Water intake and consumption prediction method and device coupled with system dynamics and deep learning

Abstract

The invention provides a water taking and consumption prediction method and device coupled with system dynamics and deep learning, and the method comprises the steps: building a water resource system dynamics model based on the water demands of industrial development, domestic water and agricultural irrigation, obtaining a first unit time period water taking amount of a target area determined by the water resource system dynamic model; the influence factor data of the target area and the first unit time period water taking amount of the target area are input into a pre-trained water amount prediction model, the second unit time period water taking amount of the target area is determined according to the output of the water amount prediction model, and the duration of the second unit time period is smaller than that of the first unit time period. According to the method, accurate prediction of the water taking amount is realized by coupling the water resource system dynamic model and the long-short-term memory network.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a method and device for predicting water consumption by coupling system dynamics and deep learning. Background Art

[0002] With the development of social economy, people's demand for water resources has also increased, so it is particularly important to carry out reasonable planning and management of water resources.

[0003] The prediction of daily water demand can help implement the most stringent water resource management system, formulate water resource optimization allocation plan, and accelerate the control of total water resource consumption and intensity. There are many factors that affect water demand, including socio-economic conditions and meteorological conditions.

[0004] Traditional water demand prediction models mostly take regional water demand as the target, such as considering the socio-economic development status of the region (population density, regional GDP, etc.), and predicting the daily water demand of the city based on statistical methods.

[0005] However, as cities become larger and larger, the spatial variation of water use within cities cannot be ignored. At the same time, as society's awareness of sustainable development of water resources and water ecological protection has gradually increased in recent years, in order to ensure the ecological flow downstream, the formulation of social water use plans is also affected by the state of natural water resources. Therefore, it is difficult to meet the current demand for human water use activity data in water resources management and planning by predicting the total water use of cities based only on socio-economic conditions. Summary of the invention

[0006] The present invention provides a method and device for predicting water consumption by coupling system dynamics and deep learning, so as to solve the defect that the prediction methods in the prior art are difficult to meet the demand for human water use activity data in current water resources management and planning, and realize a method for predicting water consumption at a fine spatiotemporal scale.

[0007] The present invention provides a method for predicting water consumption by coupling system dynamics and deep learning, comprising: A water resource system dynamics model is constructed based on the water demand for industrial development, domestic water use and agricultural irrigation, and the water consumption in the first unit period of the target area determined by the water resource system dynamics model is obtained; The influencing factor data of the target area and the water consumption in the first unit time period of the target area are input into a pre-trained water quantity prediction model, and the water consumption in the second unit time period of the target area is determined according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

[0008] According to a method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention, the step of constructing a water resource system dynamics model based on water demand for industrial development, domestic water and agricultural irrigation specifically includes: Determine the water demand for industrial development based on industrial added value and water consumption per unit of industrial added value; Determine the water demand for domestic water based on regional population and water consumption per unit of population; Determine the water demand for agricultural irrigation based on the irrigated area and water consumption per unit irrigated area; Constructing a social water demand model according to the sum of the water demand for industrial development, the water demand for domestic water and the water demand for agricultural irrigation, and determining a social water withdrawal model according to the social water demand model; A water resource system dynamics model is constructed based on the social water demand model and the social water withdrawal model.

[0009] According to a method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention, the influencing factor data of the target area includes watershed attribute data, meteorological data, hydrological data and socio-economic development data of the target area.

[0010] According to a method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention, before the step of inputting the influencing factor data of the target area and the water consumption of the target area in the first unit period into a pre-trained water consumption prediction model, the method further includes: Collecting historical impact factor data of the target area and preprocessing the historical impact factor data of the target area; Based on the preprocessed historical impact factor data, a long short-term memory network is trained as a water volume prediction model.

[0011] According to a method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention, the step of collecting historical impact factor data of the target area specifically includes: Performing grid processing on the research area to which the target area belongs; Determine a grid corresponding to the target area, and use a plurality of grids within a preset range centered on the grid corresponding to the target area as an influence area of ​​the target area; The historical impact factor data of the target area is determined based on the historical impact factor data of each grid in the impact area of ​​the target area.

[0012] According to a method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention, the step of preprocessing the historical impact factor data of the target area specifically includes: Performing spatiotemporal interpolation of meteorological data, hydrological data and socio-economic development data in the historical impact factor data of the target area; Spatial interpolation is performed on the watershed attribute data in the historical impact factor data of the target area.

