Water conservancy dispatching method and system based on neural network technology
The system optimizes water resource scheduling by dividing areas, evaluating hydrological data, and using neural networks to enhance prediction reliability, ensuring timely adjustments and accurate scheduling to support crop growth.
Patent Information
- Application Number
- CN202411136381.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-19
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-08-19
AI Technical Summary
Existing water conservancy dispatching systems using neural networks struggle with unreliable predictions during water resource scheduling, particularly in areas with groundwater level and soil moisture abnormalities, leading to unsuitable scheduling plans that do not align with actual crop growth conditions.
The system divides the dispatching area into sub-areas, evaluates hydrological data to identify unqualified regions, constructs condition evaluation coefficients, and uses neural networks to optimize water resource status prediction models, ensuring timely early warnings and accurate scheduling plans by integrating reliability coefficients and knowledge graphs.
This approach ensures reliable and accurate water resource scheduling by identifying unqualified areas, optimizing prediction models, and adjusting water distribution to improve crop growth status, reducing risks and aligning with actual conditions.
Smart Images

Figure CN118657352B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of water conservancy dispatching, and in particular to a water conservancy dispatching method and system based on neural network technology. Background Art
[0002] Water conservancy dispatch refers to a technical and management measure to formulate corresponding water conservancy project operation plans and control the use of water conservancy facilities such as reservoirs, hydropower stations, and dams according to the natural water supply, water demand, flood control safety, and ecological protection requirements of the river basin or region. The main purpose of water conservancy dispatch is to achieve the rational allocation and effective utilization of water resources, meet the needs of people's lives and economic development, and ensure the ecological and environmental health of river basins and regions. Specifically, water conservancy dispatch generally includes the following contents:
[0003] Water dispatching: According to the water resource demand of each region and the water storage of the reservoir, the water outflow of the reservoir is reasonably adjusted to meet the needs of downstream water users. Irrigation dispatching: During the agricultural irrigation season, according to the water demand of farmland and the water storage of the reservoir, a water release plan of the reservoir is formulated to meet the needs of farmland irrigation. Ecological dispatching: To ensure the health of the river's ecological environment, corresponding dispatching plans are formulated, such as water level control, water allocation, etc., to protect the river ecosystem.
[0004] A neural network is a computing model whose main job is to input data, process data, and then give a result as output. Neurons are the most basic units of a neural network. Their functions are the same as those of neurons in the human body. They receive input signals, process the data, and then give a result as output or as input for the next neuron. In order to improve the reliability of water resources scheduling, neural networks are gradually being applied in water resources management.
[0005] The invention patent with application publication number CN110472840A discloses an agricultural water conservancy dispatching method and system based on neural network technology, including a genetic algorithm improved by a particle swarm algorithm, setting multiple monitoring sensors in the agricultural irrigation area, monitoring various environmental data and various crop data in the area, and processing them to optimize the water conservancy dispatching method and system. BP Initial weights and thresholds of the neural network, and establish an improved genetic algorithm BP Neural network prediction model. BP The data predicted by the neural network model is used to determine the method of dispatching agricultural water resources in the region, so as to achieve the rational dispatch, optimization and allocation of agricultural water resources.
[0006] After determining the water resource dispatching area, the scheme described in the above application is used BWhen neural networks are used in conjunction with genetic algorithms, although the predicted output is more stable and does not fluctuate greatly, the prediction of agricultural water resources scheduling is more reliable. However, the actual input data parameters are relatively large, and the corresponding workload is also large when the actual output predicted data is output, especially when there is a certain water supply risk for crops in the scheduling area. For example, when there are certain abnormalities in the groundwater level and soil moisture content in the scheduling area, early warnings cannot be given in time. If water resources are blindly scheduled under this condition, the scheduling plan given may not be fully consistent with the actual situation, and the degree of improvement in the growth status of crops in the scheduling area is limited.
[0007] To this end, the present invention provides a water conservancy dispatching method and system based on neural network technology. Summary of the invention
[0008] 1. Technical issues to be resolved
[0009] In view of the shortcomings of the prior art, the present invention provides a water conservancy dispatching method and system based on neural network technology. Determine whether the water resource status prediction model needs to be optimized; predict the water resource status data in the dispatching area, and construct the water resource safety factor in the dispatching area based on the predicted data , if the water resources safety factor obtained is Below the safety threshold, the corresponding scheduling objectives and constraints are matched for water resource scheduling in the scheduling area, and then the trained water resource scheduling model outputs the scheduling plan. The scheduling scheme is screened and the water resources in the scheduling area are adjusted to improve the growth state of crops in the scheduling area, thus solving the technical problems raised in the background technology.
