-stage centrifugal pump under the current working condition are obtained by fusing the regression outputs of the plurality of regression trees, and the performance parameters include actual lift, hydraulic efficiency, theoretical flow deviation, shaft power and running resistance change rate, and the adaptive modeling of the relationship between the non-stationary and nonlinear operation characteristics and the performance response can be realized.

[0031] As an embodiment of the present application, the neural regression forest is trained offline using a historical operation data set of the super-high lift centrifugal pump to obtain an initial performance prediction model, and in the real-time operation process of the super-high lift centrifugal pump, new working condition parameters and performance feedback data are continuously collected as online learning samples to incrementally update part of the regression trees in the neural regression forest, so that the neural regression forest can adapt to changes in the operating environment in a timely manner. For example, when the hydraulic efficiency of the super-high lift centrifugal pump decreases due to changes in water quality, the updated neural regression forest can automatically correct the performance prediction and output shaft power and running resistance change rate closer to the actual operation. In this way, the neural regression forest can maintain prediction accuracy and generalization ability under non-stationary working conditions and long-term operation, and improve the intelligent operation control level of the super-high lift centrifugal pump.

[0032] The performance parameters are converted into performance scores by using a comprehensive performance score function, including: The performance parameters are normalized; optionally, the normalization method is the maximum-minimum normalization method; The normalized performance parameters are taken as the input of the comprehensive performance score function to obtain the performance scores corresponding to the performance parameters: ; Wherein, The normalized performance parameter vector includes the normalized actual lift , the normalized hydraulic efficiency , the normalized theoretical flow deviation , and the normalized shaft power and the running resistance change rate , denotes the performance parameter vector after normalization processing corresponding performance score are all score coefficients, and are set are all equal and are 0.2.

[0033] S4: taking the current working condition, performance parameter and performance score of the super-high-lift single-stage centrifugal pump as a state vector, selecting an optimal control vector corresponding to the state vector from the control space by using a reinforcement learning model, and adjusting the frequency conversion frequency, main shaft speed and valve opening degree of the super-high-lift single-stage centrifugal pump by using the optimal control vector.

[0034] constructing a state vector as the input of the reinforcement learning model: [current working condition, performance parameter, performance score of the super-high-lift single-stage centrifugal pump]; constructing a control space C of the super-high-lift single-stage centrifugal pump, the control space C including the following three continuous control dimensions: frequency conversion frequency of the frequency converter in the super-high-lift single-stage centrifugal pump ; main shaft speed ; valve opening degree of the outlet of the super-high-lift single-stage centrifugal pump ; wherein, is the output range of the frequency conversion frequency in the control space C, denotes the minimum output value (for example, 20 Hz) of the frequency conversion frequency, denotes the maximum output value (for example, 60 Hz) of the frequency conversion frequency, is the output range of the main shaft speed in the control space C, denotes the minimum output value (for example, 30% rated speed) of the main shaft speed, denotes the maximum output value (for example, 120% rated speed) of the main shaft speed, is the output range of the valve opening degree in the control space C. constructing a policy network in the reinforcement learning model by using a policy gradient type reinforcement learning method, the policy network taking the state vector as the input and selecting the frequency conversion frequency, main shaft speed and valve opening degree optimal for the state vector from the three continuous control dimensions in the control space C to form an optimal control vector, and adjusting the frequency conversion frequency, main shaft speed and valve opening degree of the super-high-lift single-stage centrifugal pump by using the optimal control vector.

