Centrifugal water pump control method based on multi-source data fusion and self-adaptive model

Abstract

The invention provides a centrifugal water pump control method based on multi-source data fusion and an adaptive model. The control method comprises the steps that static data of a water pump motor, dynamic environment data of a pipe network and a water pump performance curve of a water pump are obtained; based on the static data, the pipe network dynamic environment data and the water pump performance curve, constructing a target motor-pump combined model; the target motor-pump combined model is a multi-dimensional model constructed based on a Gaussian process regression model; and dynamic data of a water pump motor are obtained, and a real-time working mode of the water pump is selected according to the dynamic data and the target motor-pump combined model. By means of the control method, the problems of sensing blind areas, control lag and energy efficiency black holes existing in an existing centrifugal water pump control technology are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of centrifugal pumps, and particularly relates to a control method for centrifugal pumps based on multi-source data fusion and an adaptive model. Background Art

[0002] In industrial fluid transportation systems, as a core power device, the control technology of centrifugal pumps directly affects the operation efficiency and stability of the system. The current mainstream control schemes are mainly based on traditional sensor monitoring and fixed parameter adjustment mechanisms.

[0003] However, the current control of centrifugal pumps has the following problems:

[0004] Perception blind spot: Relying on external sensors such as flow meters and differential pressure transmitters, the installation is complex and the failure rate is high (MTBF < 2 years).

[0005] Control lag: PID regulation relies on empirical parameters and cannot adapt to the dynamically changing characteristics of the pipe network (such as sudden changes in valve opening).

[0006] Energy efficiency black hole: Data from the International Energy Agency (IEA) shows that only 20% of pumps operate in the high-efficiency zone (BEP), and the annual average ineffective energy consumption exceeds 200 billion kWh. Coarse adjustment: More than 70% of users set the speed only based on experience, resulting in a redundant power consumption ratio exceeding 40%. Due to the lack of real-time monitoring of pump efficiency, enterprises often cannot accurately adjust the operating parameters of the pumps, resulting in high energy consumption. Summary of the Invention

[0007] The present application provides a control method for centrifugal pumps based on multi-source data fusion and an adaptive model to solve the problems of perception blind spot, control lag, and energy efficiency black hole existing in the existing centrifugal pump control technology.

[0008] The control method includes:

[0009] Obtain the static data of the pump motor, the dynamic data of the pipe network environment, and the pump performance curve of the pump;

[0010] Construct a target motor-pump combined model based on the static data, the dynamic data of the pipe network environment, and the pump performance curve; the target motor-pump combined model is a multi-dimensional model constructed based on a Gaussian process regression model;

[0011] Obtain the dynamic data of the pump motor, and select the real-time working mode of the pump according to the dynamic data and the target motor-pump combined model.

[0012] Preferably, the steps of obtaining the static data of the pump motor, the dynamic data of the pipe network environment, the pump performance curve of the pump, and the dynamic data of the pump motor include:

[0013] Obtain the static data, the dynamic data, the dynamic environment data of the pipe network, and the pump performance curve through the data perception layer.

[0014] Preferably, the data perception layer includes:

[0015] A static data acquisition layer, which is configured to obtain the static data of the pump motor and the pump performance curve of the pump;

[0016] A dynamic data acquisition layer, which is configured to obtain the dynamic data of the pump motor;

[0017] A pipe network environment data acquisition layer, which is configured to obtain the dynamic environment data of the pipe network.

[0018] Preferably, the static data includes the rated power, rated current, rated voltage, rated speed, power factor, motor efficiency, and frequency reference value of the pump motor;

[0019] The pump performance curve is a mapping table between pump efficiency, head, and flow rate;

[0020] The dynamic environment data of the pipe network is used to characterize the internal fluid resistance of the pipe connected to the pump.

[0021] Preferably, the steps of constructing the target motor-pump combined model further include:

[0022] Perform continuous modeling on the pump performance curve using Gaussian process regression. During the continuous modeling process, use the static data and the dynamic environment data of the pipe network as modeling parameters, and use the dynamic data as input variables and output variables.

