Optimized scheduling method and system for water storage project

Abstract

The invention discloses a water storage project optimization scheduling method and system, and relates to the technical field of water conservancy project intelligent scheduling, and the method comprises the steps: respectively mapping a flood control risk weight and a power generation gain coefficient in a multi-dimensional constraint feature matrix into a positive diagonal term and a negative diagonal term of a QUBO Hamiltonian matrix, and mapping the hydraulic correlation strength into a non-diagonal coupling term, analyzing and distributing to generate quantum computable coded data; loading the quantum computable coding data to a quantum processor for annealing optimization, generating a non-dominated solution set through quantum parallel search, and outputting an optimal scheduling strategy vector; based on real-time attributes of water conservancy facility nodes of the dynamic hydraulic topological graph, a flood discharge gate opening value and a generator set output value of the optimal scheduling strategy vector are extracted, and a preliminary control instruction is generated; the water resource utilization efficiency and the safety and stability of hydropower station operation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent scheduling of water conservancy projects, and in particular to a water storage project optimization scheduling method and system. BACKGROUND

[0002] In modern water conservancy projects, the optimization scheduling technology of water storage projects mainly relies on a framework combining classical mathematical programming methods and artificial intelligence algorithms. Traditional methods are usually based on linear programming, nonlinear programming and mixed integer programming, combined with dynamic programming or genetic algorithms and other optimization tools, to realize the solution of scheduling strategies by constructing objective functions and constraint conditions. In recent years, with the improvement of computer computing power, scheduling models based on big data analysis have been gradually introduced, such as feature extraction of historical hydrological data through machine learning algorithms to assist in generating scheduling rules.

[0003] The prior art usually simplifies the multi-objective scheduling problem into a single objective optimization, or handles the multi-objective conflict through weighted summation, resulting in difficulty in balancing the dynamic balance of flood control risk and power generation gain in the scheduling strategy. For example, during the post-flood storage period, the joint scheduling of cascade reservoir groups needs to consider multi-dimensional constraints such as upstream inflow prediction, downstream ecological water demand and power generation demand, and the traditional optimization algorithm often cannot effectively decouple these nonlinear correlations when constructing the objective function, resulting in the generation of local optimal solutions. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a water storage project optimization scheduling method to solve the problem of local optimal solution caused by insufficient constraint decoupling of traditional methods.

[0006] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a water storage project optimization scheduling method, which comprises collecting a multi-source heterogeneous industrial data set, performing graph modeling processing, and constructing a dynamic water power topology graph; performing water power physical feature extraction and coupled constraint decoupling calculation on the dynamic water power topology graph to generate a multi-dimensional constraint feature matrix; mapping the flood control risk weight and the power generation gain coefficient in the multi-dimensional constraint feature matrix into positive and negative diagonal items of a QUBO Hamiltonian matrix, respectively, while mapping the water power correlation strength into a non-diagonal coupling item, and performing analytical distribution to generate quantum computable coding data; loading the quantum computable coding data to a quantum processor for annealing optimization, generating a non-dominated solution set through quantum parallel search, and outputting an optimal scheduling strategy vector; extracting the flood discharge gate opening value and the generator output value of the optimal scheduling strategy vector based on the real-time attributes of the water conservancy facility nodes of the dynamic water power topology graph to generate preliminary control instructions; and dynamically scaling the generator output value of the preliminary control instructions based on the power generation gain coefficient, and encapsulating to generate standard control instructions.

[0007] As a preferred scheme of the reservoir project optimal scheduling method, the dynamic hydraulic topology graph is constructed, and the specific steps are as follows: collecting original real-time data streams of flow meters of hydrological stations, PLCs of water plants and gate sensors to generate a multi-source heterogeneous industrial data set; performing timestamp synchronization and spatial positioning matching on the multi-source heterogeneous industrial data set to output a space-time aligned industrial data table; automatically mapping the space-time aligned industrial data table into an initial topology graph containing water conservancy facility nodes and water system connection edges based on reservoir hub topology rules; incrementally correcting real-time attributes of the water conservancy facility nodes and dynamic weights of the water system connection edges in the initial topology graph to generate the dynamic hydraulic topology graph.