[0013] The present invention also provides a water consumption prediction device for coupling system dynamics and deep learning, comprising: A determination module is used to construct a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtain the water consumption in the first unit time period of the target area determined by the water resource system dynamics model; The prediction module is used to input the influencing factor data of the target area and the water consumption of the target area in the first unit time period into a pre-trained water quantity prediction model, and determine the water consumption of the target area in the second unit time period according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting water consumption by coupling system dynamics and deep learning as described in any one of the above-mentioned methods is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a water consumption prediction method of coupling system dynamics and deep learning as described in any one of the above.

[0016] The present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-mentioned methods for predicting water consumption by coupling system dynamics and deep learning.

[0017] The water consumption prediction method and device of the coupled system dynamics and deep learning provided by the present invention determine the water consumption of the target area in a longer period of time by constructing a water resource system dynamics model that takes into account industrial development, agricultural irrigation and domestic water use, and then inputs it into a pre-trained water consumption prediction model. The water consumption data of a shorter period of time is predicted based on the water consumption and water resource influencing factor data of a longer period of time, so that the water consumption data of the second unit time period obtained integrates the changes in the coordinated development of social economy and nature in the target area within a certain period of time, thereby obtaining a more accurate prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 It is one of the flow diagrams of the method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention; Figure 2 It is a schematic diagram of grid division of the water consumption prediction method of coupling system dynamics and deep learning provided by the present invention; Figure 3 It is a schematic diagram of a water resource system dynamics model of the water consumption prediction method of the coupling system dynamics and deep learning provided by the present invention; Figure 4 This is the second flow chart of the method for predicting water consumption by coupling system dynamics and deep learning provided by the present invention; Figure 5 It is a comparison chart of prediction results of the water consumption prediction method based on coupled system dynamics and deep learning provided by the present invention; Figure 6 It is a schematic diagram of the structure of the water consumption prediction device for coupling system dynamics and deep learning provided by the present invention; Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Combine the following Figures 1 to 5 The water consumption prediction method of the present invention coupling system dynamics and deep learning is introduced. Figure 1 As shown, including: Step 101, constructing a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtaining the water consumption in the first unit time period of the target area determined by the water resource system dynamics model; The water resources system dynamics model is a mechanistic model that takes water resources as the research object, examines them in the form of a system, and studies the relationship between internal and external factors in the system.

[0022] In this embodiment, a macro-social water resource system dynamics model is pre-constructed that takes into account industrial development, people's domestic water use and irrigation water, wherein the output of the water resource system dynamics model includes at least the water consumption in the first unit time period of the target area.

[0023] Optionally, the target area may be a study area where water withdrawal is to be predicted, e.g. Figure 2 As shown by the red dot represented by W6, at this time, the water consumption in the first unit period of the target area is the water consumption in the first unit period of the study area whose water consumption is to be predicted.

[0024] Optionally, the target area includes a study area where water withdrawal is to be predicted, e.g. Figure 2 As shown in the red shaded area containing the red dot represented by W6, at this time, the water consumption in the first unit period of the target area represents the water consumption in the first unit period of the range of the study area where the water consumption to be predicted is located, that is, in subsequent studies, the influence of the surrounding environment of the study area where the water consumption to be predicted is also considered.

[0025] Step 102, input the influencing factor data of the target area and the water consumption of the target area in the first unit time period into a pre-trained water quantity prediction model, and determine the water consumption of the target area in the second unit time period according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

[0026] A water volume prediction model is obtained by pre-training so that the water volume prediction model can predict the water volume in a second unit period according to the water volume in a first unit period and the influencing factor data of the target area, wherein the duration of the second unit period is shorter than the first unit period.

[0027] Optionally, the first unit period is set to year, and the second unit period is set to day, that is, the daily water consumption of the target area is predicted based on the annual water consumption of the target area.

[0028] Optionally, the water volume prediction model may be a LSTM (Long Short-Term Memory) network.

[0029] Optionally, the influencing factor data of the target area are factors such as the natural environment and human social activities that affect the water volume in the target area, and may include one or more of the watershed properties, meteorological data and socio-economic development information of the target area, subject to the data that can be actually obtained.