[0023] Preferably, the dynamic data includes the real-time current of the pump motor, the real-time speed of the pump motor, and the real-time shaft power of the pump;

[0024] The steps of constructing the target motor-pump combined model further include using the real-time current and the real-time speed as the input variables, and using the real-time shaft power as the output variable.

[0025] Preferably, the dynamic data acquisition layer is further configured to obtain historical data of the pump under variable frequency conditions and rated conditions;

[0026] Use the historical data to train the motor-pump combined model to obtain the target motor-pump combined model.

[0027] Preferably, the target motor-pump combined model includes a first-layer model and a second-layer model;

[0028] The first-layer model is the mapping relationship among the shaft power, flow rate, and head of the water pump at the rated speed of the water pump motor;

[0029] The second-layer model is a full-condition three-dimensional model constructed based on the speed of the water pump motor as a variable, the speed of the water pump motor, the shaft power of the water pump, the flow rate of the water pump, and the head of the water pump.

[0030] Preferably, the second-layer model is constructed based on the radial basis function and the motor similarity law, and the full-condition three-dimensional model is a mixed kernel function;

[0031] In the second-layer model, the flow rate of the water pump is directly proportional to the speed of the water pump motor, and the head of the water pump is related to the square of the speed of the water pump motor.

[0032] Preferably, the working real-time mode includes a constant flow rate control mode, a constant head control mode, and an energy efficiency optimal control mode;

[0033] The control method further includes:

[0034] Setting an objective function for the target motor-pump combined model, and using an improved particle swarm optimization algorithm to search for the global optimal speed to obtain the speed of the water pump motor corresponding to the energy efficiency optimal control mode.

[0035] As can be seen from the above, the present application provides a centrifugal water pump control method based on multi-source data fusion and an adaptive model. The control method includes obtaining static data of the water pump motor, dynamic environment data of the pipe network, and the water pump performance curve of the water pump; constructing a target motor-pump combined model based on the static data, the dynamic environment data of the pipe network, and the water pump performance curve; the target motor-pump combined model is a multi-dimensional model constructed based on the Gaussian process regression model; obtaining dynamic data of the water pump motor, and selecting the working real-time mode of the water pump according to the dynamic data and the target motor-pump combined model. The present application solves the problems of perception blind spots, control lag, and energy efficiency black holes existing in the existing centrifugal water pump control technology through the above control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a flowchart of a centrifugal water pump control method based on multi-source data fusion and an adaptive model of the present application;

[0038] Figure 2Schematic diagram of the data perception layer in a centrifugal pump control method based on multi-source data fusion and adaptive model of the present application;

[0039] Figure 3 Schematic diagram of the principle of a centrifugal pump control method based on multi-source data fusion and adaptive model of the present application. Detailed implementation manners

[0040] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0041] It should be noted that the brief description of the terms in the present application is only for the convenience of understanding the following described implementation manners, rather than intending to limit the implementation manners of the present application. Unless otherwise specified, these terms should be understood in their ordinary and general meanings.

[0042] It should be noted that in the present application, words such as "exemplary" or "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, using words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0043] In the field of fluid transportation, centrifugal pumps, as key equipment, are widely used in many scenarios such as industrial production, building water supply, and agricultural irrigation. However, many problems that need to be solved urgently have emerged in the actual application of traditional centrifugal pump control technologies. These problems seriously affect the operation efficiency, stability and reliability of the pump system and become a bottleneck restricting the development of the industry.

[0044] (1) Traditional centrifugal pump control systems mainly rely on external sensors such as flow meters and differential pressure transmitters to obtain relevant parameters of pump operation. The installation process of these external sensors is extremely complex. Not only professional technicians are required to operate, but also many factors need to be considered during the installation process, such as the installation position of the sensor and the connection method with the pump. Once the installation is improper, it may cause the sensor to be unable to accurately obtain data, affecting the normal operation of the entire control system.

[0045] More seriously, the failure rate of these external sensors is extremely high. Due to their long-term exposure to the external environment, they are easily affected by factors such as temperature, humidity, and dust, resulting in a gradual decline in the performance of the sensors and even failures. Generally speaking, the mean time between failures (MTBF) of these external sensors is relatively short, usually less than 2 years. Frequent failures not only increase the maintenance cost of the equipment but also may cause the water pump system to malfunction at critical moments, bringing great inconvenience to the production and life of enterprises.