[0010] (II) Technical solution
[0011] To achieve the above objectives, the present invention is implemented through the following technical solutions:
[0012] The water conservancy dispatching method based on neural network technology includes dividing the dispatching area into several sub-areas, detecting the hydrological data in each sub-area, and constructing the condition evaluation coefficient of the sub-area in the current monitoring period based on the detection data. , in order to screen out unqualified areas in each sub-area, if the proportion of unqualified areas exceeds expectations, an early warning instruction is issued; among them, the condition evaluation coefficient The method of obtaining includes: Ds and soil moisture content Ts Perform linear normalization and map the corresponding data values to the interval Then follow the following formula: in, , is the number of data collection nodes, is the mean value of groundwater level, is the qualified reference value of groundwater level, is the mean soil moisture content, It is the qualified reference value of soil moisture content;
[0013] After receiving the warning instruction, data is collected in the dispatching area to construct the second hydrological data set, and the coefficient of variation of the collected data is used to calculate the second hydrological data set. , filter out model parameters from the second hydrological data set;
[0014] use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resource status prediction model needs to be optimized, and if necessary, optimize it;
[0015] The optimized water resource status prediction model is used to predict the water resource status data in the dispatching area, and the water resource safety factor in the dispatching area is constructed based on the predicted data. , if the water resources safety factor obtained is If the value is below the safety threshold, a dispatch instruction is issued;
[0016] After receiving the dispatch instruction, the pre-built water resource dispatch knowledge graph matches the corresponding dispatch target and constraint conditions for the water resource dispatch in the dispatch area, and then the trained water resource dispatch model outputs the dispatch plan. If there are several dispatch plans, the dispatch coefficients are used to calculate the dispatch plan. Filter it.
[0017] Furthermore, the dispatching area is predefined, and the dispatching area is divided into several sub-areas using the trained classifier combined with the terrain data in the dispatching area; a data collection point is set in each sub-area, and after setting the monitoring period, the hydrological data in the sub-area is collected at equal intervals at each data collection point, and the collected data are aggregated to construct a first hydrological data set; the condition evaluation coefficient of the sub-area in the current monitoring period is constructed from the first hydrological data set , to obtain the conditional evaluation coefficient The corresponding sub-regions are marked.
[0018] Furthermore, the data collected in the dispatching area are preprocessed and the preprocessed data are summarized as the second hydrological data set; several independent variable parameters and their values at each collection node are selected in the second hydrological data set, and the coefficient of variation of each independent variable parameter is calculated in the following manner: : Perform linear normalization on the data values of the independent variable parameters and map the corresponding data values to the interval Inside: in, refers to the independent variable parameter, For independent variable parameters in the collection node i The value on is its mean value, is the weight, , ,and , is the number of independent variable parameters;
[0019] Sort the obtained coefficients of variation of each independent variable parameter by the coefficient of variation. The highest three independent variable parameters are used as model parameters.
[0020] Further, use B The neural network constructs an initial model, extracts part of the data from the second hydrological data set as the first sample data and the second sample data respectively, uses the first sample data to train the initial model, and uses the trained initial model as the water resources status prediction model in the scheduling area; uses the second sample data to test the reliability of the water resources status prediction model, and constructs a test data set based on the acquired test data.
[0021] Furthermore, the reliability coefficient is constructed from the test data set , if the obtained reliability coefficient If the water resource status prediction model is lower than the reliability threshold, the water resource status prediction model is optimized and the optimized water resource status prediction model is obtained; the method is as follows: the recall rate in the test data set and coefficient of determination After normalization, according to the following formula: Among them, the parameter meaning is: recall factor , , coefficient of determination factor , , is a constant correction coefficient, whose value falls within .
[0022] Furthermore, the optimized water resource status prediction model is used to predict the water resource status data in the dispatching area with the model parameters as input features, a number of equally spaced prediction nodes are set, prediction data are obtained at the prediction nodes, and a water resource prediction data set is constructed after the prediction data are aggregated;
[0023] Obtaining groundwater levels in non-water areas from resource prediction data sets Ds and soil moisture content Ts , the obtained prediction data are arranged in order according to the extension direction of the time axis, and the difference between two adjacent prediction nodes is obtained in turn as the water level increment and water volume increase , the obtained differences are summarized to construct a difference data set.
[0024] Furthermore, the water resources safety factor in the dispatching area is constructed by the difference data set. , as follows: Increase the water level by and water volume increase Perform linear normalization and map the corresponding data values to the interval In the following way: in, is the mean value of water level increment, For its i The value at the position; is the mean value of water volume increment, For its i The value at the position; , Is the number of predicted nodes, weight coefficient: , ,and .