[0030] The present invention determines the water consumption of the target area over a longer period of time by constructing a water resource system dynamics model that takes into account industrial development, agricultural irrigation and domestic water use, and then inputs it into a pre-trained water quantity prediction model. The water consumption data of a shorter period of time is predicted based on the water consumption over a longer period of time and the water resource influencing factor data. The water consumption data of the second unit time period obtained integrates the changes in the coordinated development of social economy and nature in the target area within a certain period of time, thereby obtaining a more accurate prediction result.

[0031] In the water consumption prediction method of coupling system dynamics and deep learning of the present invention, the step of constructing a water resource system dynamics model based on the water demand for industrial development, domestic water and agricultural irrigation specifically includes: Determine the water demand for industrial development based on industrial added value and water consumption per unit of industrial added value; Optionally, a water resource system dynamics model constructed in this embodiment is as follows: Figure 3 As shown, in order to construct a water resources system dynamics model, it is necessary to determine the main socioeconomic factors that affect social water demand.

[0032] Optionally, identify the main socio-economic factors affecting water demand by including water availability. S W ,population S P , industrial added value S G , agricultural irrigation area S R And the irrigation water requirement per unit area WDI R etc. as important factors affecting social water demand.

[0033] In a feasible implementation, the impact of industrial development, domestic water use and agricultural irrigation on water resources is comprehensively considered. Therefore, the social water demand in the water resources system dynamics model is WD It also consists of three parts: water demand for industrial development, water demand for domestic water and water demand for agricultural irrigation.

[0034] Alternatively, water demand for industrial development is industrial value added S G Water consumption per unit of industrial added value WDI G The product of .

[0035] Optionally, the water consumption per unit of industrial added value represents the change in water consumption per unit of industrial added value, and in this implementation manner is specifically set to the water consumption per 10,000 yuan of industrial added value.

[0036] Alternatively, industrial added value SG Determined based on dynamic changes in regional population.

[0037] For example, construct a population equation that follows the Malthusian law of population growth to model regional populations. S P Dynamic changes: ; On this basis, the economic development equation is constructed. According to the experience of developed and developing countries in economic development, economic growth will not always be high-speed. To a certain extent, the development speed will slow down. Therefore, the regional per capita GDP level is predicted to be s The type function describes a s The characteristic of the curve is that GDP per capita grows faster at first, slows down in the middle stage, and finally almost stops growing. S G The dynamic changes of can be expressed as: .

[0038] Determine the water demand for domestic water based on regional population and water consumption per unit of population; Optionally, the domestic water demand is the regional population S P Water consumption per unit population WDI P The product of .

[0039] Optionally, the water consumption per unit population represents the change in domestic water demand resulting from each unit change in population.

[0040] Determine the water demand for agricultural irrigation based on the irrigated area and water consumption per unit irrigated area; Alternatively, the water demand for agricultural irrigation is the irrigated area S R and water consumption per unit irrigation area WDI R The product of .

[0041] Optionally, fit irrigated area based on historical data S R Dynamic changes: .

[0042] Optionally, simulate water usage per unit irrigated area WDI R Dynamic changes: .

[0043] Constructing a social water demand model according to the sum of the water demand for industrial development, the water demand for domestic water and the water demand for agricultural irrigation, and determining a social water withdrawal model according to the social water demand model; In summary, the social water demand model in the water resources system dynamics model is constructed WD It can be expressed as: ; Due to the continuous development of social economy, social water demand is also changing with the dynamic changes of population and industrial production. Based on historical data analysis, water consumption per 10,000 yuan of industrial added value WDI G and water consumption per capita WDI P The changes are small and gradually tend to be stable, so constant value calculations are used in future forecasts.

[0044] On this basis, the social water withdrawal model WU It can be expressed as: ; In the formula, WD is the total water demand of society, R 1 is the water resource supply-demand ratio, which can be set according to actual demand and is set to 1 in this implementation.

[0045] In the above formulas, R 1, p The value of 0 needs to be determined based on the relevant prediction model established; t is the forecast year; To determine the parameters, the historical test method can be used to verify the effectiveness of the prediction model; the estimated value of the parameter is used as the initial value, and the simulated value of the model is compared with the actual value of the verification year (historical statistical value series). The parameter value with a smaller relative error is the value of each parameter in the model.

[0046] A water resource system dynamics model is constructed based on the social water demand model and the social water withdrawal model.