[0046] (2) The commonly used PID regulation method in traditional centrifugal water pump control relies heavily on empirical parameters. In practical applications, the characteristics of the pipe network are dynamically changing. For example, the opening degree of the valve may suddenly change. However, the PID regulation based on empirical parameters cannot adapt to this dynamic change in a timely manner. When the characteristics of the pipe network change, the PID regulation requires a certain amount of time to readjust the parameters, resulting in a control lag phenomenon. This control lag will cause the operating state of the water pump to fail to keep up with the changes in the pipe network characteristics in a timely manner, thus affecting the conveying efficiency and stability of the water pump, and even may cause problems such as overload and vibration of the water pump, shortening the service life of the water pump.

[0047] (3) From a global perspective, the energy consumption problem of centrifugal water pumps is very prominent. Relevant data from the International Energy Agency (IEA) show that only 20% of water pumps can operate in the high-efficiency area (BEP). This means that most water pumps are in an inefficient state during actual operation, resulting in a large amount of energy waste. According to statistics, the annual ineffective energy consumption of centrifugal water pumps exceeds 200 billion kWh. Such a huge amount of energy consumption not only increases the operating cost of enterprises but also causes a serious burden on the environment.

[0048] (4) In practical applications, more than 70% of users set the water pump speed solely based on experience. This rough adjustment method lacks a scientific basis, resulting in the water pump speed often not matching the actual demand. When the speed is set too high, the water pump will consume too much energy, causing redundant power consumption; while when the speed is set too low, it cannot meet the actual conveying demand. According to statistics, due to this rough adjustment method, the redundant power consumption of the water pump accounts for more than 40%.

[0049] In addition, due to the lack of real-time monitoring of the water pump efficiency in the traditional centrifugal water pump control system, enterprises cannot timely understand the operating efficiency status of the water pump, and thus cannot accurately adjust the operating parameters of the water pump. This causes the water pump to operate in an inefficient state for a long time, with high energy consumption, seriously affecting the economic and social benefits of enterprises.

[0050] Based on the above problems, the present application provides the following implementation manners.

[0051] Figure 1This is a flowchart of a centrifugal pump control method based on multi-source data fusion and an adaptive model in this application.

[0052] Figure 3 This is a schematic diagram of a centrifugal pump control method based on multi-source data fusion and an adaptive model in this application.

[0053] See Figure 1 and Figure 3 It can be known that this embodiment provides a centrifugal pump control method based on multi-source data fusion and an adaptive model. The control method includes:

[0054] S100, obtain the static data of the pump motor, the dynamic data of the pipe network environment, and the pump performance curve of the pump. Specifically, in this embodiment, before controlling the pump, it is necessary to first obtain the relevant data of the corresponding pump, that is, obtain the static data of the pump motor, the dynamic data of the pipe network environment, and the pump performance curve of the pump.

[0055] The static data of the pump motor and the pump performance curve of the pump both belong to the inherent parameters of the corresponding pump. Therefore, they can be directly obtained by reading the nameplate of the pump. The static data can be understood as the rated parameters of the pump, specifically including the rated power, rated current, rated voltage, rated speed, power factor, motor efficiency, and frequency reference value of the pump motor; the pump performance curve is used to characterize the performance of the pump. For a pump, its main performance is the flow rate, head, and the required energy. Therefore, the pump performance curve is a mapping table used to characterize the relationship between pump efficiency, head, and flow rate.

[0056] Considering that the same pump will inevitably be affected by external factors such as the opening or closing of the valve of the pipeline connected to the pump, the blockage of the pipeline, and / or the rupture of the pipeline during long-term use. Therefore, if more accurate control of the pump is desired, it is necessary to obtain the situation of the external factors affecting the pump operation. Therefore, this embodiment obtains the dynamic data of the pipe network environment, and uses the dynamic data of the pipe network environment to characterize the internal fluid resistance of the pipeline connected to the pump, so as to add external factors to the pump control process, thereby realizing accurate control of the pump.

[0057] It should be noted that the static data and pump performance curves of different pumps are different, and the ways to obtain these data can also be different. The obtaining methods include but are not limited to scanning the QR code on the nameplate information of the pump and automatically writing the data, and manually referring to the nameplate to write the data.

