Scheduling method, device and equipment of water pump unit and medium

Abstract

The invention discloses a scheduling method, device and equipment for a water pump unit and a storage medium. Historical operation data of a water supply system and operation condition data of the water pump unit are collected; according to the historical operation data, training a corresponding water supply demand prediction model, and predicting a water supply flow demand in a future time period through the trained water supply demand prediction model; physical characteristic information of the water pump unit is determined according to the operation condition data; establishing a multi-target mixed integer nonlinear programming model related to the scheduling of the water pump unit by taking the water supply flow demand and the operation limitation of the water pump unit as constraint conditions and taking the minimization of the power cost and the minimization of the equipment loss as optimization targets; and solving the multi-target mixed integer nonlinear programming model, and scheduling the water pump unit in a future time period according to a solving result. The method can effectively improve the scheduling efficiency and benefit of the water pump unit. The device can be widely applied to the technical field of water supply.

Description

Technical Field

[0001] The present application relates to the technical field of water supply, and in particular to a scheduling method, device, equipment and storage medium for a water pump unit. Background Art

[0002] The water pump unit in the water supply system is the core power equipment, mainly used to transport water sources (such as reservoirs, rivers or groundwater) to the water treatment plant, water tower or directly to the user end through pipelines. The scheduling of the water pump unit is a key link to ensure that the water use needs of urban residents and enterprises are met. Reasonable scheduling of the water pump unit can not only improve the water supply efficiency, but also reduce energy consumption, reduce equipment wear, extend the service life of the equipment, and ensure the stability and safety of the water supply system.

[0003] In the related art, the existing scheduling of water pump units generally captures the trend of water supply demand and arranges the scheduling of water pump units according to the future water supply demand situation. This implementation method ignores the situation of the water pump unit, which easily leads to the water pump unit being difficult to exert its optimal scheduling potential and has low practicability.

[0004] In summary, the problems existing in the related art need to be solved urgently. Summary of the Invention

[0005] An object of the present application is to solve at least to a certain extent one of the technical problems existing in the related art.

[0006] To this end, an object of an embodiment of the present application is to provide a scheduling method, device, equipment and storage medium for a water pump unit.

[0007] In order to achieve the above technical object, the technical solutions adopted in the embodiments of the present application include:

[0008] On the one hand, an embodiment of the present application provides a scheduling method for a water pump unit, and the method includes:

[0009] Collect the historical operation data of the water supply system and the operation condition data of the water pump unit; wherein, the historical operation data includes historical water consumption data;

[0010] According to the historical operation data, train the corresponding water supply demand prediction model, and predict the water supply flow demand in the future period through the trained water supply demand prediction model;

[0011] According to the operation condition data, determine the physical characteristic information of the water pump unit; wherein, the physical characteristic information includes the correlation curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency and flow velocity of the water pump unit;

[0012] Taking the water supply flow demand and the operation limits of the water pump unit as constraint conditions, and minimizing the electricity cost and equipment loss as optimization objectives, a multi-objective mixed-integer non-linear programming model for the scheduling of the water pump unit is established;

[0013] Solving the multi-objective mixed-integer non-linear programming model, and scheduling the water pump unit in the future time period according to the solution result.

[0014] In addition, according to the scheduling method of the water pump unit in the above embodiments of the present application, the following additional technical features may also be included:

[0015] Further, in an embodiment of the present application, after collecting the historical operation data of the water supply system and the operation conditions data of the water pump unit, the method further includes:

[0016] Preprocessing the historical operation data and the operation conditions data;

[0017] Among them, the operations of the preprocessing include at least one of outlier detection and correction, data normalization processing, and creating a sliding window dataset.

[0018] Further, in an embodiment of the present application, training the corresponding water supply demand prediction model according to the historical operation data includes:

[0019] Dividing the historical water consumption data by time period to obtain sub-water consumption data in multiple time periods; wherein, each time period includes several sampling time points, and the number of sampling time points in each time period is the same;

[0020] According to the first sub-water consumption data in the first time period and the second sub-water consumption data in the previous time period of the first time period, determining the kurtosis value corresponding to the first time period; wherein, the first time period is any one time period;

[0021] According to the kurtosis value, determining the training weight corresponding to each time period;

[0022] Inputting the sub-water consumption data of each time period into the water supply demand prediction model, and predicting the water consumption in the next time period of each time period through the water supply demand prediction model to obtain predicted water consumption data;

[0023] According to the sub-water consumption data in the next time period of each time period and the predicted water consumption data corresponding to the time period, determining the initial loss value of the time period;

[0024] According to the training weight, performing weighted summation on the initial loss value to obtain a summarized loss value;

[0025] Update the parameters of the water supply demand prediction model according to the aggregated loss value to obtain a trained water supply demand prediction model.

[0026] Further, in an embodiment of the present application, the determining the kurtosis value corresponding to the first time period according to the first sub-water consumption data in the first time period and the second sub-water consumption data in the previous time period of the first time period includes:

[0027] Associate and correspond the sampling time points of the first time period and the second time period in sequence;

[0028] Compare the magnitude relationship between the first sub-water consumption data and the second sub-water consumption data at each corresponding sampling time point;

[0029] If the first sub-water consumption data is greater than the second sub-water consumption data at the sampling time point, determine the sampling time point as the target time point;

[0030] Determine the kurtosis value corresponding to the first time period according to the number of target time points corresponding to the first time period.

[0031] Further, in an embodiment of the present application, the determining the training weight corresponding to each time period according to the kurtosis value includes:

[0032] Determine the smallest first value and the largest second value from the kurtosis values corresponding to each time period;

[0033] Calculate the first difference between the first value and the second value, and calculate the second difference between the kurtosis value corresponding to each time period and the first value;

[0034] Calculate the ratio of the second difference corresponding to each time period to the first difference, and determine the sum of the ratio and 1 as the preliminary weight corresponding to the time period;

[0035] Normalize the preliminary weights of each time period to obtain the training weights corresponding to each time period.

[0036] Further, in an embodiment of the present application, the water supply demand prediction model is built based on LSTM.

[0037] Further, in an embodiment of the present application, the solving of the multi-objective mixed-integer non-linear programming model includes:

[0038] Solve the multi-objective mixed-integer non-linear programming model by using the non-dominated sorting genetic algorithm.