[0025] Furthermore, water resource scheduling is used as the target word, and the data in the second hydrological data set is summarized to pre-construct a water resource scheduling knowledge graph; the predicted data in the resource prediction data set is obtained, and after setting the abnormal standard, feature recognition is performed on the predicted data to obtain the corresponding abnormal data features, and the trained matching model is used to match the corresponding scheduling goals and constraints for the water resource scheduling in the scheduling area from the water resource scheduling knowledge graph according to the correspondence between the scheduling plan and the abnormal data features; the third sample data is extracted from the second hydrological data set, and the initial model constructed by the linear programming algorithm is trained by the third sample data to obtain the water resource scheduling model, and the trained water resource scheduling model is used to output a number of scheduling plans under the pre-constructed scheduling goals and constraints.
[0026] Furthermore, taking the scheduling plan as input, the optimized water resources status prediction model is used to predict the feasibility of the scheduling plan and obtain the corresponding prediction data, and the scheduling coefficient is constructed based on the prediction data. , the scheduling coefficient The smallest dispatching scheme is used as the target scheme, and the next smallest one is used as the backup scheme. The target scheme is implemented in the dispatching area to dispatch water resources.
[0027] The method is as follows: After obtaining the execution scheduling plan from the forecast data and completing the water resource scheduling, the water supply in each sub-area Co Perform linear normalization and map the corresponding data values to the interval Then follow the steps below: , is the water demand in the sub-area, For the i The water supply in each sub-region, is its mean value, n is the number of sub-regions, , weight coefficient: , ,and .
[0028] The water conservancy dispatching system based on neural network technology includes:
[0029] The early warning unit divides the dispatching area into several sub-areas, detects the hydrological data in each sub-area, and constructs the condition evaluation coefficient of the sub-area in the current monitoring period based on the detection data. , so as to screen out unqualified areas in each sub-area. If the proportion of unqualified areas exceeds expectations, an early warning instruction is issued;
[0030] After receiving the warning instruction, the screening unit collects data in the dispatching area to construct a second hydrological data set, and , filter out model parameters from the second hydrological data set;
[0031] Optimize the unit, use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resource status prediction model needs to be optimized, and if necessary, optimize it;
[0032] The dispatching unit uses the optimized water resource status prediction model to predict the water resource status data in the dispatching area, and constructs the water resource safety factor in the dispatching area based on the predicted data. , if the water resources safety factor obtained is If the value is below the safety threshold, a dispatch instruction is issued;
[0033] After receiving the dispatch instruction, the analysis unit matches the corresponding dispatch target and constraint conditions for the water resources dispatch in the dispatch area through the pre-built water resources dispatch knowledge graph, and then outputs the dispatch plan through the trained water resources dispatch model. If there are several dispatch plans, the dispatch coefficient Filter it.
[0034] (III) Beneficial effects
[0035] The present invention provides a water conservancy dispatching method and system based on neural network technology, which has the following beneficial effects:
[0036] 1. Evaluation coefficient based on conditions Determine whether the current sub-area is a qualified area, obtain the proportion of unqualified areas, determine whether the water resources in the sub-area are evenly distributed, and take timely action to avoid ineffective water supply to some areas in the scheduling area, reduce water supply risks, and cause greater negative impact on crop growth.
[0037] 2. Based on the coefficient of variation By selecting model parameters from several potential independent variables and using them as input features, it is possible to reduce workload and interference between data when scheduling water resources in the scheduling area, thus ensuring the reliability of the predicted data. The size of the factor determines the corresponding degree of concern. On this basis, linear regression analysis can be used to determine the degree of influence on the growth status of crops in the scheduling area, which makes it easier to take targeted protection measures when ensuring the growth of crops.
[0038] 3. Train and obtain the water resource status prediction model, test the reliability of the water resource status prediction model, and construct the reliability coefficient based on the test data , with the reliability coefficient It is judged whether the water resource status prediction model needs to be optimized. If it needs to be optimized, it is optimized to improve the reliability of the water resource status prediction model. When it is used to predict the water resource status in the scheduling area, the predicted data will be more accurate and reliable. When it is used as the basis for water resource scheduling, it will be more in line with the actual situation in the scheduling area.
[0039] 4. Use the water resource status prediction model to predict the water resource status in the dispatching area, and construct the water resource safety factor based on the corresponding prediction data To obtain the water resources safety factor The water resources development status in the dispatching area is judged. If the water resources status in the dispatching area needs to be adjusted and improved, a dispatching instruction is issued to the outside. Based on the distribution data of water resources in the dispatching area and its neighboring areas, the water resources in the dispatching area are adjusted to improve the growth status of crops in the dispatching area.
[0040] 5. In order to match the corresponding scheduling objectives and constraints in the scheduling area, the trained water resource scheduling model is used to give the corresponding scheduling plan for the scheduling area. The water resources in the scheduling area are scheduled through the scheduling plan. When the water resource distribution status in the scheduling area does not meet expectations, the water resources in the scheduling area are scheduled. Therefore, with the assistance of knowledge graphs and linear programming algorithms, the growth status of crops in the scheduling area is guaranteed; the feasibility of each scheduling plan is verified, and the best part is selected to ensure the growth status of crops in the scheduling area. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 It is a schematic diagram of the flow of the water conservancy dispatching method based on the neural network technology of the present invention;
[0042] Figure 2 It is a schematic diagram of the structure of the water conservancy dispatching system based on neural network technology of the present invention. DETAILED DESCRIPTION
[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.