[0035] As an embodiment of the present application, the training process of the policy network is: Construct a simulation environment for an ultra-high head single-stage centrifugal pump. Specifically, refer to Figure 2 The structural schematic diagram of the ultra-high-head single-stage centrifugal pump is shown. A three-dimensional structural model of the ultra-high-head single-stage centrifugal pump is generated in the simulation environment. High-quality volume meshing is performed on the three-dimensional structural model, including the use of structured or near-structured meshes in the high-specific-speed closed impeller region to ensure blade boundary layer resolution, dense meshing in the gap / gap region, and unstructured transition meshing around the rotor. The continuity and momentum conservation of the fluid in the pump chamber are modeled using the fluid mechanics equations (Navier–Stokes equations), thereby obtaining the velocity distribution, pressure distribution, and energy transfer laws of the fluid in the ultra-high-head centrifugal pump. Combined with the energy transfer process of the high-specific-speed closed impeller to the fluid and referring to the Euler turbine equation, a theoretical head model of the ultra-high-head centrifugal pump is constructed to reflect the energy enhancement effect provided by the high-specific-speed closed impeller to the fluid. At the same time, combined with the performance curves of the ultra-high-head centrifugal pump obtained from actual tests, the theoretical model is compared and corrected with the empirical curves, so that the three-dimensional structural model in the simulation environment can accurately reflect the actual operating performance of the ultra-high-head centrifugal pump under different control conditions such as flow rate, speed, and valve opening. Initialize the trainable parameters in the policy network, which is used to receive the state vector, calculate the policy selection probability of the control vector that meets the variable frequency, spindle speed, and valve opening constraints in the control space C, and select the control vector with the highest policy selection probability as the optimal control vector corresponding to the state vector; Extract the state vector of the current 3D model structure in the simulation environment, use the performance score of the state vector after taking the control vector as the reward function, use the policy optimization algorithm to interactively iterate the state vector and trainable parameters until the preset maximum number of interactive iterations (50 times) is reached, and construct a policy network based on the trainable parameters obtained from the final iteration. The trainable parameters of the policy network include weight matrix parameters and bias parameters; The interactive iterative process is as follows: the policy network generates policy selection probabilities for different control vectors under the current state vector, selects the control vector with the highest policy selection probability to optimize the control of the 3D model structure, generates the state vector of the 3D model structure in the new state, calculates the reward function value of the selected control vector, and iteratively updates the trainable parameters using a policy optimization method; Optionally, the policy optimization method adopts a PPO (Proximal Policy Optimization) method, a REINFORCE with baseline (REINFORCE policy optimization method with baseline) or an AdvantageActor-Critic method.

[0036] It should be noted that the neural regression forest model can accurately predict the key performance parameters of the super-high lift single-stage centrifugal pump, such as actual lift, hydraulic efficiency, theoretical flow deviation, shaft power and operating resistance change rate, so as to comprehensively characterize the multi-dimensional performance of the super-high lift single-stage centrifugal pump, and the comprehensive performance scoring function can quantitatively evaluate the multiple heterogeneous performance indicators, thereby enhancing the comparability and sensitivity of the decision. Further, the reinforcement learning model can dynamically decide the optimal control strategy based on the state vector, so as to realize the adaptive adjustment of the frequency, shaft speed and valve opening of the variable frequency under the changing conditions, thereby improving the energy efficiency ratio, stability and service life of the super-high lift single-stage centrifugal pump.

[0037] As a preferred embodiment of the present application, the actual lift is the energy height of the actual output of the super-high lift single-stage centrifugal pump, and insufficient lift will lead to substandard fluid delivery, and too high lift will waste energy, both of which will reduce energy efficiency. High-lift operation is often accompanied by high load, which can easily aggravate impeller fatigue and bearing wear. The hydraulic efficiency is the proportion of the input shaft power used for water delivery, and the hydraulic efficiency directly determines the effective delivery capacity under unit power consumption. Low efficiency operation is often accompanied by unstable factors such as turbulent flow and cavitation, which affects the stability of the super-high lift single-stage centrifugal pump. The theoretical flow deviation is the difference between the actual flow and the theoretically designed flow, and a large deviation means that the pump operation deviates from the optimal operating point, and the energy utilization rate decreases. Long-term non-theoretical flow operation can easily lead to component fatigue or local overheating. The shaft power is the total power input required to drive the super-high lift single-stage centrifugal pump to operate, and continuous high shaft power operation will overload the motor and bearing, accelerating aging. The operating resistance change rate is the change rate of the resistance of the fluid in the pump cavity and the pipeline system of the super-high lift single-stage centrifugal pump. Sudden resistance means that the system is blocked, scaled or cavitated, which is prone to cause vibration. The above performance indicators all obviously affect the energy efficiency ratio, stability and operating life of the super-high lift single-stage centrifugal pump.

[0038] It should be understood that the above-mentioned embodiments are only for illustration, and the scope of the patent application is not limited by the structure.

[0039] It should be noted that the above-mentioned embodiment numbers of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. And the terms "include", "contain" or any other variants in this paper are intended to cover non-exclusive inclusion, so that the process, device, article or method including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, device, article or method. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, device, article or method including the element.

[0040] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, also can be through hardware, but in many cases the former is the better embodiment. Based on such understanding, the technical solutions of the present application essentially or say the part of the contribution to the prior art can be embodied in the form of software products, the computer software product is stored in a storage medium (such as ROM / RAM, magnetic disc, optical disc) as described above, including a number of instructions to make a terminal device (may be a mobile phone, computer, server, or network equipment, etc.) executes the method described in various embodiments of the present application.

