Secondary water supply control method and system based on artificial intelligence

Abstract

The invention provides a secondary water supply control method and system based on artificial intelligence, and relates to the technical field of water supply control. The method comprises the following steps: collecting real-time flow data of a water supply pipeline through a flow meter; receiving the real-time flow data through a controller; the method comprises the following steps: receiving real-time traffic data through local deployment equipment, and determining predicted traffic data at one or more moments in the future according to the real-time traffic data; obtaining water pump lift data at one or more moments in the future; and the water pump lift data and the predicted flow data are transmitted to a water pump controller, and water pump operation parameters at one or more moments in the future are determined through the water pump controller. According to the invention, based on the real-time flow and the predicted future flow, the operation parameters of the water pump are controlled, water is supplied according to the energy consumption required by a pipe network, the current situation that the lift of the traditional water pump is operated according to the most unfavorable working condition is changed, the operation energy consumption, the operation cost and the power consumption of the water pump are greatly reduced, and the aims of saving energy and reducing carbon are fulfilled.

Description

Technical Field

[0001] This invention relates to the field of water supply control technology, and in particular to a secondary water supply control method and system based on artificial intelligence. Background Technology

[0002] Secondary water supply systems are crucial for ensuring the normal operation of buildings and are also a type of end-user energy consumption during building operation. With the continuous advancement of urbanization in my country, the number and density of high-rise buildings are gradually increasing, leading to a surge in demand for secondary water supply technology. Energy conservation and efficiency improvement in secondary water supply is one of the advanced application technologies for achieving carbon reduction during building operation. Among related technologies, first-generation water supply technology is cumbersome to install, requires significant investment, necessitates elevated water tanks, increases structural load, and sets tank height and volume according to the most unfavorable operating conditions, resulting in high energy consumption. Second-generation water supply technology provides constant pressure output, but excessively high excess head during low-flow water use easily leads to energy waste. Its water supply is still based on the most unfavorable operating conditions, disconnected from the actual water usage conditions of the pipe network. Therefore, problems such as excessive excess head and energy surplus exist, making it difficult to supply water according to the energy consumption required by the pipe network. This results in continuous high pressure in the pipe network, increasing the probability of pipe network leakage and causing unnecessary water waste.

[0003] The information disclosed in the background section of this application is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] This invention provides a secondary water supply control method and system based on artificial intelligence, which can solve the technical problem that related technologies cannot supply water according to the energy consumption required by the pipeline network.

[0005] According to a first aspect of the present invention, an artificial intelligence-based secondary water supply control method is provided, comprising: Real-time flow data of the water supply pipeline is collected using a flow meter. The controller receives the real-time traffic data. Receive real-time traffic data through locally deployed devices, and determine predicted traffic data for one or more future times based on the real-time traffic data; The predicted flow rate data is transmitted to the controller, and the controller processes the predicted flow rate data to obtain pump head data for one or more future time periods. The pump head data and the predicted flow rate data are transmitted to the pump controller, and the pump controller determines the pump operating parameters for one or more future times.

[0006] According to the present invention, determining predicted traffic data for one or more future time periods includes: The real-time traffic data is received via the Modbus TCP protocol; By processing real-time traffic data using a trained traffic prediction model, predicted traffic data for one or more future times can be obtained.

[0007] According to the present invention, obtaining predicted traffic data for one or more future time periods includes: The traffic prediction model is used to process real-time traffic data and multiple historical traffic data through 1D convolutional layers to obtain traffic feature information. By processing traffic feature information through the multi-layer perceptual network of the trained traffic prediction model, predicted traffic data for one or more future time points can be obtained.

[0008] According to the present invention, the method further includes: If the current moment reaches the end of the monitoring period, the real-time traffic data of multiple moments in the monitoring period will be transmitted to the cloud server. The cloud server uses real-time traffic data from multiple moments in the monitoring period to train the traffic prediction model, thereby obtaining the trained traffic prediction model. The trained traffic prediction model is deployed to the locally deployed device.