[0044] See also Figure 1 The present invention provides a water conservancy dispatching method based on neural network technology, comprising:
[0045] Step 1: Divide the dispatching area into several sub-areas, detect the hydrological data in each sub-area, and construct the condition evaluation coefficient of the sub-area in the current monitoring period based on the detection data. , so as to screen out unqualified areas in each sub-area. If the proportion of unqualified areas exceeds expectations, an early warning instruction is issued;
[0046] The step 1 includes the following contents:
[0047] Step 101: When water resource dispatching is required, a dispatching area is predefined, and the dispatching area is divided into a plurality of sub-areas using a trained classifier combined with terrain data in the dispatching area, and each sub-area is numbered respectively;
[0048] A data collection point is set in each sub-area, and after setting a monitoring period, hydrological data in the sub-area is collected at equal intervals at each data collection point, and the collected data is aggregated to construct a first hydrological data set;
[0049] Step 102: Construct the condition evaluation coefficient of the sub-region in the current monitoring period , including the groundwater level in the first hydrological data set Ds and soil moisture content Ts Perform linear normalization and map the corresponding data values to the interval Then follow the following formula: in, i is the serial number of the data collection node within the monitoring period, , is the number of data collection nodes, is the mean value of groundwater level, is the qualified reference value of groundwater level, is the mean soil moisture content, is the qualified reference value of soil moisture content; weight coefficient: , ,and ;The weight coefficient is determined by referring to the hierarchical analysis method; The groundwater level is i The value of the collection node, The soil moisture content is i The value of each collection node;
[0050] To obtain the conditional evaluation coefficient The corresponding sub-areas are marked, and the condition thresholds and quantity thresholds are pre-set based on historical data and management expectations for water resources scheduling. If the condition evaluation coefficient of the sub-area is If the area is lower than the condition area, it will be determined as an unqualified area; if the proportion of unqualified areas exceeds expectations, that is, exceeds the quantity threshold, an early warning instruction will be issued;
[0051] When using, combine the contents in steps 101 to 103:
[0052] After defining the water resources dispatching area, the dispatching area is divided into several sub-areas, and the condition evaluation coefficients in the sub-areas are constructed. , according to the conditional evaluation coefficient Determine whether the current sub-region is a qualified region, obtain the proportion of unqualified regions, and determine whether the water resources in the sub-region are evenly distributed. If the water resources are very unevenly distributed, or the water resources distribution conditions in some areas are poor, timely processing is required to avoid some areas in the scheduling area from not being effectively supplied with water, which will have a greater negative impact on the growth of crops.
[0053] Step 2: After receiving the warning instruction, collect data in the dispatching area to construct a second hydrological data set, and calculate the coefficient of variation of the collected data. , filter out model parameters from the second hydrological data set;
[0054] The step 2 includes the following contents:
[0055] Step 201: after receiving the early warning instruction, data is collected in the dispatching area, including collecting historical meteorological data, soil data, crop type and growth cycle data, historical irrigation data, etc. of the farmland, collecting water supply data, including historical and current water level and flow information of rivers, reservoirs, groundwater, etc., and marking each data with the collection time; pre-processing the collected data, including processing abnormal data, including noise, missing values, etc., and summarizing the pre-processed collected data as the second hydrological data set;
[0056] Step 202: After setting a number of equally spaced collection nodes, select a number of independent variable parameters and their values at each collection node in the second hydrological data set, such as rainfall, temperature, evaporation, soil moisture, etc., and calculate the coefficient of variation of each independent variable parameter in the following manner: perform linear normalization on the data values of the independent variable parameters and map the corresponding data values to the interval Inside: in, refers to the independent variable parameter, For independent variable parameters in the collection node i The value on is its mean value, is the weight, , ,and , the specific value is set by the user. is the number of independent variable parameters;
[0057] Sort the obtained coefficients of variation of each independent variable parameter by the coefficient of variation. The highest three independent variable parameters were used as model parameters;
[0058] When using, combine the contents in steps 201 to 202:
[0059] After receiving the early warning instruction, multi-dimensional data collection is carried out in the dispatching area, and the coefficient of variation is constructed after limiting several potential independent variables. , with coefficient of variation Selecting model parameters from several potential independent variables and using model parameters as input features can reduce workload and interference between data when scheduling water resources in the scheduling area, thus ensuring the reliability of prediction data;
[0060] Furthermore, after obtaining the coefficient of variation of the independent variable parameter Then, according to the coefficient of variation The size of the factor determines the corresponding degree of concern. On this basis, through linear regression analysis, the degree of influence of the crop growth status in the scheduling area can be determined, so that targeted protection measures can be taken when ensuring the growth of crops.