[0008] As a preferred scheme of the reservoir project optimal scheduling method, the dynamic hydraulic topology graph is constructed, and the specific steps are as follows: collecting original real-time data streams of flow meters of hydrological stations, PLCs of water plants and gate sensors to generate a multi-source heterogeneous industrial data set; performing timestamp synchronization and spatial positioning matching on the multi-source heterogeneous industrial data set to output a space-time aligned industrial data table; automatically mapping the space-time aligned industrial data table into an initial topology graph containing water conservancy facility nodes and water system connection edges based on reservoir hub topology rules; incrementally correcting real-time attributes of the water conservancy facility nodes and dynamic weights of the water system connection edges in the initial topology graph to generate the dynamic hydraulic topology graph.

[0009] As a preferred scheme of the reservoir project optimal scheduling method, the dynamic hydraulic topology graph is constructed, and the specific steps are as follows: collecting original real-time data streams of flow meters of hydrological stations, PLCs of water plants and gate sensors to generate a multi-source heterogeneous industrial data set; performing timestamp synchronization and spatial positioning matching on the multi-source heterogeneous industrial data set to output a space-time aligned industrial data table; automatically mapping the space-time aligned industrial data table into an initial topology graph containing water conservancy facility nodes and water system connection edges based on reservoir hub topology rules; incrementally correcting real-time attributes of the water conservancy facility nodes and dynamic weights of the water system connection edges in the initial topology graph to generate the dynamic hydraulic topology graph.

[0010] As a preferred scheme of the reservoir project optimal scheduling method, the output optimal scheduling strategy vector is as follows: quantum computable coded data is loaded into a task execution queue of a quantum processor, and a quantum optimization task instance is output; the quantum optimization task instance is loaded into the quantum processor for annealing evolution, and quantum state evolution is performed in a potential energy field constructed by combining flood control targets and power generation targets, and an original solution set bit stream is output; the original solution set bit stream is subjected to Pareto front screening, and dominated solutions are filtered out and a non-dominated solution set is retained, and an optimal solution feature set is output; the optimal solution feature set is converted into a numerical vector of the floodgate opening degree and the generator unit output through linear weighting aggregation, and the optimal scheduling strategy vector is output.

[0011] As a preferred scheme of the reservoir project optimal scheduling method, the generation of the preliminary control instruction is as follows: based on the real-time attributes of the water conservancy facility nodes, a unique facility identifier is located, and the floodgate opening degree value and the generator unit output value in the optimal scheduling strategy vector are extracted, and a facility-control parameter key-value pair is generated; the real-time spatial coordinates are extracted from the water conservancy facility node attributes of the dynamic water power topology graph, the standardized node path identifier is generated through the OPC UA address mapping rule, and the OPC UA node address is output; the facility-control parameter key-value pair is bound with the OPC UA node address in the field, and the preliminary control instruction is generated.

[0012] As a preferred scheme of the reservoir project optimal scheduling method, the encapsulation of the standard control instruction is as follows: based on the power generation gain coefficient and the preset rated gain coefficient, a normalized proportion factor is calculated; the generator unit output value and the standardized node path identifier in the preliminary control instruction are analyzed, and an output sequence is output; the output sequence and the proportion factor are scaled with the generator unit output value, and a dynamic scaled output value sequence is generated; the dynamic scaled output value and the floodgate opening degree value are combined to form a control parameter set, and the control parameter set is encapsulated through the OPC UA binary encoder to generate the standard control instruction.