[0047] The above method can be used to construct the social water demand model in the water resources system dynamics model. WD and social water withdrawal models WU , thus predicting the water consumption of the target area in the first unit time period.

[0048] In addition, the water resources system dynamics model can also construct a water resources equation to describe the available water resources in the basin. S W dynamic balance.

[0049] The available amount of water resources is replenished by inflow and precipitation. Due to the consumption of water in the region due to social water withdrawal, part of the unconsumed water is returned to the river basin or underground reservoir through drainage, and finally discharged out of the region through outflow. That is, the change in the available amount of water resources includes the change in the available amount of water resources caused by the influence of precipitation and the influence of water withdrawal from the water resource system by the social and economic system. The specific formula is expressed as: ; In the formula, I is the regional inflow of water, It is the runoff produced by rainfall after a series of hydrological processes. O is the regional outflow water volume. The water resource equation is simulated using the distributed hydrological model Wetspa, which can simulate the dynamic changes of the available water resources in each grid within the study area under the driving of meteorology.

[0050] The present invention realizes the prediction of water consumption of a target area in the first unit time period by constructing a social water demand model that couples the water demands of industrial development, domestic water use and agricultural irrigation; and constructs a social water withdrawal model based on the social water demand model to predict the water withdrawal of the target area in the first unit time period, thereby obtaining the water withdrawal of the target area in the first unit time period.

[0051] In the water consumption prediction method of coupling system dynamics and deep learning of the present invention, the influencing factor data of the target area include watershed attribute data, meteorological data, hydrological data and socio-economic development data of the target area.

[0052] Optionally, the watershed attribute data includes but is not limited to the geographic elevation and land use type of the target area.

[0053] Optionally, the meteorological data includes, but is not limited to, rainfall, temperature, relative humidity, and sunshine duration data of the target area.

[0054] Optionally, the hydrological data includes but is not limited to soil moisture data of the target area.

[0055] Optionally, the socio-economic development data include, but are not limited to, population density, regional gross product, night light index and irrigation area data of the target area.

[0056] In the water consumption prediction method of coupling system dynamics and deep learning of the present invention, before the step of inputting the influencing factor data of the target area and the water consumption of the first unit period of the target area into the pre-trained water consumption prediction model, it also includes: Collecting historical impact factor data of the target area and preprocessing the historical impact factor data of the target area; The impact factor data of the target area in the historical time period are collected as the historical impact factor data, and are preprocessed to construct a data set for training the water volume prediction model, wherein the length of the historical time period is not less than the first unit period.

[0057] It is understandable that, since the water volume prediction model needs to predict the data of the second unit period based on the data of the first unit period, the historical impact factor data can be collected based on the second unit period. For example, the impact factor data of the target area is collected every day in the historical period, and the impact factor data of 365 consecutive days is collected as the impact factor data of the target area every year.

[0058] Optionally, preprocessing includes data denoising and data interpolation to obtain sufficient basic data.

[0059] Optionally, the preprocessing includes normalization processing to map different types of influencing factor data and water consumption data to a range of [0,1].

[0060] Based on the preprocessed historical impact factor data, a long short-term memory network is trained as a water volume prediction model.

[0061] Taking the historical influencing factor data of the target area and the water consumption of the target area in the first unit period as input, and the water consumption of the target area in the second unit period as the target value, an LSTM model is constructed and trained to form a water volume prediction model that can simulate the water consumption allocation in the second unit period based on the water consumption in the first unit period.

[0062] In a specific implementation, a network with 2 hidden layers and 64 hidden layer nodes is constructed. The data series of water consumption in the first unit period is normalized and expressed as ,in, x is the water intake in the first unit period, x n For the n The water withdrawal per second unit period, to The sum of is 1.

[0063] On this basis, compile the network, specify the optimization algorithm and loss function, and convert the defined simple layer sequence into an efficient matrix transformation sequence so that it can be executed in the computer. In this example, the root mean square error is used as the loss function, tanh is used as the activation function, and the Adam algorithm is used as the optimizer.

[0064] After the network is compiled, the weights can be adjusted on the training data set. Prepare the training data, including the input of various influencing factor data (X) and the daily allocation ratio within the year (Y), and set the number of iterations to 300. Use the back propagation algorithm to train the network and optimize it according to the optimization algorithm and loss function specified when compiling the model.