[0041] The above is only the preferred embodiment of the present application, not therefore limit the patent scope of the present application, any equivalent structure or equivalent flow transformation made by using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for controlling a single-stage centrifugal pump with intelligent control and multi-operating conditions and ultra-high head, characterized in that: The method comprises: S1: Use acquisition equipment to periodically collect multi-source operating condition data during the operation of an ultra-high-lift single-stage centrifugal pump, and perform attention-guided dimensionality reduction fusion on the multi-source operating condition data to obtain a multi-source operating condition fusion tensor; S2: Perform multi-scale feature extraction on the multi-source operating condition fusion tensor to obtain the multi-scale operating characteristics of the ultra-high head single-stage centrifugal pump. Use a density-based time clustering algorithm to perform time evolution processing on the multi-scale operating characteristics, and cluster the multi-scale operating characteristics after time evolution processing into clusters; S3: The operating condition corresponding to the cluster to which the multi-scale operating characteristics belong is taken as the current operating condition of the ultra-high head single-stage centrifugal pump. Based on the current operating condition and the multi-scale operating characteristics, the neural regression forest is used to output the performance parameters of the ultra-high head single-stage centrifugal pump under the current operating condition. The performance parameters are converted into a performance score using a comprehensive performance scoring function. S4: The current operating conditions, performance parameters, and performance scores of the ultra-high lift single-stage centrifugal pump are used as the state vector. The reinforcement learning model is used to select the optimal control vector corresponding to the state vector from the control space. The optimal control vector is then used to adjust the variable frequency, spindle speed, and valve opening of the ultra-high lift single-stage centrifugal pump.

2. The intelligent control method for a multi-operating-condition ultra-high-lift single-stage centrifugal pump according to claim 1, characterized in that: The acquisition equipment includes a pressure sensor, an electromagnetic flowmeter, a current sensor, a power meter, a speed / frequency sensor, a temperature sensor, and an acceleration sensor; the acquisition equipment is used to periodically acquire multi-source operating condition data during the operation of the ultra-high head single-stage centrifugal pump, including: The multi-source operating condition data includes a variety of operating condition data, which are respectively the inlet instantaneous pressure data, outlet instantaneous pressure data, speed data, frequency data, temperature data, acceleration data, current data, power data of the ultra-high head single-stage centrifugal pump and the instantaneous flow data of the ultra-high head single-stage centrifugal pump; The operating condition data in the multi-source operating condition data are all sequence data with a length of N, where N represents the number of sampling moments in the process of collecting the multi-source operating condition data. The time interval between adjacent sampling moments is set to 100 milliseconds, and N is set to 50. The missing values ​​in the multi-source working condition data are interpolated and completed using a linear weighted interpolation method to obtain the completed multi-source working condition data; Adaptive weighted differential filtering is used to perform differential filtering on the data values ​​of all working condition data in the completed multi-source working condition data.

3. The intelligent control method for a multi-operating-condition ultra-high-lift single-stage centrifugal pump according to claim 2, characterized in that: Perform attention-guided dimensionality reduction fusion on multi-source operating condition data after differential filtering, including: Convert the multi-source working condition data after differential filtering into matrix form: ； in, for In matrix form with M rows and M columns, Located in the matrix The Row m column, Indicates the The working condition data is The differential filtering processing results of the data values ​​at the sampling moments are obtained. M represents the number of categories of the working condition data in the multi-source working condition data. The first to M-th working condition data are the inlet instantaneous pressure data, outlet instantaneous pressure data, speed data, frequency data, temperature data, acceleration data, current data, power data and instantaneous flow data of the ultra-high head single-stage centrifugal pump, respectively. M is 9. Indicates the sequence data length of the operating condition data in the multi-source operating condition data after differential filtering; The Rows are represented as Indicates that the ultra-high head single-stage centrifugal pump is A vector composed of the data values ​​of M types of working condition data at sampling moments; Calculate the time attention of each sampling moment in the multi-source working condition data in matrix form, where the time attention at the sampling moment is calculated as follows: ； Where T represents transpose, Indicates the The temporal attention at each sampling moment is a scalar between 0 and 1; represents an exponential function with a natural constant as the base, Q represents a trainable query vector, Represents a trainable linear transformation matrix; The matrix Convert to temporal attention weighted covariance matrix ; Weighted covariance matrix of temporal attention Perform eigendecomposition, select the eigenvectors corresponding to the largest K eigenvalues ​​to form the principal component matrix, and use the principal component matrix to transform the matrix Projected into low-dimensional space as a multi-source working condition fusion tensor, the multi-source working condition fusion tensor is in the form of a matrix with N-1 rows and K columns. Behavior The multi-source dimension reduction projection vector at each sampling moment.