[0013] In the second aspect, the present invention provides a water storage project optimization scheduling system, including a graph modeling module for collecting multi-source heterogeneous industrial data sets, performing graph modeling processing, and constructing a dynamic hydraulic topology map; a feature decoupling module for extracting hydraulic physical features and performing coupling constraint decoupling calculations on the dynamic hydraulic topology map to generate a multidimensional constraint feature matrix; a Hamiltonian compilation module for mapping the flood control risk weights and power generation gain coefficients in the multidimensional constraint feature matrix into the positive diagonal terms and negative diagonal terms of the QUBO Hamiltonian matrix, and mapping the hydraulic correlation strength into non-diagonal coupling terms, and performing The system performs row parsing and allocation to generate quantum computable coded data; the optimization solution module is used to load the quantum computable coded data into the quantum processor for annealing optimization, generate a non-dominated solution set through quantum parallel search, and output the optimal scheduling strategy vector; the instruction conversion module is used to extract the flood gate opening value and generator output value of the optimal scheduling strategy vector based on the real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology graph, and generate preliminary control instructions; the dynamic optimization module is used to dynamically scale the generator output value of the preliminary control instruction based on the power generation gain coefficient, and encapsulate and generate standard control instructions.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the water storage project optimization scheduling method as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the method for optimizing scheduling of water storage projects as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: by vectorizing and graph-structure encapsulating the real-time attributes of water conservancy facility nodes and the dynamic weights of water system connection edges in a dynamic hydraulic topology graph, accurate modeling of the complex hydrodynamic characteristics of water storage projects is achieved; it can effectively improve the adaptability to changes in actual operating conditions and accurately capture the mutual influence between different facilities, thereby enhancing the scientific nature and accuracy of optimized scheduling decisions; it achieves the purpose of maximizing power generation benefits while ensuring flood control safety, and improves the efficiency of water resource utilization and the safety and stability of hydropower station operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Fig. 1 A flow chart of the water storage project optimal scheduling method.

[0019] Fig. 2 A schematic diagram of the water storage project optimal scheduling system.

[0020] Fig. 3 A flow chart of the dynamic water hydraulic topology construction.

[0021] Fig. 4 A flow chart of the control instruction generation and packaging. DETAILED DESCRIPTION

[0022] In order to make the above objectives, features and advantages of the present application more apparent, specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0023] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the concept of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0024] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0025] REFERENCE Figs. 1-4 For one embodiment of the present application, the embodiment provides a water storage project optimal scheduling method, comprising the following steps: S1, collecting multi-source heterogeneous industrial data sets, performing graph modeling processing, and constructing a dynamic water hydraulic topology.

[0026] S1.1, collecting the original real-time data stream of the hydrological station flow meter, water plant PLC and gate sensor, and generating a multi-source heterogeneous industrial data set.

[0027] Specifically, the flow data is periodically read through the sensor interface of the hydrological station flow meter, the sampling frequency is set to once per second, the data format is JSON, and the data contains a timestamp field, a flow value field, and a device identifier field; the running parameters are read through the Modbus TCP protocol interface of the water plant PLC, including current, voltage, power factor, running time, inlet and outlet pressure, and liquid level fields, the sampling frequency is once per minute, and the data format is CSV; the real-time state data is subscribed through the MQTT protocol interface of the gate sensor, including gate opening value, opening and closing state, and fault code fields, the sampling frequency is set to once per second, and the data format is binary protocol; the original real-time data stream is subjected to conflict detection, a deduplication algorithm based on hash values is used to eliminate duplicate records, and a linear interpolation method is used to complete the missing fields, and a multi-source heterogeneous industrial data set is output.

[0028] S1.2, time stamp synchronization and spatial positioning matching of the multi-source heterogeneous industrial data set are performed, and a space-time aligned industrial data table is output.

[0029] Specifically, the NTP protocol is used to calibrate to the UTC+8 time zone reference, and the calibration error is controlled within, for example, ±10 milliseconds; the KNN nearest neighbor matching algorithm is performed on the latitude and longitude coordinate fields in the hydrological station flow meter data, and the coordinate data of the water system topology graph is spatially associated with the water conservancy facility nodes, and the matching radius is, for example, 50 meters; the Euclidean distance calculation is performed on the installation location coordinate fields in the water plant PLC data, and the coordinate data of the water plant building plan is matched with the facility nodes, and the matching is, for example, 10 meters; the topological adjacency relationship matching algorithm is performed on the installation coordinate fields in the gate sensor data, and the spatial connection relationship of the water conservancy facility nodes is compared; the data records after time stamp calibration and the corresponding spatial matching results are spliced according to the fields, a standardized data table containing a time stamp, spatial coordinates, device type, and parameter value is generated, the total number of fields is, for example, 15, and the data format is Parquet; a deduplication algorithm based on hash values is used to eliminate redundant records, and a linear interpolation method is used to complete the missing fields, and a space-time aligned industrial data table is output.