[0039] On the other hand, an embodiment of the present application provides a scheduling device for a water pump unit, and the device includes:

[0040] A collection unit for collecting historical operation data of a water supply system and operation condition data of a water pump unit; wherein, the historical operation data includes historical water consumption data.

[0041] A prediction unit for training a corresponding water supply demand prediction model according to the historical operation data and predicting the water supply flow demand in a future period through the trained water supply demand prediction model.

[0042] A processing unit for determining physical characteristic information of the water pump unit according to the operation condition data; wherein, the physical characteristic information includes an association curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency and flow velocity of the water pump unit.

[0043] An establishment unit for establishing a multi-objective mixed integer non-linear programming model for scheduling the water pump unit with the water supply flow demand and the operation limit of the water pump unit as constraint conditions and minimizing the electricity cost and equipment loss as optimization objectives.

[0044] An execution unit for solving the multi-objective mixed integer non-linear programming model and scheduling the water pump unit in a future period according to the solution result.

[0045] On the other hand, an embodiment of the present application provides an electronic device, including:

[0046] At least one processor;

[0047] At least one memory for storing at least one program;

[0048] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned scheduling method of the water pump unit.

[0049] On the other hand, an embodiment of the present application further provides a computer-readable storage medium, in which a program executable by a processor is stored, and the program executable by the processor is used to implement the above-mentioned scheduling method of the water pump unit when executed by the processor.

[0050] The advantages and beneficial effects of the present application will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present application:

[0051] The scheduling method, device, equipment and storage medium of the water pump unit disclosed in the embodiments of the present application collect the historical operation data of the water supply system and the operation condition data of the water pump unit; wherein, the historical operation data includes historical water consumption data; according to the historical operation data, a corresponding water supply demand prediction model is trained, and the water supply flow demand in a future period is predicted through the trained water supply demand prediction model; according to the operation condition data, the physical characteristic information of the water pump unit is determined; wherein, the physical characteristic information includes the correlation curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency and flow velocity of the water pump unit; taking the water supply flow demand and the operation limit of the water pump unit as constraint conditions, and taking the minimization of power cost and the minimization of equipment loss as optimization objectives, a multi-objective mixed-integer non-linear programming model for the scheduling of the water pump unit is established; the multi-objective mixed-integer non-linear programming model is solved, and the water pump unit in the future period is scheduled according to the solution result. This method predicts the water supply flow demand in the future period, establishes a multi-objective optimized scheduling model based on the physical characteristic information of the water pump unit, and finds the optimal scheduling plan by weighing the power cost, equipment loss, water supply flow demand and the operation limit of the water pump unit, which can effectively improve the scheduling efficiency and revenue of the water pump unit and is beneficial to ensuring the stability and reliability of water supply. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following introduces the accompanying drawings of the related technical solutions in the embodiments of the present application or the prior art. It should be understood that the accompanying drawings in the following introduction are only for conveniently and clearly expressing some embodiments of the technical solutions in the present application, and those skilled in the art can also obtain other drawings based on these drawings without creative efforts.

[0053] Figure 1 It is a schematic diagram of the implementation environment of a scheduling method for a water pump unit provided in an embodiment of the present application;

[0054] Figure 2 It is a schematic flowchart of a scheduling method for a water pump unit provided in an embodiment of the present application;

[0055] Figure 3 It is a schematic diagram of the overall implementation of a scheduling method for a water pump unit provided in an embodiment of the present application;

[0056] Figure 4 is a schematic diagram of water supply flow demand data provided in an embodiment of the present application;

[0057] Figure 5 It is a schematic diagram of normalized demand data and sample weights provided in an embodiment of the present application;

[0058] Figure 6 This is a comparison schematic diagram of a prediction result provided in an embodiment of the present application;

[0059] Figure 7 This is a schematic diagram of the fitting effect of a water pump provided in an embodiment of the present application;

[0060] Figure 8 This is a schematic diagram of a Pareto optimal boundary provided in an embodiment of the present application;

[0061] Figure 9 This is a structural schematic diagram of a scheduling device for a water pump unit provided in an embodiment of the present application;

[0062] Figure 10 This is a structural schematic diagram of an electronic device provided in an embodiment of the present application. Detailed implementation manners

[0063] The present application will be further described below in conjunction with the accompanying drawings of the specification and specific embodiments. The described embodiments should not be regarded as limitations on the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0064] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0066] 1) Artificial Intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, a theory, method, technology, and application system that can perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines, enabling the machines to have the functions of perception, reasoning, and decision-making.

[0067] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware and software technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, pre-trained model technologies, operation / interaction systems, mechatronics, etc. Among them, pre-trained models, also known as large models or foundation models, can be widely applied to downstream tasks in various directions of artificial intelligence after fine-tuning. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0068] 2) Machine Learning (ML) is an interdisciplinary field that involves multiple disciplines such as probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve their own performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent, and its applications cover all fields of artificial intelligence. Machine learning and deep learning usually include technologies such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and rote learning. Pre-trained models are the latest development results of deep learning, integrating the above technologies.

[0069] 3) LSTM (Long Short-Term Memory) is a special type of Recurrent Neural Network (RNN) that can learn long-term dependencies. Traditional RNNs are prone to problems such as vanishing gradients or exploding gradients when processing long sequence data, making it difficult for the model to capture information at long time intervals. LSTM solves this problem by introducing special structural units.

[0070] The pump unit in the water supply system is the core power equipment, mainly used to transport water sources (such as reservoirs, rivers, or groundwater) to water treatment plants, water towers, or directly to the user end through pipelines. The scheduling of the pump unit is a key link to ensure that the water use needs of urban residents and enterprises are met. Reasonable scheduling of the pump unit can not only improve the water supply efficiency, but also reduce energy consumption, reduce equipment wear, extend the service life of the equipment, and ensure the stability and safety of the water supply system.

[0071] In related technologies, the existing scheduling of pump units generally captures the trend of water supply demand and arranges the scheduling of pump units according to the future water supply demand situation. This implementation method ignores the situation of the pump unit, which easily leads to the pump unit being unable to exert its optimal scheduling potential and has low practicality.