4. The intelligent control method for a multi-operating-mode ultra-high-lift single-stage centrifugal pump according to claim 3, characterized in that: Perform multi-scale feature extraction on multi-source working condition fusion tensors, including: Calculate the local similarity and global similarity of the multi-source dimensionality reduction projection vectors at any two sampling moments in the multi-source working condition fusion tensor, use the local similarity and global similarity as matrix elements, and construct the local adjacency graph and global adjacency graph of N-1 sampling moments respectively. The local adjacency graph and the global adjacency graph are both in the form of a matrix with N-1 rows and N-1 columns. The multi-source working condition fusion tensor, local adjacency graph, and global adjacency graph are used as inputs to the graph neural network. The graph neural network includes a plurality of graph convolutional layers nested in a hierarchical structure; A residual channel enhancement module is introduced between each graph convolutional layer to enhance the stability and nonlinear expression ability of feature transfer; The residual channel enhancement module includes a SE channel attention mechanism and a residual connection path; After the graph neural network is processed by multi-layer graph convolution, the embedded features of the last graph convolution layer are output as the multi-scale operation features F that characterize the operating state of the ultra-high lift single-stage centrifugal pump.

5. The intelligent control method for multi-operating-condition ultra-high lift single-stage centrifugal pump according to claim 4, characterized in that: A density-based time clustering algorithm is used to process the time evolution of multi-scale operation features, and the multi-scale operation features after time evolution are clustered into clusters, including: The time evolution factor of the multi-scale operation feature F is generated using a density-based time clustering algorithm. ; Using time evolution factors Perform time evolution processing on the multi-scale operation feature F to obtain the multi-scale operation feature after time evolution processing , Indicates the control coefficient, set is 0.1; Calculate the multi-scale operation characteristics after time evolution processing The Euclidean distance between the core points of different clusters, the multi-scale operation characteristics after time evolution processing Cluster them into the cluster with the closest Euclidean distance, and use the operating condition label of the cluster as the current operating condition of the ultra-high head single-stage centrifugal pump.

6. The intelligent control method for a multi-operating-mode ultra-high-lift single-stage centrifugal pump according to claim 1, characterized in that: Based on the current operating conditions and multi-scale operating characteristics of the ultra-high head single-stage centrifugal pump, the neural regression forest is used to output the performance parameters of the ultra-high head single-stage centrifugal pump under the current operating conditions, including: The neural regression forest includes multiple parallel regression tree structures, and the splitting function of each regression tree is parameterized and modeled through a neural network, so that it has stronger nonlinear fitting ability; the current operating conditions and multi-scale operating characteristics of the ultra-high head single-stage centrifugal pump are jointly constructed into a high-dimensional feature vector, and the high-dimensional feature vector is used as the input of the regression tree in the neural regression forest. Each regression tree performs nonlinear regression mapping on the high-dimensional feature vector through an embedded shallow neural network to obtain the regression output of each regression tree, wherein the regression output is the mapping result of the high-dimensional feature vector to the performance parameter dimension; by fusing the regression outputs of multiple regression trees, the performance parameters of the ultra-high head single-stage centrifugal pump under the current operating conditions are obtained, and the performance parameters include actual head, hydraulic efficiency, theoretical flow deviation, shaft power and operating resistance change rate.

7. The intelligent control method for a multi-operating-mode ultra-high-lift single-stage centrifugal pump according to claim 6, characterized in that: The performance parameters are converted into performance scores using a comprehensive performance scoring function, including: performing normalization processing on the performance parameters; The normalized performance parameters are used as the input of the comprehensive performance scoring function to obtain the performance scores corresponding to the performance parameters: ； in, represents the normalized performance parameter vector, which includes the normalized actual head , hydraulic efficiency , theoretical flow deviation , shaft power And the rate of change of running resistance , Represents the performance parameter vector after normalization The corresponding performance score, Both are rating coefficients.

8. The intelligent control method for a multi-operating-mode ultra-high-lift single-stage centrifugal pump according to claim 1, characterized in that: The reinforcement learning model is used to select the optimal control vector corresponding to the state vector from the control space, including: Construct the state vector as the input of the reinforcement learning model: [current operating conditions, performance parameters, and performance score of the ultra-high head single-stage centrifugal pump]; A control space C for an ultra-high head single-stage centrifugal pump is constructed, and the control space C includes the following three continuous control dimensions: Frequency conversion of inverter in ultra-high head single-stage centrifugal pump ; Spindle speed ; Valve opening at the outlet of ultra-high lift single-stage centrifugal pump ; in, To control the frequency conversion in space C The output range of is the minimum output value of the variable frequency, is the maximum output value of the variable frequency, To control the spindle speed in space C The output range of Indicates the minimum output value of the spindle speed. Indicates the maximum output value of the spindle speed. To control the valve opening in space C Output range; The policy network in the reinforcement learning model is constructed using a policy gradient reinforcement learning method. The policy network takes a state vector as input and selects the optimal frequency conversion frequency, spindle speed, and valve opening corresponding to the state vector in three continuous control dimensions from the control space C to form an optimal control vector. The optimal control vector is used to adjust the frequency conversion frequency, spindle speed, and valve opening of an ultra-high lift single-stage centrifugal pump.

Images