[0009] According to the present invention, obtaining a trained traffic prediction model includes: The real-time traffic data from time k to time k+n within the monitoring period is processed by the traffic prediction model to obtain the predicted traffic data from time k+n+1 to time k+n+m, where n is the number of input data of the traffic prediction model, m is the number of output data of the traffic prediction model, and k is a positive integer. Based on the real-time traffic data and predicted traffic data from the (k+n+1)th to (k+n+m)th time within the monitoring period, the loss function of the traffic prediction model is obtained. The traffic prediction model is trained based on the loss function to obtain the trained traffic prediction model.

[0010] According to the present invention, obtaining pump head data for one or more future time points includes: A preset physical model of water pump head is loaded into the controller; The predicted flow rate data is processed using a preset physical model of the water pump head to obtain the water pump head data.

[0011] According to the present invention, transmitting the pump head data and the predicted flow rate data to a pump controller, and determining the pump operating parameters for one or more future times using the pump controller, includes: The pump head data and the predicted flow rate data are transmitted to the pump controller via the Modbus TCP communication protocol. The pump controller processes the pump head data and the predicted flow rate data to obtain the pump operating parameters for one or more future times.

[0012] According to a second aspect of the present invention, an artificial intelligence-based secondary water supply control system is provided, comprising: The data acquisition module collects real-time flow data from the water supply pipeline via a flow meter. The receiving module receives the real-time traffic data through the controller; The traffic prediction data module receives real-time traffic data through a locally deployed device and determines the predicted traffic data for one or more future times based on the real-time traffic data. The head data module transmits the predicted flow data to the controller, and the controller processes the predicted flow data to obtain pump head data for one or more future times. The pump operation parameter module transmits the pump head data and the predicted flow rate data to the pump controller, and determines the pump operation parameters for one or more future times through the pump controller.

[0013] According to a third aspect of the present invention, an artificial intelligence-based secondary water supply control device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the artificial intelligence-based secondary water supply control method.

[0014] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having computer program instructions stored thereon, which, when executed by a processor, implement the artificial intelligence-based secondary water supply control method.

[0015] By adopting the above technical solution, the present invention can achieve the following technical effects: According to the present invention, real-time flow data of water supply pipelines can be collected by a flow meter, received by a controller, and received by locally deployed equipment. This allows for the determination of predicted flow data for one or more future time periods, which is then transmitted to the controller. This provides pump head data for one or more future time periods. The pump head data and predicted flow data are then transmitted to the pump controller, which determines the pump operating parameters for one or more future time periods. Water supply can be based on the energy consumption required by the pipeline network, changing the traditional practice of operating pumps at the most unfavorable conditions. This significantly reduces pump operating energy consumption and secondary water supply operating costs, meeting water demand under different conditions while achieving high energy efficiency. It also avoids continuous high pressure in the pipeline network, reduces pipeline leakage, and achieves energy saving and carbon reduction goals. Furthermore, the controller can receive real-time flow data collected by the flow meter, providing basic data for determining predicted flow data for one or more future time periods. When determining the predicted flow rate data, the flow prediction model can be trained based on the flow rate data within the monitoring period to obtain the trained flow prediction model. The parameters of the flow prediction model are updated at the end of each monitoring period to obtain the flow prediction model for the next monitoring period. Then, at the end of each monitoring period, the updated flow prediction model is deployed to the local deployment device, which better reflects the dynamic changes in flow usage trends and improves the objectivity, accuracy, and comprehensiveness of flow prediction. Furthermore, pump head data and predicted flow rate data can be transmitted to the pump controller, thereby determining the pump operating parameters for one or more future moments. This can change the traditional situation where pumps operate at the most unfavorable operating conditions, significantly reducing pump operating energy consumption, lowering secondary water supply operating costs and electricity consumption, and achieving energy conservation and carbon reduction goals.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Other features and aspects of the invention will become clearer from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort. Figure 1 An exemplary flowchart of an artificial intelligence-based secondary water supply control method according to an embodiment of the present invention is shown. Figure 2A schematic diagram of the physical model of the water pump head in the water supply mode of the power frequency pump + high-level domestic water tank according to an embodiment of the present invention is shown exemplarily. Figure 3 A schematic diagram of the physical model of the pump head in the constant pressure variable frequency water supply mode according to an embodiment of the present invention is shown as an example. Figure 4 A schematic diagram of an artificial intelligence-based secondary water supply control system according to an embodiment of the present invention is shown as an example; Figure 5 A block diagram of an artificial intelligence-based secondary water supply control system according to an embodiment of the present invention is shown as an example. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0020] Figure 1 An exemplary flowchart illustrates a secondary water supply control method based on artificial intelligence according to an embodiment of the present invention, the method comprising: Step S1: Collect real-time flow data of the water supply pipeline using a flow meter; Step S2: Receive the real-time traffic data through the controller; Step S3: Receive real-time traffic data through the locally deployed device, and determine the predicted traffic data for one or more future times based on the real-time traffic data; Step S4: The predicted flow rate data is transmitted to the controller, and the controller processes the predicted flow rate data to obtain pump head data for one or more future time periods. Step S5: Transmit the pump head data and the predicted flow rate data to the pump controller, and use the pump controller to determine the pump operating parameters for one or more future times.