[0072] In view of this, in the embodiments of the present application, a scheduling method for a water pump unit is provided, which collects historical operation data of a water supply system and operation condition data of the water pump unit; wherein, the historical operation data includes historical water consumption data; according to the historical operation data, a corresponding water supply demand prediction model is trained, and the water supply flow demand in a future period is predicted through the trained water supply demand prediction model; according to the operation condition data, the physical characteristic information of the water pump unit is determined; wherein, the physical characteristic information includes the correlation curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency and flow velocity of the water pump unit; taking the water supply flow demand and the operation limit of the water pump unit as constraint conditions, and taking the minimization of power cost and the minimization of equipment loss as optimization objectives, a multi-objective mixed integer non-linear programming model for the scheduling of the water pump unit is established; the multi-objective mixed integer non-linear programming model is solved through a non-dominated sorting genetic algorithm, and the water pump unit in the future period is scheduled according to the solution result. This method predicts the water supply flow demand in the future period, establishes a multi-objective optimized scheduling model based on the physical characteristic information of the water pump unit, and finds the optimal scheduling plan by weighing the power cost, equipment loss, water supply flow demand and the operation limit of the water pump unit, which can effectively improve the scheduling efficiency and revenue of the water pump unit and is beneficial to ensuring the stability and reliability of water supply.

[0073] Please refer to Figure 1 , Figure 1 FIG. shows a schematic diagram of an implementation environment of a scheduling method for a water pump unit provided in an embodiment of the present application. In this implementation environment, the main software and hardware entities involved include a terminal device 110 and a background server 120. The terminal device 110 and the background server 120 are communicatively connected.

[0074] Specifically, the scheduling method for the water pump unit provided in the embodiments of the present application can be executed solely on the side of the terminal device 110, or solely on the side of the background server 120, or executed based on data interaction between the terminal device 110 and the background server 120.

[0075] Among them, the terminal device 110 in the above embodiments may include a mobile phone, a computer, a smart wearable device, a PDA device, etc., but is not limited thereto. The background server 120 may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0076] A communication connection can be established between the terminal device 110 and the background server 120 through a wireless network or a wired network. The wireless network or the wired network uses standard communication technologies and / or protocols. The network can be set as the Internet or any other network, such as including but not limited to any combination of a Local Area Network (LAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), a mobile, wired or wireless network, a private network or a virtual private network.

[0077] Of course, it can be understood that Figure 1 the implementation environment in Figure 1 is only some optional application scenarios of the scheduling method of the water pump unit provided in the embodiments of the present application. The actual application is not fixed to

[0078] Next, in combination with the introduction of the foregoing implementation environment, a scheduling method of a water pump unit provided in the embodiments of the present application will be introduced and described.

[0079] Please refer to Figure 2 , Figure 2 which is a schematic diagram of a scheduling method of a water pump unit provided in the embodiments of the present application. The scheduling method of the water pump unit includes but is not limited to:

[0080] Step 210, collect historical operation data of the water supply system and operation condition data of the water pump unit; wherein, the historical operation data includes historical water consumption data;

[0081] Step 220, train a corresponding water supply demand prediction model according to the historical operation data, and predict the water supply flow demand in a future period through the trained water supply demand prediction model;

[0082] Step 230, determine the physical characteristic information of the water pump unit according to the operation condition data; wherein, the physical characteristic information includes the correlation curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency and flow rate of the water pump unit;

[0083] Step 240, taking the water supply flow demand and the operation limit of the water pump unit as constraint conditions, and taking the minimization of power cost and the minimization of equipment loss as optimization objectives, establish a multi-objective mixed integer nonlinear programming model for the scheduling of the water pump unit;

[0084] Step 250, solve the multi-objective mixed integer nonlinear programming model, and schedule the water pump unit in a future period according to the solution result.

[0085] In an embodiment of the present application, a scheduling method for a water pump unit is provided. This method predicts the water supply flow demand in future periods, establishes a multi-objective optimization scheduling model based on the physical characteristic information of the water pump unit, and finds the optimal scheduling plan by weighing the electricity cost, equipment loss, water supply flow demand, and the operation limits of the water pump unit, which can effectively improve the scheduling efficiency and benefits of the water pump unit and is conducive to ensuring the stability and reliability of water supply.

[0086] In an embodiment of the present application, when scheduling the water pump unit, historical operation data of the water supply system can be collected. The operation data here can include, but is not limited to, historical water consumption data, water temperature data, water pressure data, etc. The present application does not limit this. In addition, operation condition data of the water pump unit is also collected. The operation data here can include, but is not limited to, data such as the flow rate, head, efficiency, and power of the water pump.

[0087] In some embodiments, after obtaining the historical operation data of the water supply system and the operation condition data of the water pump unit, these data can be preprocessed. Exemplarily, the preprocessing operations can include at least one of outlier detection and correction, data normalization processing, and creating a sliding window dataset. Among them, an outlier refers to a data point that is significantly different from other observed values, which may be caused by measurement errors, equipment failures, or special events. These outliers may affect the subsequent analysis results, so they need to be detected and processed. Common outlier detection methods include statistical methods, machine learning methods, or rule-based methods. When correcting outliers, it can be deletion or replacement processing. The present application does not limit the specific detection and correction methods. Data normalization is to make data with different dimensions comparable, and the specific normalization method can be implemented with reference to existing methods. Creating a sliding window dataset can help capture the dependencies in time. The specific approach is to extract part of the data in a fixed-length time window from the original time series data, take this part of the data as a sub-data, and then move the time window to take the next sub-data to obtain the sliding window dataset.

[0088] In the embodiments of the present application, a corresponding water supply demand prediction model can be trained based on the historical operation data of the water supply system, and the water supply flow demand for a future period can be predicted through the trained water supply demand prediction model. For example, the water supply demand prediction model in the embodiments of the present application can be built based on LSTM. LSTM (Long Short-Term Memory Neural Network) is a special recurrent neural network (RNN) that performs well in processing sequential data. Through its gated unit structure, LSTM (Long Short-Term Memory Neural Network) effectively solves the problems of gradient disappearance and gradient explosion that easily occur in traditional recurrent neural networks (RNN) when processing long-sequence data, so as to better capture the long-term dependence relationships in water supply demand prediction. This enables LSTM to effectively process multi-factor data such as a large amount of historical water consumption, meteorological data, and holiday information, and learn the long-term dependence relationships therein, so as to more accurately predict the water supply flow demand for a future period, providing important decision-making support for the optimal scheduling of pump units.