[0065] After the training is completed, the model is evaluated. During the test period, only the influencing factors (X) and the trained network are provided to predict the daily allocation ratio (Y) within the year. The simulation results are compared with the actual annual allocation of water consumption, and the evaluation indicators are calculated to evaluate the network.

[0066] The water volume prediction model trained in the above manner can predict the daily water withdrawal allocation ratio within the year based on the input influencing factors and the annual water withdrawal data, and then calculate the daily water withdrawal within the year based on the daily water withdrawal allocation ratio within the year and the total water withdrawal data within the year.

[0067] In other feasible implementations, the water volume prediction model may also be trained to directly output the daily water consumption within a year.

[0068] In addition, in order to obtain sufficient data for training the water volume prediction model, the historical impact factor data of the area with the same water use type as the target area can also be used as the historical impact factor data of the target area. The water use type includes water source, water use type, and daily water use monitoring value.

[0069] In the water consumption prediction method of coupling system dynamics and deep learning of the present invention, the step of collecting historical impact factor data of the target area specifically includes: Performing grid processing on the research area to which the target area belongs; like Figure 2 As shown, the study area is delineated, wherein the study area includes the target area. The study area is converted into raster data with a unit of 1 km in ArcGIS software and divided into grids.

[0070] Determine a grid corresponding to the target area, and use a plurality of grids within a preset range centered on the grid corresponding to the target area as an influence area of ​​the target area; The grid to which the target area belongs is determined as the grid corresponding to the target area, and the area within a preset range centered on the grid corresponding to the target area is used as the influence range of the target area.

[0071] Optionally, the preset range is determined according to the size of each grid when dividing the grid. In this embodiment, the grids within a 5×5 range centered on the grid corresponding to the target area are used as the influence area of ​​the target area, such as Figure 2 Shown in red area.

[0072] The historical impact factor data of the target area is determined based on the historical impact factor data of each grid in the impact area of ​​the target area.

[0073] The historical impact factor data of each grid in the influence area of ​​the target area are collected, and the historical impact factor data of all grids in the influence area of ​​the target area are used as the historical impact factor data of the target area, so that when predicting the water consumption, not only the changes in the water volume of the target area itself but also the changes in the water volume of the radiation area related to the target area are considered.

[0074] It can be understood that when the water volume prediction model trained with the data set is constructed in this way, the influencing factor data of the image area of ​​the target area is input as the influencing factor data of the target area when it is used.

[0075] The present invention constructs a daily water consumption prediction model by statistically analyzing the factors affecting water use at a grid scale. The model prediction result is a grid scale, which can be applied to the calculation of hydrological models considering water use activities at any scale, better simulate the water cycle process under the interference of human activities, and improve the accuracy of hydrological forecasting.

[0076] In the water consumption prediction method of coupling system dynamics and deep learning of the present invention, the step of preprocessing the historical impact factor data of the target area specifically includes: Performing spatiotemporal interpolation of meteorological data, hydrological data and socio-economic development data in the historical impact factor data of the target area; Spatial interpolation is performed on the watershed attribute data in the historical impact factor data of the target area.

[0077] In order to construct the dataset used to train the water volume prediction model and ensure the consistency and completeness of the spatiotemporal accuracy of the input data in the dataset, the meteorological data, hydrological data, and socioeconomic development data in the historical influencing factor data of the target area are spatiotemporally interpolated, and the watershed attribute data in the historical influencing factor data of the target area are spatially interpolated.

[0078] Specifically, the point data of meteorological data is converted into surface data through the Thiessen polygon method, and then converted into grid data of uniform scale. In terms of time scale, since the prediction target is on a daily scale, linear interpolation is used for the data on the annual scale to obtain daily data information within the year.

[0079] The present invention performs temporal or spatial interpolation on different categories of historical impact factor data of the target area, thereby obtaining more complete and accurate historical impact factor data of the target area for training a water volume prediction model.

[0080] By coupling the water resource system dynamics model and the LSTM-based water volume prediction model in the above manner, the overall process of predicting the water consumption in the second unit period of the target area is as follows: Figure 4 As shown in the figure, the comparison between the predicted results and the actual results of some grids is as follows: Figure 5 shown.

[0081] The present invention introduces system dynamics theory and deep learning into the problem of water consumption prediction method by combining the system dynamics model with the long short-term memory network for daily water consumption prediction. Compared with the traditional quota statistical annual water consumption calculation method, the prediction accuracy of annual water consumption and intra-annual water consumption in long-term prediction is effectively improved, which makes the water use more accurate and reasonable, so as to be better used for water resource allocation and scheduling, and has certain social and economic value.