[0058] The control method further includes:

[0059] S200. Construct a target motor-pump combined model based on the static data, the dynamic environment data of the pipe network, and the performance curve of the water pump. The target motor-pump combined model is a multi-dimensional model constructed based on the Gaussian process regression model.

[0060] Specifically, in this embodiment, since the existing centrifugal water pump control technology needs to rely on a large number of sensors to obtain data such as flow rate and pressure difference to achieve the control of the water pump, it is necessary to frequently replace the sensor equipment, and it cannot adapt to sudden changes in pipe network valves, resulting in a lag in the control of the water pump. Therefore, a target motor-pump combined model for predicting the operation of the water pump is proposed. Through pre-training, the operation conditions of the water pump under all working conditions are used as training data, and thus a single model can be used to automatically control the water pump.

[0061] Among them, the target motor-pump combined model is constructed through the Gaussian process regression model, and thus the model can simultaneously capture the non-linear characteristics of the data and physical laws, thereby improving the accuracy of the control method.

[0062] The control method further includes:

[0063] S300. Obtain the dynamic data of the water pump motor, and select the real-time working mode of the water pump according to the dynamic data and the target motor-pump combined model.

[0064] Specifically, in this embodiment, based on step S200, after the target motor-pump combined model is constructed, the real-time working mode of the water pump can be adjusted according to the user's set requirements. Among them, the real-time working mode includes a constant flow control mode, a constant head control mode, and an energy efficiency optimal control mode.

[0065] It should be noted that the constant flow control mode, the constant head control mode, and the energy efficiency optimal control mode are only general descriptions of the working modes of the water pump. Combining with the control method of this embodiment, these modes can be understood as adjusting the flow rate, head, or energy efficiency to the specified values required by the user. The energy efficiency optimal control mode is the working mode of the water pump based on the premise that the user aims to improve energy efficiency. If the user's specified requirements for constant flow or constant head exist, it is necessary to find the corresponding best energy efficiency under this requirement according to the user's needs.

[0066] Based on Figure 3 it can be seen that, by way of example, if a constant head is to be set, the head can be set to the fixed value required, and the target motor-pump combined model is used to calculate other parameters, and these parameters are used as the operating parameters of the water pump and the water pump motor.

[0067] Among them, before selecting the real-time working mode of the water pump, it is also necessary to set an objective function for the target motor-pump combined model, and use an improved particle swarm optimization algorithm to search for the global optimal rotational speed to obtain the rotational speed of the water pump motor corresponding to the energy efficiency optimal control mode.

[0068] Figure 2 It is a schematic diagram of the data perception layer in a centrifugal water pump control method based on multi-source data fusion and adaptive model of the present application.

[0069] See Figure 2 It can be known that, further, in some embodiments, the steps of obtaining the static data of the water pump motor, the dynamic data of the pipe network environment, the water pump performance curve of the water pump, and the dynamic data of the water pump motor include:

[0070] Obtain the static data, the dynamic data, the dynamic data of the pipe network environment, and the water pump performance curve through the data perception layer.

[0071] Specifically, in this embodiment, a data perception layer for obtaining multi-dimensional data is designed, and data with different dimensions are respectively obtained through the data perception layer.

[0072] Further, in some embodiments, the data perception layer includes:

[0073] A static data acquisition layer configured to obtain the static data of the water pump motor and the water pump performance curve of the water pump;

[0074] A dynamic data acquisition layer configured to obtain the dynamic data of the water pump motor;

[0075] A pipe network environment data acquisition layer configured to obtain the dynamic data of the pipe network environment.

[0076] Specifically, in this embodiment, the static data acquisition layer is used to obtain the static data of the water pump motor and the water pump performance curve of the water pump, and the acquisition method can be to directly obtain it from the water pump through communication; the dynamic data acquisition layer is used to obtain the dynamic data of the water pump motor and monitor the data situation in the water pump in real time through the dynamic data acquisition layer; the pipe network environment data acquisition layer is used to obtain the dynamic data of the pipe network environment, and the pipe network environment data acquisition layer can calculate the dynamic data of the pipe network environment through the pipe diameter of the pipeline connected to the water pump and the flow velocity of the fluid flowing inside the pipeline.