[0021] According to an embodiment of the present invention, an artificial intelligence-based secondary water supply control method can collect real-time flow data of the water supply pipeline through a flow meter, receive the real-time flow data through a controller, and receive the real-time flow data through a locally deployed device. This allows for the determination of predicted flow data for one or more future time periods, which is then transmitted to the controller. This provides pump head data for one or more future time periods. The pump head data and predicted flow data are then transmitted to the pump controller, which determines the pump operating parameters for one or more future time periods. This method allows water supply to be based on the energy consumption required by the pipeline network, changing the traditional practice of operating pumps at the most unfavorable conditions. This significantly reduces pump operating energy consumption and secondary water supply operating costs, meeting water demand under different conditions while achieving high energy efficiency. It also avoids continuous high pressure in the pipeline network, reduces pipeline leakage, and achieves the goals of energy saving and carbon reduction.

[0022] Example 1: According to an embodiment of the present invention, in step S1, real-time flow data of the water supply pipeline is collected using a flow meter. For example, a flow meter (e.g., a differential pressure flow meter, a thermal mass flow meter, etc.) is installed inside the water supply pipeline, and the water flow in the pipeline is measured every preset time interval (e.g., 1 hour), which is the real-time flow data. The present invention does not limit the monitoring period or the time interval between each collection.

[0023] Example 2: According to an embodiment of the present invention, in step S2, the real-time flow data is received by a controller. The controller (e.g., a PLC programmable logic controller) can receive the real-time flow data collected by the flow meter via the 485 protocol to determine predicted flow data for one or more future times.

[0024] In this way, based on the real-time flow data collected by the flow meter and received by the controller, basic data can be provided to determine the predicted flow data for one or more future times.

[0025] Example 3: According to an embodiment of the present invention, in step S3, receiving real-time traffic data through a locally deployed device and determining predicted traffic data for one or more future times based on the real-time traffic data includes: receiving the real-time traffic data through the Modbus TCP protocol; and processing the real-time traffic data through a trained traffic prediction model to obtain predicted traffic data for one or more future times.

[0026] According to an embodiment of the present invention, the locally deployed device can receive real-time traffic data sent by the controller via the Modbus TCP protocol. After collecting real-time traffic data at multiple times, it can also store the data on the cloud server to provide basic data for training the traffic prediction model. For example, multiple real-time traffic data within a monitoring period (e.g., 1 month) can be transmitted to the cloud server for training the traffic prediction model.

[0027] According to an embodiment of the present invention, a trained traffic prediction model is used to process real-time traffic data to obtain predicted traffic data for one or more future times. This includes: processing real-time traffic data and multiple historical traffic data through a 1D convolutional layer of the trained traffic prediction model to obtain traffic feature information; and processing the traffic feature information through a multilayer perceptron layer of the trained traffic prediction model to obtain predicted traffic data for one or more future times.