[0082] Combining deep learning with social water demand forecasting, and taking into account the impact of climate variables, water resources, socio-economic development, and historical water consumption, a system dynamics model that considers the coordinated development of society and nature, as well as a multi-input LSTM network, can more accurately predict the annual water consumption and annual distribution of the grid in the future long-term process. It can better provide data support for water resources management and planning, assist in the scheduling of urban and rural water use, and give full play to the value of data.

[0083] The water consumption prediction device based on the coupling system dynamics and deep learning provided by the present invention is described below. The water consumption prediction device based on the coupling system dynamics and deep learning described below and the water consumption prediction method based on the coupling system dynamics and deep learning described above can be referenced to each other.

[0084] like Figure 6 As shown, the water consumption prediction device of coupling system dynamics and deep learning includes a determination module 601 and a prediction module 602; A determination module 601 is used to construct a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtain the water consumption in the first unit time period of the target area determined by the water resource system dynamics model; The water resources system dynamics model is a mechanistic model that takes water resources as the research object, examines them in the form of a system, and studies the relationship between internal and external factors in the system.

[0085] In this embodiment, a macro-social water resource system dynamics model is pre-constructed that takes into account industrial development, people's domestic water use and irrigation water, wherein the output of the water resource system dynamics model includes at least the water consumption in the first unit time period of the target area.

[0086] Optionally, the target area may be a study area where water withdrawal is to be predicted, e.g. Figure 2 As shown by the red dot represented by W6, at this time, the water consumption in the first unit period of the target area is the water consumption in the first unit period of the study area whose water consumption is to be predicted.

[0087] Optionally, the target area includes a study area where water withdrawal is to be predicted, e.g. Figure 2 As shown in the red shaded area containing the red dot represented by W6, at this time, the water consumption in the first unit period of the target area represents the water consumption in the first unit period of the range of the study area where the water consumption to be predicted is located, that is, in subsequent studies, the influence of the surrounding environment of the study area where the water consumption to be predicted is also considered.

[0088] The prediction module 602 is used to input the influencing factor data of the target area and the water consumption of the target area in the first unit time period into a pre-trained water quantity prediction model, and determine the water consumption of the target area in the second unit time period according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

[0089] A water volume prediction model is obtained by pre-training so that the water volume prediction model can predict the water volume in a second unit period according to the water volume in a first unit period and the influencing factor data of the target area, wherein the duration of the second unit period is shorter than the first unit period.

[0090] Optionally, the first unit period is set to year, and the second unit period is set to day, that is, the daily water consumption of the target area is predicted based on the annual water consumption of the target area.

[0091] Optionally, the water volume prediction model may be a LSTM (Long Short-Term Memory) network.

[0092] Optionally, the influencing factor data of the target area are factors such as the natural environment and human social activities that affect the water volume in the target area, and may include one or more of the watershed properties, meteorological data and socio-economic development information of the target area, subject to the data that can be actually obtained.

[0093] The present invention determines the water consumption of the target area over a longer period of time by constructing a water resource system dynamics model that takes into account industrial development, agricultural irrigation and domestic water use, and then inputs it into a pre-trained water quantity prediction model. The water consumption data of a shorter period of time is predicted based on the water consumption over a longer period of time and the water resource influencing factor data. The water consumption data of the second unit time period obtained integrates the changes in the coordinated development of social economy and nature in the target area within a certain period of time, thereby obtaining a more accurate prediction result.

[0094] Figure 7An example of a physical structure diagram of an electronic device is shown in FIG. Figure 7 As shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730 and a communication bus 740, wherein the processor 710, the communication interface 720 and the memory 730 communicate with each other through the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the water consumption prediction method of coupling system dynamics and deep learning, the method comprising: constructing a water resource system dynamics model based on the water demand of industrial development, domestic water and agricultural irrigation, and obtaining the first unit time period water consumption of the target area determined by the water resource system dynamics model; inputting the influencing factor data of the target area and the first unit time period water consumption of the target area into the pre-trained water consumption prediction model, and determining the second unit time period water consumption of the target area according to the output of the water consumption prediction model, wherein the second unit time period is shorter than the first unit time period.