[0089] Exemplarily, in some embodiments, the training of the corresponding water supply demand prediction model according to the historical operation data includes:

[0090] Dividing the historical water consumption data by time period to obtain sub-water consumption data for multiple time periods; wherein, each time period includes a number of sampling time points, and the number of sampling time points in each time period is the same;

[0091] Determining the kurtosis value corresponding to the first time period according to the first sub-water consumption data in the first time period and the second sub-water consumption data in the time period immediately preceding the first time period; wherein, the first time period is any one of the time periods;

[0092] Determining the training weights corresponding to each of the time periods according to the kurtosis value;

[0093] Inputting the sub-water consumption data for each time period into the water supply demand prediction model, and predicting the water consumption for the next time period of each time period through the water supply demand prediction model to obtain predicted water consumption data;

[0094] Determining the initial loss value for the time period according to the sub-water consumption data for the next time period of each time period and the predicted water consumption data corresponding to the time period;

[0095] Performing weighted summation on the initial loss values according to the training weights to obtain a total loss value;

[0096] Updating the parameters of the water supply demand prediction model according to the total loss value to obtain a trained water supply demand prediction model.

[0097] It should be noted that the traditional LSTM model has limitations in dealing with the prediction of water supply demand with unbalanced importance during periods, because the water supply flow demand fluctuates greatly with time periods. For example, there are differences in water usage in different seasons and also in different times of a day. Considering this situation, in the embodiments of the present application, a weighted method is adopted to train the water supply demand prediction model. By means of different weights, the importance of water supply in different time periods is reflected, that is, a weight function is introduced to make corresponding adjustments to the original loss value.

[0098] Specifically, in the embodiments of the present application, when training the water supply demand prediction model, the historical water consumption data can be divided according to time periods to obtain sub-water consumption data under different time periods. Here, the divided time periods can be determined according to actual needs. For example, in some embodiments, when the prediction period is relatively large, such as quarterly prediction, the time periods can be divided by days; in some embodiments, when the prediction period is relatively small, such as intraday prediction, the time periods can be divided by hours. The present application does not limit this. For each time period, there are several sampling time points corresponding to it, and the intervals of each sampling time point can be the same. Since the lengths of the time periods themselves are the same, the number of sampling time points within each time period is also the same. For example, within each hour, a sampling time point can be set every ten minutes. The present application does not limit the specific situation of the sampling time points. It can be understood that in the embodiments of the present application, for each time period, the sampling time points among them can be associated and corresponding in the order of time.

[0099] In the embodiments of the present application, for any time period of the historical water consumption data, the sub-water consumption data within this time period can be compared with the sub-water consumption data of the previous time period. The sub-water consumption data within this time period is denoted as the first sub-water consumption data, and the sub-water consumption data of the previous time period is denoted as the second sub-water consumption data. By comparing the situations of the two, the kurtosis value corresponding to the first time period can be determined. Specifically, in the embodiments of the present application, it is stipulated that the kurtosis value is the number of values of the sampling time point of the current time period (i.e., the first sub-water consumption data) greater than the values of the sampling time point of the past time period (i.e., the second sub-water consumption data) within a sliding window. Its formula can be expressed as:

[0100]

[0101] where T is the size of the sliding window, that is, the length of the time period, j is the sampling time point of the current time period, and t is the sampling time point of the previous time period. G t represents the kurtosis of the sampling time point j, and PV j represents the kurtosis value of the current time period.

[0102] Specifically, in the embodiments of the present application, the magnitude relationship between the first sub - water consumption data and the second sub - water consumption data at each corresponding sampling time point can be compared. In the above formula, d t represents the second sub - water consumption data, and d j represents the first sub - water consumption data. If the first sub - water consumption data is greater than the second sub - water consumption data, it can be determined that the sampling time point is the target time point, and the kurtosis at this sampling time point is 1. By counting the total number of target time points within the statistical period, the kurtosis value of this period can be determined.

[0103] It can be understood that in the embodiments of the present application, the larger the kurtosis value of a period, the more water is consumed within this period. During model prediction, more attention needs to be paid to such periods. In the embodiments of the present application, their different importance can be reflected by means of different weights, that is, a weight function is introduced to correct the loss value, and the obtained weighted loss function is as follows:

[0104]

[0105] where, is the predicted value of the model, and is the actual observed value. is the basic loss function, which measures the difference between the predicted value and the actual value. W(d) is a weight function that measures the importance of different samples and is used to adjust the loss values of different observed values.

[0106] Specifically, in the embodiments of the present application, the training weights corresponding to each period can be determined according to the kurtosis value.

[0107] When training the water supply demand prediction model, the sub - water consumption data of each period can be input into the water supply demand prediction model, and the water consumption of the next period of each period can be predicted through the water supply demand prediction model to obtain the predicted water consumption data. After obtaining the predicted water consumption data, a loss value, denoted as the initial loss value, can be determined according to the actual sub - water consumption data of the next period of each period and the predicted water consumption data corresponding to the period. This loss value can be determined by any loss function, and the present application does not limit this. For example, the MSE loss function, RMSE loss function, or coefficient of determination can be used.

[0108] After obtaining the initial loss value, the weighted sum can be performed through the training weight corresponding to each period and the initial loss value to obtain the aggregated loss value, and then the parameters of the water supply demand prediction model can be updated using the aggregated loss value, thereby realizing the training of the water supply demand prediction model.

[0109] It can be understood that in the training process of the water supply demand prediction model in the embodiments of the present application, the weight optimization based on the kurtosis value within a time period is introduced, which can effectively improve the attention of the model to the peak time period, thereby improving the prediction effect of the water supply demand prediction model.