[0061] Step 3: Use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resource status prediction model needs to be optimized, and if necessary, optimize it;
[0062] The step three includes the following contents:
[0063] Step 301: Use B The neural network constructs an initial model, extracts part of the data from the second hydrological data set as the first sample data and the second sample data respectively, uses the first sample data to train the initial model, and uses the trained initial model as the water resources state prediction model in the scheduling area; uses the second sample data to perform reliability testing on the water resources state prediction model, and constructs a test data set from the acquired test data;
[0064] Step 302: Construct reliability coefficient from test data set , as follows: by the recall rate in the test data set and coefficient of determination After normalization, according to the following formula: Among them, the parameter meaning is: recall factor , , coefficient of determination factor , , is a constant correction coefficient, whose value falls within ;
[0065] It should be noted that a technician in this field collects multiple groups of sample data and sets a corresponding preset proportional coefficient for each group of sample data; substitutes the preset proportional coefficient and the collected sample data into the formula, any two formulas constitute a binary linear equation system, and the calculated coefficients are screened and averaged to obtain and The value of
[0066] The size of the coefficient is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the coefficient depends on the amount of sample data and the corresponding preset proportional coefficient initially set by technical personnel in this field for each set of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.
[0067] Based on historical data and expectations of model reliability, a reliability threshold is pre-established; if the obtained reliability coefficient If it is lower than the reliability threshold, it means that the currently constructed water resource status prediction model still needs further optimization, and an optimization instruction is sent to the outside. After receiving the optimization instruction, the water resource status prediction model is optimized and the optimized water resource status prediction model is obtained.
[0068] When using, combine the contents in steps 301 to 302:
[0069] Based on the sample data, the water resources status prediction model is trained and the reliability of the water resources status prediction model is tested, and the reliability coefficient is constructed based on the test data. , with the reliability coefficient It is judged whether the water resource status prediction model needs to be optimized. If it needs to be optimized, it is optimized to improve the reliability of the water resource status prediction model. When it is used to predict the water resource status in the scheduling area, the predicted data will be more accurate and reliable. When it is used as the basis for water resource scheduling, it will be more in line with the actual situation in the scheduling area.
[0070] Step 4: Use the optimized water resource status prediction model to predict the water resource status data in the dispatching area, and construct the water resource safety factor in the dispatching area based on the predicted data. , if the water resources safety factor obtained is If the value is below the safety threshold, a dispatch instruction is issued;
[0071] The step 4 includes the following contents:
[0072] Step 401: using the model parameters as input features, using the optimized water resource status prediction model to predict the water resource status data in the dispatching area, such as the water level in the water area, the water area, and the groundwater level in the non-water area; setting a number of equally spaced prediction nodes, obtaining prediction data at the prediction nodes, and constructing a water resource prediction data set after aggregating the prediction data;
[0073] Step 402: Obtain the groundwater level in the non-water area from the resource prediction data set Ds and soil moisture content Ts , the obtained prediction data are arranged in order according to the extension direction of the time axis, and the difference between two adjacent prediction nodes is obtained in turn as the water level increment and water volume increase , the obtained differences are summarized to construct a difference data set;
[0074] Step 403: Constructing the water resources safety factor in the dispatching area from the difference data set , as follows: Increase the water level by and water volume increase Perform linear normalization and map the corresponding data values to the interval In the following way: in, is the mean value of water level increment, For its i The value at the position; The mean value of water volume increment, For its i The value at the position; , Is the number of predicted nodes, weight coefficient: , ,and ,The specific value of the weight coefficient can be obtained by software simulation, or can be adjusted by the user according to actual usage;
[0075] Based on historical data and the management expectations of water resources in the dispatching area, a safety threshold is pre-established; if the water resource safety factor obtained is lower than the safety threshold, it means that the water resource changes in the dispatching area exceed expectations and there may be certain safety hazards. At this time, a dispatching instruction is issued to the outside;
[0076] When using, combine the contents in steps 401 to 403:
[0077] Based on the model parameters and water resource status prediction model, the prediction nodes are limited, the water resource status prediction model is used to predict the water resource status in the dispatching area, and the water resource safety factor is constructed based on the corresponding prediction data. To obtain the water resources safety factor The water resources development status in the dispatching area is judged. If the water resources status in the dispatching area needs to be adjusted and improved, a dispatching instruction is issued to the outside. Based on the distribution data of water resources in the dispatching area and its neighboring areas, the water resources in the dispatching area are adjusted to improve the growth status of crops in the dispatching area.