[0030] S1.3, based on the reservoir hub topology rules, the space-time aligned industrial data table is automatically mapped to an initial topology graph containing water conservancy facility nodes and water system connection edges.

[0031] Specifically, the equipment type field, the spatial coordinate field and the equipment identifier field are extracted from the space-time alignment industrial data table; according to a preset reservoir hub topology rule, records with the equipment type of "hydrological station flow meter" are mapped as water system monitoring nodes, records with the equipment type of "gate sensor" are mapped as control and adjustment nodes, and records with the equipment type of "water plant PLC" are mapped as water intake nodes; the connection relationship between adjacent nodes is calculated by using a Delaunay triangulation algorithm according to the spatial coordinate field of the nodes, and a water system connection edge set is generated; for each water system connection edge, a direction attribute is given according to an upstream and downstream relationship, and is checked in combination with an actual connection relationship of the water conservancy facilities; finally, all the water conservancy facility nodes and the water system connection edges are stored in a graph structure form, and an initial topology graph is generated.

[0032] It should also be noted that the specific steps of the preset reservoir hub topology rule include: first, determining the node category corresponding to the equipment type, for example, identifying "hydrological station flow meter" as a monitoring node, "gate sensor" as a control and adjustment node, and "water plant PLC" as a water intake node; then, calculating the Euclidean distance between all nodes according to the spatial coordinate field, for example, if the distance between two nodes is less than or equal to, for example, 1000 meters, then further checking whether it meets the upstream and downstream relationship logic; then, using the Delaunay triangulation algorithm to process the nodes that meet the distance condition to generate a set of potential connection edges; for each potential connection edge, according to the physical connection principle of the equipment type and the water flow direction, determining whether it really represents the connection between the actual water conservancy facilities, and giving the correct flow direction attribute; finally, correcting the errors in the automatic mapping result through checking to ensure that all the connection edges accurately reflect the actual layout of the water conservancy facilities and the water flow direction.

[0033] S1.4, incrementally correcting the real-time attributes of the water conservancy facility nodes and the dynamic weights of the water system connection edges in the initial topology graph to generate a dynamic water conservancy topology graph.

[0034] Specifically, in the initial topology graph, the real-time attributes of the water conservancy facility nodes are incrementally corrected, including obtaining the latest monitoring data such as the flow value of the "hydrological station flow meter" and the opening state of the "gate sensor", and associating them with the corresponding nodes; based on a time interval (for example, updated every 10 minutes), the dynamic weight of the water system connection edge is calculated using the basic principles of fluid mechanics, such as adjusting the weight of the connection edge according to the flow difference and water flow speed between upstream and downstream nodes. By traversing the entire initial topology graph, comparing the current attributes of each node with the historical record, for example, if it is found that the attribute change exceeds the flow change by more than ±20%, the state of the node is immediately updated and the weight of all water system connection edges connected to it is re-evaluated. For newly added or deleted nodes and connection relationships, adjust them according to the latest monitoring data and physical connection principles. Integrate all updated information to generate a dynamic water conservancy topology graph containing the latest real-time attributes of water conservancy nodes and dynamic weights of water system connection edges.

[0035] S2, hydraulic physical feature extraction and coupling constraint decoupling calculation are performed on the dynamic water conservancy topology graph to generate a multi-dimensional constraint feature matrix.

[0036] S2.1, the real-time attribute vector of the water conservancy facility node in the dynamic water conservancy topology graph is encapsulated as a node feature tensor, and the dynamic weight of the water system connection edge is constructed into an adjacency matrix, and the graph structure is encapsulated to generate a graph structure data tensor.