[0110] Specifically, in some embodiments, determining the training weight corresponding to each of the time periods according to the kurtosis value includes:

[0111] Determining a minimum first value and a maximum second value from the kurtosis values corresponding to each of the time periods;

[0112] Calculating a first difference between the first value and the second value, and calculating a second difference between the kurtosis value corresponding to each time period and the first value;

[0113] Calculating a ratio of the second difference corresponding to each time period to the first difference, and determining the sum of the ratio and 1 as the preliminary weight corresponding to the time period;

[0114] Normalizing the preliminary weights of each of the time periods to obtain the training weights corresponding to each of the time periods.

[0115] In the embodiments of the present application, when determining the training weight, a minimum first value and a maximum second value can be first determined from the kurtosis values corresponding to each time period. Then an initial weight is calculated. The specific calculation method is to first calculate a first difference between the first value and the second value, and calculate a second difference between the kurtosis value corresponding to each time period and the first value, and then calculate a ratio of the second difference to the first difference, and add 1 to obtain the preliminary weight corresponding to the time period. Then, the preliminary weights of the time periods can be normalized to obtain the training weights corresponding to each time period.

[0116] In the embodiments of the present application, the physical characteristic information of the water pump unit can also be determined according to the operating condition data. Here, the physical characteristic information may include the correlation curve between the performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency, and flow velocity of the water pump unit. Specifically, the flow velocity of the water pump refers to the flow velocity of the water extracted and released by the water pump, the head of the water pump is a potential energy that the water pump itself has to push the water, and the efficiency of the water pump refers to the percentage of the effective work converted from electrical energy in the electrical energy. According to experience, there is an approximate quadratic function relationship between the head, efficiency of the water pump and the flow velocity of the water pump, and the relational expression is as follows:

[0117]

[0118] Wherein, a i2 , a i1 , a i0 and b i2 , b i1 , bi0 is the coefficient to be fitted, q ij is the water pump flow rate. According to real life, due to the design and physical limitations of the pump, at high flow rates, its energy is used to push the water flow, resulting in a decrease in the energy it has to push the water flow. Therefore, generally, a i2 < 0.

[0119] The average power p of water pump i during period j ij can be expressed as:

[0120]

[0121] where e is a physical constant, calculated from the physical constants for calculating the pump power, generally ρ is the density of water, g is the acceleration due to gravity, h ij , n ij , q ij are the pump head, pump efficiency, and pump flow rate respectively.

[0122] In the embodiments of the present application, the water supply flow demand and the operation restrictions of the water pump unit can be used as constraint conditions, and the minimization of electricity cost and the minimization of equipment loss can be used as optimization objectives to establish a multi-objective mixed-integer non-linear programming model for the scheduling of the water pump unit. Then, the multi-objective mixed-integer non-linear programming model can be solved by the non-dominated sorting genetic algorithm, and the water pump unit in the future period can be scheduled according to the solution results.

[0123] Specifically, in terms of the objective function, the pump head, efficiency, and power can be calculated using the pump flow rate, and the objective of minimizing electricity consumption can be derived, that is, the total work done by the water pump unit is minimized. The objective of minimizing equipment loss is measured by the minimum of the pump switch changes and the total number of starts. In terms of the constraint conditions, one is that the total flow supply in each period needs to meet the predicted flow demand, and the other is that the pump flow rate is restricted by its rated flow rate, minimum head, etc.

[0124] Based on the above objectives and restrictions, a multi-objective mixed-integer non-linear programming model for the scheduling of the water pump unit is established. The specific steps are as follows:

[0125] Objective: (1) Minimum electricity cost

[0126] The electricity cost of the pump operation is mainly related to the pump power, operation time, and electricity price. The cumulative electricity cost of all pumps in all periods can be used to calculate the minimum total electricity cost of the entire scheduling cycle, which can be expressed as follows.

[0127]

[0128] where p ij is the power of water pump i in period j, Δ jis the time interval for period j, c 1 is the unit electricity price.

[0129] (2) Minimize equipment losses

[0130] Generally speaking, for pump equipment with normal usage frequency, reducing the number of pumps running simultaneously can reduce equipment losses. Additionally, and more importantly, frequent switching can lead to accelerated equipment wear and increased risk of failure. Therefore, to reduce equipment losses, we mainly start from two aspects: minimizing the number of switching operations and minimizing the total number of pumps in operation. By optimizing the scheduling strategy, such as using variable frequency drives to adjust the equipment operating speed instead of fully starting or stopping, the number of switching operations can be reduced. At the same time, adjusting the number of operating pumps according to real-time water supply demand can effectively match the water supply volume with the demand and avoid over-supply. Minimizing equipment losses can be expressed by the following formula:

[0131]

[0132] where c 2 is the estimated fixed cost and maintenance cost per hour of the pump, which can be calculated by dividing the total cost of purchasing and repairing the pump by the total available working hours of the pump, w i,j , w i,j-1 is the switching state of pump i between period j and j - 1. For w i,0 it needs to be given in advance, Δ j is the time interval for period j.

[0133] The constraints include the following conditions:

[0134] (1) The total water supply in each period should be greater than the water supply demand in that period:

[0135]

[0136] where q ij represents the flow rate of pump i in period j, d j represents the water supply demand at time j, which is the minimum water supply that must be met to ensure the service level of the water supply system.

[0137] (2) The operating flow rate of the pump is within the allowable limits:

[0138]

[0139] When the pump is closed, the flow rate of the pump is 0. After the pump is started, the flow rate of each pump should be between its minimum allowable flow rate and maximum allowable flow rate Within the formed interval, this ensures that the operation of the water pump is neither inefficient or causing pipeline blockage due to too low a flow rate, nor exceeding the pump's own bearing capacity due to too high a flow rate, thus guaranteeing the high efficiency, stability, and safety of the entire water supply system.

[0140] (3) The operating head of the water pump is greater than or equal to the minimum allowable head

[0141]

[0142] After the water pump is started, the actual operating head h of each pump ij must be greater than or equal to the minimum allowable head This ensures that the water pump can still maintain sufficient head to meet system requirements when facing different water flow rates and operating conditions, avoiding problems such as insufficient flow rate, insufficient head, or pump overload.