[0028] According to an embodiment of the present invention, the real-time traffic data corresponding to multiple times prior to the current time stored in the cloud server is the historical traffic data. Arranging the current real-time traffic data and the historical traffic data from multiple times prior to the current time in chronological order forms a new vector, which can be used as a traffic feature vector. For example, if the current real-time traffic data is 14 L / min, and the historical traffic data from multiple times prior to the current time are 16 L / min…12 L / min, 12 L / min, 10 L / min respectively, predicting the traffic data for one or more future times requires traffic data from multiple times. Therefore, arranging the current real-time traffic data and the historical traffic data from the previous four times in chronological order yields the traffic feature vector, i.e., [10, 12, 12,…, 16, 14]. Convolving the traffic feature vector with a 1D convolutional layer of a trained traffic prediction model (e.g., a 1D convolutional layer containing one or more 1D convolutional layers) yields the corresponding output vector, which is the traffic feature information. When processing traffic feature vectors through 1D convolutional layers, the data in the traffic feature vectors can be convolved by the convolution kernels in the convolutional layers to obtain the output vector, which is the traffic feature information. When processing traffic feature vectors using 1D convolutional layers containing multiple 1D convolutional layers, local features of the traffic feature vectors can be extracted from multiple perspectives to obtain traffic feature information. For example, if the flow feature vector is [10,12,12,…,16,14], the parameters of the convolution kernel are [0.5,1,0.5], and the stride is 1, then the window data corresponding to the stride of 1 is [10,12,12]. The output value is the weighted sum of [10,12,12] and the parameter vector of the convolution kernel [0.5,1,0.5], i.e., 23. Similarly, sliding the window with a stride of 1 can obtain the output values ​​of subsequent strides. These output values ​​can be combined into a vector. Furthermore, there can be multiple convolution kernels. 1D convolution processing can be performed using convolution kernels with various parameters, and the output vectors corresponding to each convolution kernel can be concatenated to obtain the output information of a 1D convolution layer. In the next 1D convolution layer, similar processing is used. Finally, the output information of the last 1D convolution layer can be used as the flow feature information. Through the above 1D convolutional layer processing, when determining traffic feature information, features of traffic feature vectors from multiple previous time points can be extracted, so that traffic feature information can describe the traffic usage status from multiple previous time points to the current time point and reflect the trend of traffic changes.

[0029] According to an embodiment of the present invention, by processing traffic feature information through a multi-layer perceptual network (e.g., including fully connected layers and activation layers, where the activation layers are network layers processed using the ReLU activation function) of a trained traffic prediction model, the traffic feature information can be mapped to a pipeline traffic value vector for one or more future time periods, thereby obtaining the predicted traffic data. For example, the output pipeline traffic value vector is [15, 15, ..., 16], representing the predicted traffic data for multiple future time periods.

[0030] According to an embodiment of the present invention, the method further includes: when the current time reaches the end time of the monitoring period, transmitting real-time traffic data of multiple times in the monitoring period to a cloud server; the cloud server using the real-time traffic data of multiple times in the monitoring period to train a traffic prediction model to obtain a trained traffic prediction model; and deploying the trained traffic prediction model to the local deployment device.

[0031] According to an embodiment of the present invention, when the current time reaches the end of the monitoring period, real-time traffic data for multiple moments of the monitoring period are transmitted to a cloud server. For example, if the monitoring period is one month and the time interval between each moment within the monitoring period is one hour, when the current time reaches the end of the monitoring period, real-time traffic data corresponding to 720 moments can be collected. Furthermore, the real-time traffic data for the aforementioned 720 moments can be transmitted to a cloud server for training the traffic prediction model.

[0032] According to an embodiment of the present invention, the cloud server trains a traffic prediction model using traffic data from multiple moments within a monitoring period to obtain a trained traffic prediction model. This includes: processing real-time traffic data from moment k to moment (k+n) within the monitoring period using the traffic prediction model to obtain predicted traffic data from moment (k+n+1) to moment (k+n+m), where n is the number of input data points to the traffic prediction model, m is the number of output data points to the traffic prediction model, and k is a positive integer; obtaining a loss function for the traffic prediction model based on the real-time traffic data from moment (k+n+1) to moment (k+n+m) within the monitoring period and the predicted traffic data; and training the traffic prediction model based on the loss function to obtain a trained traffic prediction model.