[0078] Step 5: After receiving the dispatch instruction, the pre-built water resource dispatch knowledge graph matches the corresponding dispatch target and constraint conditions for the water resource dispatch in the dispatch area, and then the trained water resource dispatch model outputs the dispatch plan. If there are several dispatch plans, the dispatch coefficients are used to calculate the dispatch plan. Screen it;
[0079] The step five includes the following contents:
[0080] Step 501: after obtaining the dispatch instruction, data is collected with water resource dispatch as the target word, and a graph data set is generated after the data in the second hydrological data set is aggregated, part of the data is extracted from the graph data set as sample data, and a semantic network model is obtained based on machine learning training; the semantic network model is used to select data from the graph data set for fusion processing, and the knowledge graph is constructed, and the constructed knowledge graph is used as the water resource dispatch knowledge graph;
[0081] Step 502: Obtain prediction data in the resource prediction data set, and after setting the abnormal standard, perform feature recognition on the prediction data to obtain the corresponding abnormal data features, and use the trained matching model to match the corresponding scheduling targets and constraints for water resource scheduling in the scheduling area from the water resource scheduling knowledge graph based on the correspondence between the scheduling plan and the abnormal data features;
[0082] Step 503: extracting third sample data from the second hydrological data set, and using the third sample data to train the initial model constructed by the linear programming algorithm to obtain a water resources scheduling model, and using the trained water resources scheduling model to output a number of scheduling plans under the pre-constructed scheduling objectives and constraints;
[0083] Step 504: Taking the scheduling plan as input, use the optimized water resource status prediction model to predict the feasibility of the scheduling plan and obtain the corresponding prediction data, and construct the scheduling coefficient based on the prediction data. , as follows:
[0084] After obtaining the execution scheduling plan from the forecast data and completing the water resource scheduling, the water supply in each sub-area Co Perform linear normalization and map the corresponding data values to the interval Then follow the steps below: , is the water demand in the sub-area, For the i The water supply in each sub-region, is its mean value, n is the number of sub-regions, , weight coefficient: , ,and ,The specific value of the weight coefficient can be obtained by software simulation, or can be adjusted by the user according to actual usage;
[0085] The scheduling coefficient The smallest dispatching scheme is used as the target scheme, and the next smallest one is used as the backup scheme. The target scheme is implemented in the dispatching area to dispatch water resources.
[0086] When used, in combination with steps 501 to 504,
[0087] Based on the knowledge graph of water resource scheduling, the corresponding scheduling targets and constraints are matched for the scheduling area according to the subsequent water resource status data in the scheduling area. The trained water resource scheduling model is used to give the corresponding scheduling plan for the scheduling area. The water resources in the scheduling area are scheduled through the scheduling plan. Therefore, when the water resource distribution status in the scheduling area does not meet expectations, the water resources in the scheduling area are scheduled. Therefore, with the assistance of the knowledge graph and linear programming algorithm, the growth status of crops in the scheduling area is guaranteed.
[0088] As a further content, when there are multiple scheduling plans obtained, the feasibility of each scheduling plan is verified with the assistance of the neural network, so as to screen out the best part before scheduling water resources, thereby ensuring the status of crops in the scheduling area.
[0089] See also Figure 2 The present invention provides a water conservancy dispatching system based on neural network technology, comprising:
[0090] The early warning unit divides the dispatching area into several sub-areas, detects the hydrological data in each sub-area, and constructs the condition evaluation coefficient of the sub-area in the current monitoring period based on the detection data. , so as to screen out unqualified areas in each sub-area. If the proportion of unqualified areas exceeds expectations, an early warning instruction is issued;
[0091] After receiving the warning instruction, the screening unit collects data in the dispatching area to construct a second hydrological data set, and , filter out model parameters from the second hydrological data set;
[0092] Optimize the unit, use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resource status prediction model needs to be optimized, and if necessary, optimize it;
[0093] The dispatching unit uses the optimized water resource status prediction model to predict the water resource status data in the dispatching area, and constructs the water resource safety factor in the dispatching area based on the predicted data. , if the water resources safety factor obtained is If the value is below the safety threshold, a dispatch instruction is issued;
[0094] After receiving the dispatch instruction, the analysis unit matches the corresponding dispatch target and constraint conditions for the water resources dispatch in the dispatch area through the pre-built water resources dispatch knowledge graph, and then outputs the dispatch plan through the trained water resources dispatch model. If there are several dispatch plans, the dispatch coefficient Filter it.