[0077] Further, in some embodiments, the steps of constructing the target motor-pump combined model further include:

[0078] The Gaussian process regression is used to continuously model the performance curve of the water pump. During the continuous modeling process, the static data and the dynamic environment data of the pipe network are used as modeling parameters, and the dynamic data is used as input variables and output variables.

[0079] Specifically, in this embodiment, during the construction process of the target motor-pump combined model, it is necessary to set the modeling parameters, input variables, and output variables of the model. Therefore, the static data and the dynamic environment data of the pipe network are used as modeling parameters, and the dynamic data is used as input variables and output variables to perform subsequent model training.

[0080] Among them, the dynamic data includes the real-time current of the water pump motor, the real-time speed of the water pump motor, and the real-time shaft power of the water pump. The real-time shaft power is the output variable, and the real-time current and the real-time speed are the input variables.

[0081] Further, in some embodiments, the dynamic data acquisition layer is also configured to obtain historical data of the water pump under variable frequency conditions and rated conditions;

[0082] The historical data is used to train the motor-pump combined model to obtain the target motor-pump combined model.

[0083] Specifically, in this embodiment, since it is necessary to use data under different working conditions to train the model in order to enable the model to handle the control of the water pump in different situations, it is necessary to obtain the historical data. Among them, the rated condition can be understood as the operating condition of the water pump under the rated set value, and the variable frequency condition is the operating data of the water pump under all conditions except the rated working condition.

[0084] Further, in some embodiments, the target motor-pump combined model includes a first-layer model and a second-layer model;

[0085] The first-layer model is the mapping relationship between the shaft power, flow rate, and head of the water pump at the rated speed of the water pump motor;

[0086] The second-layer model is a full-condition three-dimensional model constructed based on the speed of the water pump motor as a variable, the speed of the water pump motor, the shaft power of the water pump, the flow rate of the water pump, and the head of the water pump.

[0087] Specifically, in this embodiment, the second-layer model is constructed based on the radial basis function and the motor similarity law, and the full-condition three-dimensional model is a mixed kernel function; in the second-layer model, the flow rate of the water pump is directly proportional to the speed of the water pump motor, and the head of the water pump is squared-related to the speed of the water pump motor.

[0088] This embodiment has the following advantages:

[0089] 1. High-precision real-time monitoring, eliminating the dependence on physical sensors

[0090] Sensorless design: Only through the motor current and speed data, combined with the data-mechanism fusion model, the flow rate, head, and efficiency (error rate < 1.5%) are calculated in real time, eliminating the need for hardware such as flow meters and differential pressure transmitters;

[0091] Dynamic calibration: Based on the pump performance curve table and the non-linear regression model, automatically corrects the parameter deviation caused by wear or operating conditions changes to ensure long-term operation accuracy.

[0092] 2. Precise energy efficiency control, significant energy saving and consumption reduction

[0093] Constant flow / constant head adaptation: Dynamically matches the target speed through the pipeline resistance coefficient K, without PID adjustment, achieving a flow control error < 1% and a head error < 2%;

[0094] Energy efficiency optimization mode: Real-time tracks the pump efficiency (η), automatically adjusts to the high-efficiency area (BEP ± 10%), with a comprehensive energy saving of 15% - 25%.

[0095] 3. Fully compatible intelligent integration, convenient deployment

[0096] Adaptive algorithm: Built-in AI model supports pumps of mainstream brands, without manual calibration, shortening the deployment time from 3 days to 30 minutes.

[0097] Industrial interconnection: Seamlessly accesses the SCADA system through the RS485 / Modbus protocol, supports remote monitoring and data export, and complies with the IEC 61131-3 standard.

[0098] 4. Fault warning and health management

[0099] Multi-parameter diagnosis: Integrates vibration and temperature sensors, real-time monitors abnormalities (such as vibration > 4.5mm / s, efficiency < 50%), and early warns of faults such as bearing wear and impeller cavitation.

[0100] Life optimization: Dynamic load balancing technology reduces mechanical shock, extends the pump life by more than 30%, and reduces the maintenance cost by 40%.