[0033] According to an embodiment of the present invention, similar to obtaining traffic feature information, the traffic data from time k to time (k+n) within the monitoring period are arranged in chronological order to obtain a training traffic feature vector. This training traffic feature vector is then input into the 1D convolutional layer of the traffic prediction model for convolution processing to obtain training traffic feature information. Further, the training traffic feature information is input into the multilayer perceptron layer of the traffic prediction model for processing to obtain predicted traffic data from time (k+n+1) to time (k+n+m). The number of input data points, n, for the traffic prediction model is the number of times needed to predict the predicted traffic data for the next m times. This number can be determined based on the number of traffic data points required to predict the predicted traffic data for the next m times. For example, if at least 10 traffic data points are needed to predict the predicted traffic data for the next 5 times, then m=5 and n=10.

[0034] According to an embodiment of the present invention, the average absolute error between the real-time traffic data from time (k+n+1) to time (k+n+m) and the predicted traffic data at the same time can be used as a loss function. Further, backpropagation can be performed based on this error function, and the parameters of the traffic prediction model can be adjusted using gradient descent to train the model, thereby obtaining the trained traffic prediction model. Based on the same processing method, at the end of each monitoring period, the parameters of the traffic prediction model can be updated based on the traffic data within the monitoring period to obtain the traffic prediction model for the next monitoring period. Real-time parameter updates better reflect the dynamic changes in traffic usage trends, improving the accuracy of traffic prediction.

[0035] According to an embodiment of the present invention, the trained traffic prediction model is deployed to the local deployment device. The trained traffic prediction model is converted into a more suitable deployment format, such as TensorFlow Lite or ONNXRuntime, and then deployed to the local deployment device. This is used to determine predicted traffic data for one or more future time periods after receiving real-time traffic data. That is, at the end of each monitoring period, the updated traffic prediction model can be redeployed to the local deployment device to determine predicted traffic data in the next monitoring period. The above traffic prediction model based on a neural network model is only an example; the traffic prediction model can also be an autoregressive model, a moving average model, etc. The present invention does not limit the specific type of traffic prediction model.

[0036] In this way, the traffic prediction model can be trained based on the traffic data within the monitoring period to obtain the trained traffic prediction model. At the end of each monitoring period, the parameters of the traffic prediction model are updated to obtain the traffic prediction model for the next monitoring period. Then, at the end of each monitoring period, the updated traffic prediction model is deployed to the local deployment device, which is more in line with the dynamic changes in traffic usage trends and improves the objectivity, accuracy and comprehensiveness of traffic prediction.

[0037] Example 4: According to an embodiment of the present invention, in step S4, the predicted flow rate data is transmitted to the controller, and the predicted flow rate data is processed by the controller to obtain pump head data for one or more future times, including: loading a preset pump head physical model into the controller; and processing the predicted flow rate data using the preset pump head physical model to obtain the pump head data.

[0038] Figure 2 An exemplary schematic diagram of the physical model of the water pump head in the water supply mode of the power frequency pump + high-level domestic water tank according to an embodiment of the present invention is shown.

[0039] like Figure 2 The diagram shows the physical model of the water pump head in a power frequency pump + elevated domestic water tank water supply mode. The shaded area represents the energy consumption of this mode. Hmax is the maximum head value on the pump characteristic curve. The water flow is regulated by pumping water to the tank. Its advantages include a certain water tank capacity, allowing for delayed water supply during power outages and providing relatively reliable water supply. Disadvantages include complicated installation, high investment, the need for an elevated water tank (increasing structural load), and the requirement to set the tank height and volume according to the most unfavorable operating conditions, resulting in higher energy consumption.

[0040] Figure 3 A schematic diagram of the physical model of the pump head in the constant pressure variable frequency water supply mode according to an embodiment of the present invention is shown as an example.