[0037] Specifically, for the water conservancy facility nodes in the dynamic water conservancy topology graph, the latest real-time attributes are extracted, such as the flow value of the "hydrological station flow meter" and the opening state of the "gate sensor", and are vectorized and encapsulated into node feature tensors according to a predetermined order; for example, if a node has 3 different real-time attributes, a vector of length 3 is used to represent the features of the node. According to the dynamic weight of the water system connection edge, such as the flow difference and water flow speed between upstream and downstream nodes, an adjacency matrix is constructed, where the rows and columns represent the starting and ending nodes, respectively, and the matrix elements correspond to the weight values of the connection edges; for node pairs that do not have direct connections, the corresponding matrix elements are set to 0, for example. By traversing the entire dynamic water conservancy topology graph, all node feature tensors and adjacency matrix data are updated synchronously to ensure consistency and accuracy. The node feature tensor and the adjacency matrix are graph-structured encapsulated to generate a graph structure data tensor.

[0038] S2.2, graph convolution kernel calculation is performed on the graph structure data tensor to generate a node physical field feature vector.

[0039] Specifically, the first feature extraction is realized by performing linear transformation on the feature vector of each node using a graph convolution kernel weight matrix, for example, a 3x3 graph convolution kernel weight matrix; and an aggregation operation is performed on the transformed feature vector based on an adjacency matrix, that is, for each node, the feature information of the directly connected nodes is summarized, for example, the information of adjacent nodes is integrated by using the sum or average method, to generate an intermediate node feature vector. The intermediate node feature vector is subjected to secondary linear transformation by using a graph convolution kernel weight matrix again, and the normalized processing result obtained after combining the adjacency matrix is adjusted, to update the feature representation of each node, and generate a node physical field feature vector containing richer context information.

[0040] S2.3, fluid-structure coupling strength decomposition is performed on the node physical field feature vector, and a decoupling node feature matrix is output and fused and spliced with a dynamic hydraulic topology graph to generate a multi-dimensional constraint feature matrix.

[0041] Specifically, singular value decomposition (SVD) is used to decompose the physical field feature vector of each node into multiple independent components to identify and quantify the fluid-structure coupling strength contained therein. By comparing the similarity of each component with the known hydraulic interference noise characteristics, the components representing the hydraulic interference noise between facilities are determined and removed, thereby eliminating the influence of noise on the node physical field features. Next, the processed node physical field feature vector is recombined into a decoupling node feature matrix. The dynamic hydraulic topology graph is obtained and matched and fused and spliced with the decoupling node feature matrix according to the node ID, for example, matrix splicing operation is used to realize the merging of the two in the same node dimension, to generate a multi-dimensional constraint feature matrix containing the original node physical field features, information after fluid-structure coupling strength decomposition and hydraulic interference noise elimination.

[0042] S3, the flood control risk weight and the power generation gain coefficient in the multi-dimensional constraint feature matrix are respectively mapped to the positive diagonal items and the negative diagonal items of the QUBO Hamiltonian matrix, and the hydraulic correlation strength is mapped to the non-diagonal coupling items and is analytically distributed to generate quantum computable encoding data.

[0043] S3.1, the flood control risk weight and the power generation gain coefficient are extracted from the multi-dimensional constraint feature matrix, the flood control risk weight is converted to a positive target item by a positive and negative correction coefficient, the power generation gain coefficient is converted to a negative target item, and a quantitative target value sequence is output.

[0044] Specifically, according to the field definition of the multi-dimensional constraint feature matrix, the column field representing the flood risk weight is located, for example, the column field name is "flood risk weight", and the corresponding numerical value of each row is extracted to form the flood risk weight; similarly, the column field representing the power gain coefficient is located, for example, the column field name is "power gain coefficient", and the corresponding numerical value of each row is extracted to form the power gain coefficient. Then, for example, the positive correction coefficient value is 1.5 and the negative correction coefficient value is -0.7; for each flood risk weight, proportional mapping processing is performed according to the positive correction coefficient to generate corresponding positive target item data; at the same time, for each power gain coefficient, proportional mapping processing is performed according to the negative correction coefficient to generate corresponding negative target item data; finally, all positive target items and negative target items are arranged in order according to node number to form a quantitative target value sequence.

[0045] S3.2, filtering triangular region data from the hydraulic correlation strength of the multi-dimensional constraint feature matrix, and generating a facility interaction strength matrix after scaling adjustment by a proportional coefficient.