[0143] Observing constraint (3), for w ij being 0, constraint 3 is obviously satisfied. When w ij is 1, simplifying constraint 3 gives

[0144]

[0145] As can be seen from the above, the quadratic coefficient a of this formula i2 is negative, that is is a convex-up function of q ij . When it has a solution, its two solutions (including two identical solutions) are

[0146]

[0147] That is

[0148]

[0149] Let

[0150]

[0151] Then combining constraint (2) and constraint (3) gives

[0152]

[0153] Let

[0154] λ 1 = e·c 2 ·Δ j

[0155] λ 2 = c 2 ·Δ j

[0156] The finally established multi-objective optimization model:

[0157]

[0158] In the embodiments of the present application, the non-dominated sorting genetic algorithm is used to solve the multi-objective mixed-integer non-linear programming model. Specifically, the NSGA-II algorithm can be adopted. NSGA-II (the second generation of non-dominated sorting genetic algorithm) is a multi-objective intelligent optimization algorithm, especially designed to solve multi-objective optimization problems and multi-objective constrained optimization problems. The core idea of NSGA-II is to use non-dominated sorting and crowding distance sorting to maintain the diversity of the solution set and quickly find an approximate Pareto optimal solution set. With its excellent multi-objective optimization function and powerful global search ability, the NSGA-II algorithm has become an ideal choice for solving the optimal water supply scheduling problem. In a water supply system, multiple decision variables often need to be considered simultaneously, and multiple conflicting objectives such as minimizing electricity costs and reducing equipment losses need to be balanced. The NSGA-II algorithm can efficiently handle such complex problems, find a set of balanced and optimized solution sets, and thus achieve the efficient operation and cost control of the water supply system.

[0159] Refer to Figure 3 , Figure 3 shows an overall implementation schematic diagram of a scheduling method for a water pump unit provided by the embodiments of the present application.

[0160] In the embodiments of the present application, based on the actual operation data of H Water Plant in G City from January 1, 2021 to November 30, 2022, an example analysis of demand forecasting and optimal scheduling is carried out to verify the effect of the technical solution of the present application.

[0161] In the embodiment, 8 industrial frequency water pumps (I = 8) are selected for scheduling research, and the research time period is divided into 24 time periods (J = 24) at hourly intervals, and the time interval of each time period is set to 1 hour (△ = 1). Taking the actual situation on November 30, 2022 as the scheduling comparison, combined with the actual operation data provided by the enterprise, the algorithm is implemented and solved using the Python programming language.

[0162] (I) Example analysis of demand forecasting

[0163] The demand data comes from the data of 698 days × 24 hours, a total of 16,752 samples from January 1, 2021 to November 30, 2022, including features such as water volume, water temperature, and water pressure, without missing values. As shown in Figure 4, where Figure 4a is a time-domain graph, Figure 4b is a frequency-domain graph.

[0164] 1. Data preprocessing and weight calculation

[0165] Before formally conducting the analysis of the demand forecasting example, the time series data was first preprocessed to ensure the quality and applicability of the data. The specific preprocessing steps are as follows:

[0166] (1) Outlier detection and correction: The 3σ principle (i.e., values greater than or less than three standard deviations from the mean are considered outliers) was used to identify outliers in the time series data. For the detected outliers, the three-standard-deviation bounds were corrected to avoid adverse effects of these values on the prediction results.

[0167] (2) Data normalization: To improve the convergence speed and prediction accuracy of model training, the time series data was normalized to map it into the interval [0, 1]. This process was achieved by subtracting the minimum value from the original data and dividing by the difference between the maximum and minimum values.

[0168] (3) Creating a sliding window dataset: A dataset with a sliding window size of 24 was constructed from the normalized time series data. This dataset was used for subsequent demand forecasting, where each window contained historical time point data as input features for predicting future demand.

[0169] (4) Peak and sample weight calculation: As Figure 5 shown, Figure 5 the normalized demand and sample weights for a total of 240 hours from November 21, 2022 to November 30, 2022 are shown.

[0170] 2. Model construction and demand forecasting

[0171] First, the processed time series dataset was divided into a training set and a test set in a ratio of 8:2, i.e., 80% of the data was used to train the model, and the remaining 20% was used to evaluate the performance of the model. Subsequently, two long short-term memory (LSTM) models were constructed using the PaddlePaddle deep learning framework. Each model took demand as a single input feature, with 2 LSTM hidden layers, each layer containing 64 neurons. One model used the original mean squared error (MSE) loss function, called the "original LSTM" model, and the other model used the weighted MSE loss function, called the "weighted LSTM" model.

[0172] For model training, to ensure the reproducibility and comparability of experimental results, the random seed was fixed in this example, and the same model training parameters were configured. The Adam optimizer was uniformly used, the learning rate was set to 0.005, and the number of training iterations was set to 100.

[0173] After training, the model made predictions on the test set and denormalized the prediction results. To quantify the prediction accuracy and the model's fitting effect, the root mean square error (RMSE) and the coefficient of determination (R 2 ) were selected as two performance metrics to compare the standard LSTM model and the weighted LSTM model. The experimental results are shown in Table 1.

[0174] Table 1

[0175]

[0176] Taking the period from November 21, 2022 to November 30, 2022 as the demonstration sample, the true data values, the predicted values of the original LSTM model, and the predicted values of the weighted LSTM are as Figure 6 shown.

[0177] From the above charts, by comparing the model effects and prediction results of the true values, the original LSTM model, and the weighted LSTM model, it can be intuitively concluded that both models generally perform close to the actual values, with good effects and relatively consistent performances. However, for the prediction of peaks, the enhanced LSTM model is significantly better than the original LSTM model.

[0178] (2) Analysis of Optimization Scheduling Embodiments

[0179] 1. Parameter Estimation

[0180] In the initial stage of the optimization scheduling experiment, this application first conducted an in-depth analysis of the characteristic curves of the water pumps. Based on 14,142 effective sample data obtained during the period from November 21, 2022 to November 30, 2022, curve fitting was carried out to calculate the coefficients of the relevant quadratic functions, accurately capturing the relationships between the head-flow rate and the efficiency-flow rate of 8 water pumps. The fitting effect parameters are as Figure 7 shown. After that, according to the rated flow rate limit and the minimum head limit of the water pumps, the estimated parameters of the water pumps were calculated as shown in Table 2.

[0181] Table 2

[0182]

[0183]

[0184] After completing the fitting and estimation of the water pump characteristic parameters and estimating and calculating according to the actual situation, the unit electricity price c 1 = 1.2, and the unit equipment purchase and maintenance cost c 2 = 100.