[0041] like Figure 3 The diagram shows the physical model of the pump head in a constant pressure variable frequency water supply mode. The energy consumption of this mode can be approximated by shading. It is a commonly used secondary water supply method in related technologies. It utilizes closed-loop feedback self-control technology (PID control) to achieve constant pressure and variable flow water supply. There are two main modes: a combined low-level water tank and a constant pressure variable frequency pump group, and a superimposed pressure water supply mode using municipal pressure through a variable frequency constant pressure pump group. Its advantages include a stable water supply system that meets peak flow and pressure demands. The disadvantage is that the output water is always at constant pressure (the head is always equal to H). 恒压值 When water flow is low, an excessively high excess head results in energy waste.

[0042] According to an embodiment of the present invention, similar to the physical model of the pump head in the power frequency pump + high-level domestic water tank water supply mode and the physical model of the pump head in the constant pressure variable frequency water supply mode, the horizontal axis of the preset pump head physical model is the predicted flow rate data (Q), and the vertical axis is the pump head data (H). The pump characteristic curve of the water supply mode corresponding to the preset pump head physical model can be used, along with the x-axis, y-axis, and Q... max The area enclosed by the straight lines parallel to the y-axis and perpendicular to the x-axis containing the maximum flow rate represents the energy consumption of the water supply mode corresponding to the preset pump head physical model. The preset pump head physical model is loaded into the controller, and the formula for the preset pump head physical model is: Among them, H i Let d be the actual elevation difference between the outlet and the point of use during the i-th time period (the period between the i-th time and the (i+1)-th time), h be the design pressure of the outlet, and d be the elevation difference between the outlet and the point of use. j To calculate the inner diameter of the pipe, L j To be with d j The corresponding pipe length, C h Here, q represents the Hecheng-Williams coefficient; k is the local damping loss coefficient, with a value less than 1. i For the predicted traffic data of the i-th time period, h i This represents the pump head data for the i-th time period.

[0043] Since this water supply mode can achieve on-demand water supply, the preset water pump head physical model can be used. Figure 2 as well as Figure 3 Pipe characteristic curve (H) ST The preset pump head physical model (which represents the minimum head value of the pipeline characteristic curve) is the optimal curve based on the current pipeline characteristics. This allows the pump head data to be set according to flow demand, reducing energy consumption. Furthermore, by inputting the predicted flow data into the preset pump head physical model, the pump head corresponding to the predicted flow data can be determined based on the pump characteristic curve of the preset model. This pump head data allows for the determination of pump head data at one or more future times. By controlling the pump parameters based on the aforementioned pump head data, on-demand water supply to the pipeline network can be achieved, reducing energy consumption.

[0044] Example 5: According to an embodiment of the present invention, in step S5, the pump head data and the predicted flow rate data are transmitted to the pump controller, and the pump operating parameters for one or more future times are determined by the pump controller. This includes: transmitting the pump head data and the predicted flow rate data to the pump controller via the Modbus TCP communication protocol; and processing the pump head data and the predicted flow rate data by the pump controller to obtain the pump operating parameters for one or more future times.

[0045] According to an embodiment of the present invention, since water supply pump rooms are generally located in basements where signal transmission is weak, a local area network environment can be established using the Modbus TCP communication protocol. This allows the predicted flow rate data to be transmitted to the controller for calculating pump head data. The controller then transmits the pump head data and the predicted flow rate data to the pump controller, which processes these data to enable the pump to adaptively adjust (e.g., adjust power, frequency, etc.) to achieve the required pump head and flow rate. This avoids the problem of weak signal transmission, allowing the algorithm model to predict flow rate in real time, and the pump head to change with the flow rate in real time, thus achieving real-time parameter control of the pump.

[0046] In this way, pump head data and predicted flow rate data can be transmitted to the pump controller, which can then determine the pump's operating parameters for one or more future moments. This changes the traditional practice of operating pumps at the most unfavorable head conditions, significantly reducing pump energy consumption, lowering secondary water supply operating costs and electricity consumption, and achieving energy conservation and carbon reduction goals.