[0101] 5. Strong engineering applicability and high environmental tolerance

[0102] Interference resistance: When the pipeline resistance suddenly changes (ΔK > 20%), the control response time < 200ms, and the flow rate fluctuation < ±1.5%.

[0103] For the sake of convenience in explanation, the above description has been made in conjunction with specific embodiments. However, the above discussion in some embodiments is not intended to be exhaustive or to limit the embodiments to the specific forms disclosed above. According to the above teachings, various modifications and variations can be obtained. The selection and description of the above embodiments are for the purpose of better explaining the content of the present disclosure, so that those skilled in the art can better use the embodiments.

Claims

1. A centrifugal water pump control method based on multi-source data fusion and adaptive model, characterized in that: The control system comprises: Obtain static data of the water pump motor, dynamic environmental data of the pipe network, and water pump performance curve of the water pump; A target motor-pump joint model is constructed based on the static data, the dynamic environment data of the pipe network and the water pump performance curve; the target motor-pump joint model is a multi-dimensional model constructed based on a Gaussian process regression model; The dynamic data of the water pump motor is obtained, and the real-time working mode of the water pump is selected according to the dynamic data and the target motor-pump joint model.

2. A centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 1, characterized in that: The steps of obtaining static data of the water pump motor, dynamic environmental data of the pipe network, and the water pump performance curve of the water pump and obtaining dynamic data of the water pump motor include: The static data, the dynamic data, the pipeline network dynamic environment data and the water pump performance curve are acquired through the data perception layer.

3. A centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 2, characterized in that: The data perception layer includes: A static data acquisition layer, wherein the static data acquisition layer is configured to acquire the static data of the water pump motor and the water pump performance curve of the water pump; A dynamic data acquisition layer, wherein the dynamic data acquisition layer is configured to acquire the dynamic data of the water pump motor; The pipeline network environment data collection layer is configured to obtain the pipeline network dynamic environment data.

4. The centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 1 is characterized in that: The static data includes the rated power, rated current, rated voltage, rated speed, power factor, motor efficiency and frequency reference value of the water pump motor; The water pump performance curve is a mapping table between water pump efficiency, head and flow rate; The pipeline network dynamic environment data is used to characterize the internal fluid resistance of the pipeline connected to the water pump.

5. The centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 3 is characterized in that: The step of constructing the target motor-pump joint model also includes: Gaussian process regression is used to continuously model the water pump performance curve. In the continuous modeling process, the static data and the dynamic environment data of the pipe network are used as modeling parameters, and the dynamic data are used as input variables and output variables.

6. A centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 5, characterized in that: The dynamic data includes the real-time current of the water pump motor, the real-time speed of the water pump motor and the real-time shaft power of the water pump; The step of constructing the target motor-pump joint model also includes taking the real-time current and the real-time rotational speed as the input variables, and taking the real-time shaft power as the output variable.

7. A centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 6, characterized in that: The dynamic data acquisition layer is also configured to obtain historical data of the water pump under variable frequency state and rated state; The motor-pump joint model is trained using the historical data to obtain the target motor-pump joint model.

8. The centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 6 is characterized in that: The target motor-pump joint model includes a first-layer model and a second-layer model; The first layer model is the mapping relationship between the shaft power, flow rate and head of the water pump at the rated speed of the water pump motor; The second-layer model is a full-operating condition three-dimensional model constructed based on the speed of the water pump motor, the shaft power of the water pump, the flow rate of the water pump and the head of the water pump, with the speed of the water pump motor as a variable.

9. A centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 8, characterized in that: The second layer model is constructed based on radial basis functions and motor similarity laws, and the full-operating condition three-dimensional model is a hybrid kernel function; In the second-layer model, the flow rate of the water pump is directly proportional to the speed of the water pump motor, and the head of the water pump is squarely related to the speed of the water pump motor.

10. The centrifugal water pump control method based on multi-source data fusion and adaptive model according to claim 1, characterized in that: The real-time working modes include a constant flow control mode, a constant head control mode and an energy efficiency optimal control mode; The control method further comprises: An objective function is set for the target motor-pump joint model, and an improved particle swarm algorithm is used to search for the global optimal speed to obtain the pump motor speed corresponding to the energy efficiency optimal control mode.

Images