[0185] 2. Optimization Scheduling

[0186] For the instance of the water pump scheduling problem, it should be noted first that although the flow rate is essentially a continuous variable, it is often treated as a discrete variable in practice because the loss of precision is basically negligible, while great convenience can be obtained. When using intelligent optimization algorithms to solve the problem, treating it as a discrete variable is conducive to the rapid solution of the problem. Therefore, the flow rate is set as a discrete variable here. The number of iterations of the NSGA-II algorithm is set to 5000, the population size is set to 50, the mutation probability is 0.9, and the crossover probability is 0.5. The Pareto optimal front finally solved is as Figure 8 shown.

[0187] The Pareto optimal front solved by NSGA-II is a set of optimal solutions. Here, by introducing the electricity price cost coefficient (unit electricity price) and the equipment cost coefficient (unit equipment purchase and maintenance cost), the two objectives of the solution results are made additive and comparable. The decision principle is to select the solution with the largest sum of the two. After selection and calculation, the optimal flow rate scheduling plan of 8 water pumps within 24 periods is finally obtained. The detailed scheduling results are shown in Table 3.

[0188] Table 3

[0189]

[0190]

[0191] After calculating the scheduling results, the cost comparison results before and after scheduling are shown in Table 4. By comparing the electricity cost, equipment cost and total cost before and after the optimal scheduling, it is found that the electricity cost, equipment cost and total cost after scheduling are optimized by 7.12%, 12.50% and 8.18% respectively. It can be seen that the optimal scheduling scheme based on demand prediction in this application has an obvious improvement effect compared with that before scheduling, directly reducing the economic expenditure, and can more effectively reduce energy consumption and equipment loss, which conforms to the trend of sustainable development.

[0192] Table 4

[0193]

[0194]

[0195]

[0196] On the basis of the original modeling and experimental ideas, by changing the data set and the solution algorithm, the prediction algorithms include linear regression (LR), decision tree (DTR), random forest (RFR), gradient boosting machine (GBR), support vector machine (SVR), gated recurrent unit (GRU) and their original (O), weighted (W) and combined algorithms. The new demand prediction experimental results are shown in Table 5:

[0197] Table 5

[0198]

[0199] The optimization algorithms include various improved non-dominated sorting genetic algorithms (NSGA2, NSGA3, U-NSGA3, R-NSGA3), multi-objective differential evolution algorithms (MOEA / D), and mixed-integer linear programming solvers (MILP+COPT). The new scheduling results are shown in Table 6:

[0200] Table 6

[0201]

[0202]

[0203] In summary, the present application has at least the following advantages:

[0204] 1. The present application uses a weighted long short-term memory network model (Weighted LSTM) for water supply demand prediction. By introducing a weighted loss function, it effectively solves the limitations of the traditional LSTM model in dealing with water supply demand prediction with unbalanced importance of periods, and significantly improves the prediction accuracy of water supply demand during peak periods.

[0205] 2. The present application establishes a multi-objective optimal scheduling model based on the physical characteristics of pumps. By weighing multiple factors such as electricity cost, equipment loss, and water supply demand, it finds the optimal scheduling plan, bringing the advantage of being able to effectively balance the operating efficiency and economy of the water supply system, and significantly improving the optimization effect.

[0206] 3. The present application uses the NSGA-II multi-objective optimization algorithm to solve the optimal scheduling model. The NSGA-II algorithm can find a set of balanced and optimized solution sets (Pareto optimal solution sets) among numerous effective solutions. With its excellent multi-objective optimization function and powerful global search ability, it significantly improves the solution efficiency.

[0207] Referring to Figure 9 , an embodiment of the present application also provides a scheduling device for a water pump unit, including:

[0208] An acquisition unit 910, configured to acquire historical operation data of the water supply system and operation condition data of the water pump unit; wherein, the historical operation data includes historical water consumption data;

[0209] A prediction unit 920, configured to train a corresponding water supply demand prediction model according to the historical operation data, and predict the water supply flow demand in a future period through the trained water supply demand prediction model;

[0210] A processing unit 930, configured to determine physical characteristic information of the water pump unit according to the operating condition data; wherein, the physical characteristic information includes an association curve between performance parameters when the water pump unit operates at different speeds, and the performance parameters include the head, efficiency, and flow rate of the water pump unit.

[0211] An establishing unit 940, configured to establish a multi-objective mixed-integer non-linear programming model for the scheduling of the water pump unit, with the water supply flow demand and the operation limit of the water pump unit as constraint conditions, and the minimization of power cost and the minimization of equipment loss as optimization objectives.

[0212] An execution unit 950, configured to solve the multi-objective mixed-integer non-linear programming model, and schedule the water pump unit in a future period according to the solution result.

[0213] It can be understood that the content in the above method embodiments is applicable to the device embodiments of the present application. The functions specifically implemented by the device embodiments of the present application are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0214] Referring to Figure 10 , an electronic device is provided in an embodiment of the present application, including:

[0215] At least one processor 1010;

[0216] At least one memory 1020, configured to store at least one program;

[0217] When at least one program is executed by at least one processor 1010, at least one processor 1010 is caused to implement the above scheduling method of the water pump unit.

[0218] Similarly, the content in the above method embodiments is applicable to the electronic device embodiments of the present application. The functions specifically implemented by the electronic device embodiments of the present application are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0219] An embodiment of the present application further provides a computer-readable storage medium, in which a program executable by a processor 1010 is stored, and the program executable by the processor 1010 is used to execute the above scheduling method of the water pump unit when executed by the processor 1010.

[0220] Similarly, the content in the above method embodiments is applicable to the computer-readable storage medium embodiments of the present application. The functions specifically implemented by the computer-readable storage medium embodiments of the present application are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those of the above method embodiments.

[0221] In some alternative embodiments, the functions / operations recited in the block diagrams may not occur in the order presented in the operational illustrations. For example, depending on the functions / operations involved, two blocks shown in succession may actually be executed substantially simultaneously or the blocks can sometimes be executed in reverse order. Additionally, the embodiments presented and described in the flowcharts of the present application are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated in which the order of various operations is altered and in which sub-operations described as part of a larger operation are performed independently.