[0047] According to an embodiment of the present invention, an artificial intelligence-based secondary water supply control method can collect real-time flow data of the water supply pipeline through a flow meter, receive the real-time flow data through a controller, and receive the real-time flow data through a locally deployed device. This allows for the determination of predicted flow data for one or more future time periods, which is then transmitted to the controller. This provides pump head data for one or more future time periods. The pump head data and predicted flow data are then transmitted to the pump controller, which determines the pump operating parameters for one or more future time periods. This method allows water supply to be based on the energy consumption required by the pipeline network, changing the traditional practice of operating pumps at the most unfavorable conditions. This significantly reduces pump operating energy consumption and secondary water supply operating costs, meeting water demand under different conditions while achieving high energy efficiency. It also avoids continuous high pressure in the pipeline network, reduces pipeline leakage, and achieves energy saving and carbon reduction goals. Furthermore, the controller can receive real-time flow data collected by the flow meter, providing basic data for determining predicted flow data for one or more future time periods. When determining the predicted flow rate data, the flow prediction model can be trained based on the flow rate data within the monitoring period to obtain the trained flow prediction model. The parameters of the flow prediction model are updated at the end of each monitoring period to obtain the flow prediction model for the next monitoring period. Then, at the end of each monitoring period, the updated flow prediction model is deployed to the local deployment device, which better reflects the dynamic changes in flow usage trends and improves the objectivity, accuracy, and comprehensiveness of flow prediction. Furthermore, pump head data and predicted flow rate data can be transmitted to the pump controller, thereby determining the pump operating parameters for one or more future moments. This can change the traditional situation where pumps operate at the most unfavorable operating conditions, significantly reducing pump operating energy consumption, lowering secondary water supply operating costs and electricity consumption, and achieving energy conservation and carbon reduction goals.

[0048] Example 6: Figure 4 A schematic diagram of an artificial intelligence-based secondary water supply control system according to an embodiment of the present invention is shown as an example.

[0049] like Figure 4As shown, the AI-based secondary water supply control system includes: a flow meter, a locally deployed device, a controller, and a pump controller. The flow meter collects real-time flow data and transmits it to the controller. The locally deployed device receives the real-time flow data sent by the controller via the Modbus TCP protocol, determines the predicted flow data for one or more future times based on the flow prediction model (updated once per monitoring cycle) deployed on the locally deployed device, and then transmits the predicted flow data to the controller. The controller receives the real-time flow data collected by the flow meter via the 485 protocol and transmits it to the locally deployed device. It then receives the predicted flow data determined by the locally deployed device. Further, it processes the predicted flow data using a preset pump head physical model loaded in the controller to obtain the pump head data, and transmits the pump head data and predicted flow data to the pump controller. The pump controller receives the pump head data and predicted flow data transmitted by the controller via the Modbus TCP communication protocol, performs adaptive adjustment to obtain the pump's operating parameters for one or more future times, and controls the pump to operate based on these operating parameters.

[0050] Example 7: Figure 5 An exemplary block diagram of an artificial intelligence-based secondary water supply control system according to an embodiment of the present invention is shown, the system comprising: The data acquisition module collects real-time flow data from the water supply pipeline via a flow meter. The receiving module receives the real-time traffic data through the controller; The traffic prediction data module receives real-time traffic data through a locally deployed device and determines the predicted traffic data for one or more future times based on the real-time traffic data. The head data module transmits the predicted flow data to the controller, and the controller processes the predicted flow data to obtain pump head data for one or more future times. The pump operation parameter module transmits the pump head data and the predicted flow rate data to the pump controller, and determines the pump operation parameters for one or more future times through the pump controller.

[0051] According to one embodiment of the present invention, an artificial intelligence-based secondary water supply control device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the artificial intelligence-based secondary water supply control method.

[0052] According to one embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, which, when executed by a processor, implement the artificial intelligence-based secondary water supply control method.