[0095] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0096] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the coupled system dynamics and deep learning water consumption prediction method provided by the above-mentioned methods. The method includes: constructing a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtaining the first unit time period water consumption of the target area determined by the water resource system dynamics model; inputting the influencing factor data of the target area and the first unit time period water consumption of the target area into a pre-trained water quantity prediction model, and determining the second unit time period water consumption of the target area according to the output of the water quantity prediction model, wherein the duration of the second unit time period is less than the first unit time period.

[0097] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the coupled system dynamics and deep learning water consumption prediction method provided by the above-mentioned methods, the method comprising: constructing a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtaining the first unit time period water consumption of the target area determined by the water resource system dynamics model; inputting the influencing factor data of the target area and the first unit time period water consumption of the target area into a pre-trained water quantity prediction model, and determining the second unit time period water consumption of the target area according to the output of the water quantity prediction model, wherein the duration of the second unit time period is less than the first unit time period.

[0098] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0099] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting water consumption by coupling system dynamics and deep learning, characterized in that: include: A water resource system dynamics model is constructed based on the water demand for industrial development, domestic water use and agricultural irrigation, and the water consumption per unit time period of the target area determined by the water resource system dynamics model is obtained; The influencing factor data of the target area and the water consumption in the first unit time period of the target area are input into a pre-trained water quantity prediction model, and the water consumption in the second unit time period of the target area is determined according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

2. The method for predicting water consumption by coupling system dynamics and deep learning according to claim 1, characterized in that: The steps of constructing a water resource system dynamics model based on the water demand for industrial development, domestic water and agricultural irrigation specifically include: Determine the water demand for industrial development based on industrial added value and water consumption per unit of industrial added value; Determine the water demand for domestic water based on regional population and water consumption per unit of population; Determine the water demand for agricultural irrigation based on the irrigated area and water consumption per unit irrigated area; Constructing a social water demand model according to the sum of the water demand for industrial development, the water demand for domestic water and the water demand for agricultural irrigation, and determining a social water withdrawal model according to the social water demand model; A water resource system dynamics model is constructed based on the social water demand model and the social water withdrawal model.

3. The method for predicting water consumption by coupling system dynamics and deep learning according to claim 1, characterized in that: The influencing factor data of the target area include watershed attribute data, meteorological data, hydrological data and socio-economic development data of the target area.

4. The method for predicting water consumption by coupling system dynamics and deep learning according to any one of claims 1 to 3, characterized in that: Before the step of inputting the influencing factor data of the target area and the water consumption of the target area in the first unit period into the pre-trained water quantity prediction model, the method further includes: Collecting historical impact factor data of the target area and preprocessing the historical impact factor data of the target area; Based on the preprocessed historical impact factor data, a long short-term memory network is trained as a water volume prediction model.

5. The method for predicting water consumption by coupling system dynamics and deep learning according to claim 4 is characterized in that: The step of collecting the historical impact factor data of the target area specifically includes: Performing grid processing on the research area to which the target area belongs; Determine a grid corresponding to the target area, and use a plurality of grids within a preset range centered on the grid corresponding to the target area as an influence area of ​​the target area; The historical impact factor data of the target area is determined based on the historical impact factor data of each grid in the impact area of ​​the target area.

6. The method for predicting water consumption by coupling system dynamics and deep learning according to claim 4, characterized in that: The step of preprocessing the historical impact factor data of the target area specifically includes: Performing spatiotemporal interpolation of meteorological data, hydrological data and socio-economic development data in the historical impact factor data of the target area; Spatial interpolation is performed on the watershed attribute data in the historical impact factor data of the target area.

7. A water consumption prediction device coupling system dynamics and deep learning, characterized in that: include: A determination module is used to construct a water resource system dynamics model based on the water demand for industrial development, domestic water use and agricultural irrigation, and obtain the water consumption of the target area in the first unit time period determined by the water resource system dynamics model; The prediction module is used to input the influencing factor data of the target area and the water consumption in the first unit time period of the target area into a pre-trained water quantity prediction model, and determine the water consumption in the second unit time period of the target area according to the output of the water quantity prediction model, wherein the duration of the second unit time period is shorter than the first unit time period.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the water consumption prediction method of coupling system dynamics and deep learning as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the water consumption prediction method of coupling system dynamics and deep learning as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the water consumption prediction method of coupling system dynamics and deep learning as described in any one of claims 1 to 6 is implemented.

Images