[0095] It should be noted that the AHP is a qualitative and quantitative analysis method that can decompose complex problems into multiple levels. By comparing the importance of factors at each level, it can help decision makers make decisions on complex problems and determine the final decision plan. In this process, the AHP can be used to determine the weight coefficients of these indicators. The steps of the AHP are as follows:
[0096] Clarify the problem: First, you need to clarify the decision problem and determine the decision goals and alternatives;
[0097] Establish a hierarchical model: According to the nature of the problem and the decision-making goal, the problem is decomposed into different levels, usually including the goal level, the criterion level and the solution level; the goal level is the overall goal of the decision-making problem, the criterion level is the criterion used to evaluate alternative solutions, and the solution level is the alternative solutions;
[0098] Construct a judgment matrix: Construct a judgment matrix by comparing the importance of elements in the same level relative to an element in the previous level. The elements in the judgment matrix represent the ratio of the relative importance of two elements;
[0099] Hierarchical single sorting: Based on the judgment matrix, the relative importance ranking weight of elements in the same level relative to an element in the previous level is calculated. This process is called hierarchical single sorting;
[0100] Consistency test: Check the consistency of the judgment matrix, that is, check whether the judgment matrix meets the consistency condition. If the consistency condition is met, the hierarchical single sorting result is considered reasonable;
[0101] Total ranking of levels: Calculate the composite weight of each layer element to the system goal, and perform total ranking to determine the total ranking weight of each element at the bottom layer of the hierarchical structure diagram;
[0102] Through the analytic hierarchy process, decision makers can break down complex decision-making problems into different levels and make decisions based on qualitative and quantitative analysis; this method can improve the accuracy and effectiveness of decision-making, and is particularly suitable for complex problems that are difficult to solve with quantitative methods.
[0103] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired means (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available media can be magnetic media (for example, floppy disks, hard disks, tapes), optical media (for example, DVD ), or semiconductor media. The semiconductor media may be a solid state drive.
[0104] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0105] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0106] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only some logical function divisions. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0107] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0108] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U Disk, mobile hard disk, read-only memory ( read - onlymemory , ROM ), Random Access Memory ( RandomAccessMemory , RAM ), magnetic disks, optical disks, and other media that can store program codes.
[0110] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
[0111] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.
Claims
1. A water conservancy dispatching method based on neural network technology, characterized in that: include, The dispatching area is divided into several sub-areas, the hydrological data in each sub-area is tested, and the condition evaluation coefficient of the sub-area in the current monitoring period is constructed based on the test data. , in order to screen out unqualified areas in each sub-area, if the proportion of unqualified areas exceeds expectations, an early warning instruction is issued; among them, the condition evaluation coefficient The method of obtaining the groundwater level in the first hydrological data set includes: Ds and soil moisture content Ts Perform linear normalization and map the corresponding data values to the interval Then follow the following formula: in, , is the number of data collection nodes, is the mean value of groundwater level, is the qualified reference value of groundwater level, is the mean soil moisture content, It is the qualified reference value of soil moisture content; The groundwater level is i The value of the collection node, The soil moisture content is i The value of each collection node; weight coefficient: , ,and ; After receiving the warning instruction, data is collected in the dispatching area to construct the second hydrological data set, and the coefficient of variation of the collected data is used to calculate the second hydrological data set. , filter out model parameters from the second hydrological data set; wherein, preprocess the data collected in the dispatching area, and aggregate the preprocessed data as the second hydrological data set; Select several independent variable parameters and their values at each acquisition node in the second hydrological data set, and calculate the coefficient of variation of each independent variable parameter as follows: : Perform linear normalization on the data values of the independent variable parameters and map the corresponding data values to the interval Inside: in, refers to the independent variable parameter, For independent variable parameters in the collection node i The value on is its mean value, is the weight, , ,and , is the number of independent variable parameters; Sort the obtained coefficients of variation of each independent variable parameter by the coefficient of variation. The highest three independent variable parameters were used as model parameters; use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resources status prediction model needs to be optimized, and if necessary, optimize it; among them, the reliability coefficient is constructed by the test data set , if the obtained reliability coefficient If the water resource status prediction model is lower than the reliability threshold, the water resource status prediction model is optimized and the optimized water resource status prediction model is obtained as follows: the recall rate in the test data set and coefficient of determination After normalization, according to the following formula: Among them, the parameter meaning is: recall factor , , coefficient of determination factor , , is a constant correction coefficient whose value falls within ; The optimized water resource status prediction model is used to predict the water resource status data in the dispatching area, and the water resource safety factor in the dispatching area is constructed based on the predicted data. , if the water resources safety factor obtained is If the water resource safety factor in the dispatching area is lower than the safety threshold, a dispatching instruction is issued; the water resource safety factor in the dispatching area is constructed by the difference data set. , as follows: Increase the water level by and water volume increase Perform linear normalization and map the corresponding data values to the interval In the following way: in, is the mean value of water level increment, For its i The value at the position; is the mean value of water volume increment, For its i The value at the position; , Is the number of predicted nodes, weight coefficient: , ,and ; After receiving the dispatch instruction, the pre-built water resource dispatch knowledge graph matches the corresponding dispatch target and constraint conditions for the water resource dispatch in the dispatch area, and then the trained water resource dispatch model outputs the dispatch plan. If there are several dispatch plans, the dispatch coefficients are used to calculate the dispatch plan. Screening is carried out, where the scheduling plan is used as input, and the feasibility of the scheduling plan is predicted using the optimized water resources status prediction model and the corresponding prediction data is obtained, and the scheduling coefficient is constructed based on the prediction data. , the scheduling coefficient The smallest dispatching scheme is used as the target scheme, and the next smallest one is used as the backup scheme. The target scheme is implemented in the dispatching area to dispatch water resources. The method is as follows: After obtaining the execution scheduling plan from the forecast data and completing the water resource scheduling, the water supply in each sub-area Co Perform linear normalization and map the corresponding data values to the interval Then follow the steps below: is the water demand in the sub-area, For the i The water supply in each sub-region, is its mean value, n is the number of sub-regions, , weight coefficient: , ,and ; Data is collected with water resources scheduling as the target word, and a graph data set is generated after summarizing the data in the second hydrological data set. Part of the data is extracted from the graph data set as sample data, and a semantic network model is obtained based on machine learning training. The semantic network model is used to select data from the graph data set for fusion processing to complete the construction of the knowledge graph, and the constructed knowledge graph is used as the water resources scheduling knowledge graph.