[0222] Furthermore, although the present application has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present application. Rather, given the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skill of an engineer. Thus, those skilled in the art can implement the present application as set forth in the claims without undue experimentation. It should also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present application, which is determined by the full scope of the appended claims and their equivalents.

[0223] If a function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present application, in essence or the part that contributes to the prior art or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0224] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definitional sequence of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. As used in this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device.

[0225] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which a program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0226] It should be understood that various parts of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well-known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0227] In the above description of this specification, the descriptions referring to the terms "one embodiment / example", "another embodiment / example", or "certain embodiments / examples", etc., mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0228] Although embodiments of the present application have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present application. The scope of the present application is defined by the claims and their equivalents.

[0229] The above is a specific description of the preferred embodiments of the present application, but the present application is not limited to the embodiments. Those skilled in the art can make various equivalent deformations or substitutions without departing from the spirit of the present application, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present application.

Claims

1. A method for dispatching a water pump unit, characterized in that: The method comprises: Collecting historical operation data of the water supply system and operating condition data of the water pump unit; wherein the historical operation data includes historical water consumption data; According to the historical operation data, a corresponding water supply demand prediction model is trained, and the water supply flow demand in the future period is predicted by the trained water supply demand prediction model; Determine the physical characteristic information of the water pump unit according to the operating condition data; wherein the physical characteristic information includes a correlation curve between performance parameters of the water pump unit when the water pump unit is running at different speeds, and the performance parameters include the head, efficiency and flow rate of the water pump unit; Taking the water supply flow demand and the operation limit of the water pump unit as constraints, and minimizing the power cost and equipment loss as optimization goals, a multi-objective mixed integer nonlinear programming model for the water pump unit scheduling is established; The multi-objective mixed integer nonlinear programming model is solved, and the water pump unit is scheduled in the future time period according to the solution result.

2. A method for dispatching a water pump unit according to claim 1, characterized in that: After collecting the historical operation data of the water supply system and the operation condition data of the water pump unit, the method further includes: Preprocessing the historical operation data and the operation condition data; The preprocessing operation includes at least one of outlier detection and correction, data normalization, and creation of a sliding window data set.

3. A method for dispatching a water pump unit according to claim 1, characterized in that: The step of training a corresponding water supply demand prediction model according to the historical operation data comprises: Divide the historical water consumption data into time periods to obtain sub-water consumption data in multiple time periods; wherein each time period includes a number of sampling time points, and the number of sampling time points in each time period is the same; Determine the kurtosis value corresponding to the first time period according to the first sub-water consumption data in the first time period and the second sub-water consumption data in the time period before the first time period; wherein the first time period is an arbitrary time period; Determining the training weights corresponding to each of the time periods according to the kurtosis value; Inputting the sub-water consumption data of each time period into a water supply demand prediction model, and predicting the water consumption of the next time period of each time period by the water supply demand prediction model to obtain predicted water consumption data; Determine the initial loss value of the time period according to the sub-water consumption data of the next time period of each time period and the predicted water consumption data corresponding to the time period; Performing weighted summation on the initial loss values ​​according to the training weights to obtain a summary loss value; According to the aggregated loss value, the parameters of the water supply demand prediction model are updated to obtain a trained water supply demand prediction model.

4. A method for dispatching a water pump unit according to claim 3, characterized in that: The determining, according to the first sub-water consumption data in the first time period and the second sub-water consumption data in the time period before the first time period, the kurtosis value corresponding to the first time period comprises: Correlating the sampling time points of the first time period with the sampling time points of the second time period in sequence; Comparing the magnitude relationship between the first sub-water consumption data and the second sub-water consumption data at each corresponding sampling time point; If at the sampling time point, the first sub-water consumption data is greater than the second sub-water consumption data, the sampling time point is determined as the target time point; The kurtosis value corresponding to the first time period is determined according to the number of target time points corresponding to the first time period.

5. A method for dispatching a water pump unit according to claim 3, characterized in that: Determining the training weights corresponding to the respective time periods according to the kurtosis values ​​includes: Determine a minimum first value and a maximum second value from the kurtosis values ​​corresponding to each of the time periods; Calculating a first difference between the first value and the second value, and calculating a second difference between the kurtosis value corresponding to each time period and the first value; Calculating the ratio of the second difference corresponding to each of the time periods to the first difference, and determining the sum of the ratio and 1 as the preliminary weight corresponding to the time period; The preliminary weights of each of the time periods are normalized to obtain the training weights corresponding to each of the time periods.

6. A method for dispatching a water pump unit according to claim 1, characterized in that: The water supply demand prediction model is built based on LSTM.

7. A method for dispatching a water pump unit according to claim 1, characterized in that: The solving of the multi-objective mixed integer nonlinear programming model comprises: The multi-objective mixed integer nonlinear programming model is solved by a non-dominated sorting genetic algorithm.

8. A dispatching device for a water pump unit, characterized in that: The device comprises: A collection unit, used to collect historical operation data of the water supply system and operating condition data of the water pump unit; wherein the historical operation data includes historical water consumption data; A prediction unit, used to train a corresponding water supply demand prediction model according to the historical operation data, and predict the water supply flow demand in the future period by using the trained water supply demand prediction model; A processing unit, configured to determine the physical characteristic information of the water pump unit according to the operating condition data; wherein the physical characteristic information includes a correlation curve between performance parameters of the water pump unit when the water pump unit is operated at different speeds, and the performance parameters include the head, efficiency and flow rate of the water pump unit; An establishing unit is used to establish a multi-objective mixed integer nonlinear programming model for the scheduling of the water pump unit, taking the water supply flow demand and the operation limit of the water pump unit as constraints, and taking minimizing the power cost and minimizing the equipment loss as optimization goals; The execution unit is used to solve the multi-objective mixed integer nonlinear programming model and schedule the water pump unit in the future time period according to the solution result.

9. An electronic device, characterized in that: include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements a scheduling method for a water pump unit as described in any one of claims 1-7.

10. A computer-readable storage medium storing a program executable by a processor, characterized in that: The processor-executable program is used to implement a water pump unit scheduling method as described in any one of claims 1-7 when executed by the processor.

Images