[0053] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0054] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and any variations or modifications may be made to the implementation of the present invention without departing from the stated principles.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A secondary water supply control method based on artificial intelligence, characterized in that, include: Real-time flow data of the water supply pipeline is collected using a flow meter. The controller receives the real-time traffic data. Receive real-time traffic data through locally deployed devices, and determine predicted traffic data for one or more future times based on the real-time traffic data; The predicted flow rate data is transmitted to the controller, and the controller processes the predicted flow rate data to obtain pump head data for one or more future time periods. The pump head data and the predicted flow rate data are transmitted to the pump controller, and the pump controller determines the pump operating parameters for one or more future times.

2. The artificial intelligence-based secondary water supply control method according to claim 1, characterized in that, Receive real-time traffic data through locally deployed devices, and determine predicted traffic data for one or more future time periods based on the real-time traffic data, including: The real-time traffic data is received via the Modbus TCP protocol; By processing real-time traffic data using a trained traffic prediction model, predicted traffic data for one or more future times can be obtained.

3. The artificial intelligence-based secondary water supply control method according to claim 2, characterized in that, By processing real-time traffic data using a trained traffic prediction model, predicted traffic data for one or more future time points can be obtained, including: The traffic prediction model is used to process real-time traffic data and multiple historical traffic data through 1D convolutional layers to obtain traffic feature information. By processing traffic feature information through the multi-layer perceptual network of the trained traffic prediction model, predicted traffic data for one or more future time points can be obtained.

4. The artificial intelligence-based secondary water supply control method according to claim 2, characterized in that, The method further includes: If the current moment reaches the end of the monitoring period, the real-time traffic data of multiple moments in the monitoring period will be transmitted to the cloud server. The cloud server uses real-time traffic data from multiple moments in the monitoring period to train the traffic prediction model, thereby obtaining the trained traffic prediction model. The trained traffic prediction model is deployed to the locally deployed device.

5. The artificial intelligence-based secondary water supply control method according to claim 4, characterized in that, The cloud server uses traffic data from multiple moments within the monitoring period to train the traffic prediction model, obtaining the trained traffic prediction model, including: The real-time traffic data from time k to time k+n within the monitoring period is processed by the traffic prediction model to obtain the predicted traffic data from time k+n+1 to time k+n+m, where n is the number of input data of the traffic prediction model, m is the number of output data of the traffic prediction model, and k is a positive integer. Based on the real-time traffic data and predicted traffic data from the (k+n+1)th to (k+n+m)th time within the monitoring period, the loss function of the traffic prediction model is obtained. The traffic prediction model is trained based on the loss function to obtain the trained traffic prediction model.

6. The artificial intelligence-based secondary water supply control method according to claim 1, characterized in that, The predicted flow rate data is transmitted to the controller, and the controller processes the predicted flow rate data to obtain pump head data for one or more future time periods, including: A preset physical model of water pump head is loaded into the controller; The predicted flow rate data is processed using a preset physical model of the water pump head to obtain the water pump head data.

7. The artificial intelligence-based secondary water supply control method according to claim 1, characterized in that, The pump head data and the predicted flow rate data are transmitted to the pump controller, and the pump controller determines the pump operating parameters for one or more future times, including: The pump head data and the predicted flow rate data are transmitted to the pump controller via the Modbus TCP communication protocol. The pump controller processes the pump head data and the predicted flow rate data to obtain the pump operating parameters for one or more future times.

8. A secondary water supply control system based on artificial intelligence, characterized in that, include: The data acquisition module collects real-time flow data from the water supply pipeline via a flow meter. The receiving module receives the real-time traffic data through the controller; The traffic prediction data module receives real-time traffic data through a locally deployed device and determines the predicted traffic data for one or more future times based on the real-time traffic data. The head data module transmits the predicted flow data to the controller, and the controller processes the predicted flow data to obtain pump head data for one or more future times. The pump operation parameter module transmits the pump head data and the predicted flow rate data to the pump controller, and determines the pump operation parameters for one or more future times through the pump controller.

9. A secondary water supply control device based on artificial intelligence, characterized in that, include: processor; A memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores computer program instructions that, when executed by a processor, implement the method of any one of claims 1-7.

Images