2. The water conservancy dispatching method based on neural network technology according to claim 1 is characterized in that: The dispatching area is defined in advance, and the dispatching area is divided into several sub-areas by using the trained classifier combined with the terrain data in the dispatching area; a data collection point is set in each sub-area, and after setting the monitoring period, the hydrological data in the sub-area is collected at equal intervals at each data collection point, and the collected data is aggregated to construct a first hydrological data set; the condition evaluation coefficient of the sub-area in the current monitoring period is constructed from the first hydrological data set. , to obtain the conditional evaluation coefficient The corresponding sub-regions are marked.
3. The water conservancy dispatching method based on neural network technology according to claim 1 is characterized in that: use B The neural network constructs an initial model, extracts part of the data from the second hydrological data set as the first sample data and the second sample data respectively, uses the first sample data to train the initial model, and uses the trained initial model as the water resources status prediction model in the scheduling area; uses the second sample data to test the reliability of the water resources status prediction model, and constructs a test data set based on the acquired test data.
4. The water conservancy dispatching method based on neural network technology according to claim 1 is characterized in that: The optimized water resource status prediction model is used to predict the water resource status data in the dispatching area with the model parameters as input features, and several equally spaced prediction nodes are set to obtain prediction data at the prediction nodes. After the prediction data are aggregated, a water resource prediction data set is constructed. Obtaining groundwater levels in non-water areas from resource prediction data sets Ds and soil moisture content Ts , the obtained prediction data are arranged in order according to the extension direction of the time axis, and the difference between two adjacent prediction nodes is obtained in turn as the water level increment and water volume increase , the obtained differences are summarized to construct a difference data set.
5. The water conservancy dispatching method based on neural network technology according to claim 1 is characterized in that: Obtain the forecast data in the resource forecast data set, and after setting the abnormal standard, perform feature recognition on the forecast data to obtain the corresponding abnormal data features; The third sample data is extracted from the second hydrological data set, and the initial model constructed by the linear programming algorithm is trained with the third sample data to obtain a water resource scheduling model. Under the pre-constructed scheduling objectives and constraints, the trained water resource scheduling model is used to output several scheduling plans.
6. A water conservancy dispatching system based on neural network technology, applying the water conservancy dispatching method according to any one of claims 1 to 5, characterized in that: include: The early warning unit divides the dispatching area into several sub-areas, detects the hydrological data in each sub-area, and constructs the condition evaluation coefficient of the sub-area in the current monitoring period based on the detection data. , so as to screen out unqualified areas in each sub-area. If the proportion of unqualified areas exceeds expectations, an early warning instruction is issued; After receiving the warning instruction, the screening unit collects data in the dispatching area to construct a second hydrological data set, and , filter out model parameters from the second hydrological data set; Optimize the unit, use B Neural network training obtains water resource status prediction model and constructs corresponding reliability coefficient after testing , according to the reliability coefficient Determine whether the water resources status prediction model needs to be optimized, and if so, optimize it; The dispatching unit uses the optimized water resource status prediction model to predict the water resource status data in the dispatching area, and constructs the water resource safety factor in the dispatching area based on the predicted data. , if the water resources safety factor obtained is If the value is below the safety threshold, a dispatch instruction is issued; After receiving the dispatch instruction, the analysis unit matches the corresponding dispatch target and constraint conditions for the water resources dispatch in the dispatch area through the pre-built water resources dispatch knowledge graph, and then outputs the dispatch plan through the trained water resources dispatch model. If there are several dispatch plans, the dispatch coefficient Filter it.