[0046] Specifically, a sub-matrix representing the hydraulic correlation strength in the multi-dimensional constraint feature matrix is identified, the sub-matrix is a symmetric matrix, and the elements reflect the hydraulic coupling strength between different water conservancy facilities; then the upper triangular part data of the sub-matrix is extracted, including all row and column index combinations above the main diagonal, for example, for the data item of the first row and the first column, only the element satisfying the condition is retained. The extracted upper triangular region data is scaled by a proportional coefficient, for example, 0.8, to adjust to a unified dimension range while maintaining the original relative relationship; the scaled upper triangular data is mapped back to the original matrix dimension and symmetrically filled to the lower triangular region, so that the elements at non-diagonal positions constitute a complete facility interaction strength matrix.

[0047] S3.3, assembling the quantitative target value sequence as the diagonal item and the facility interaction strength matrix as the non-diagonal coupling item into a QUBO Hamiltonian matrix, and outputting an optimization control matrix.

[0048] Specifically, let the quantitative target value sequence be , where represents the quantitative target value sequence, represents the linear term coefficient corresponding to the first water conservancy facility node, represents the linear term coefficient corresponding to the second water conservancy facility node, represents the linear term coefficient corresponding to the first water conservancy facility node, represents the linear term coefficient corresponding to the first water conservancy facility node; Let the facility interaction strength matrix be ​​,in, Indicates the Water conservancy facilities and The interaction intensity between water conservancy facilities, represents the matrix row index, represents the matrix column index, represents a symmetric matrix; Constructing the QUBO Hamiltonian matrix ,in, represents the first Water conservancy facilities and Water conservancy facilities shall be determined according to the following rules: ; That is: the quantized target value sequence is sequentially filled into the main diagonal position of the QUBO Hamiltonian matrix to form the diagonal terms; all the elements in the non-diagonal positions of the facility interaction intensity matrix are filled into the corresponding positions of the QUBO Hamiltonian matrix according to the corresponding row and column indices to form the non-diagonal coupling terms; finally, the complete QUBO Hamiltonian matrix is ​​generated and output as the optimization control matrix.

[0049] S3.4. The optimized control matrix is ​​parsed and allocated to the corresponding physical bits of the quantum processor through the quantum hardware address mapper, and compiled to generate quantum computable coded data.

[0050] Specifically, based on the optimized control matrix, a quantum hardware address mapper is used to read a pre-set physical bit mapping table, which shows the correspondence between logical variables and physical bits in the quantum processor; each element in the optimized control matrix is ​​positioned to the corresponding physical position in turn according to the physical bit mapping table, wherein the diagonal item values ​​are loaded into the spin action terms of the corresponding physical bit, and the non-diagonal coupling item values ​​are loaded into the interaction terms between the corresponding two physical bits; for example, when a matrix element corresponds to two logical variables and is mapped to two specific physical bits, the element value is written into the interaction term list as the coupling strength value between the two physical bits; finally, the spin action terms and interaction terms of all physical bits are summarized into quantum computable coded data that meets the input format requirements of the quantum processor.

[0051] It should also be noted that the establishment process of the pre-set physical bit mapping table: obtaining the physical topology description file of the quantum processor, which records all available physical bits and the connection relationship between them; determine the total number of logical variables required according to the number of variables of the optimization control matrix; then use the graph embedding algorithm to map the logical variables to the physical bits, and ensure that the coupling relationship between the logical variables matches the connection ability between the physical bits; for example, when there is a non-zero coupling term between two logical variables, it will be mapped to two physical bits with direct connection relationship; after completing the mapping, the physical bit mapping table is generated, which clearly shows the physical bit number corresponding to each logical variable and the actual position in the quantum processor.

[0052] S4, load the quantum computable encoding data to the quantum processor for annealing optimization, generate a non-dominated solution set through quantum parallel search, and output the optimal scheduling strategy vector.

[0053] S4.1, load the quantum computable encoding data to the task execution queue of the quantum processor, and output the quantum optimization task instance.

[0054] Specifically, the quantum computable encoding data includes spin interaction terms of physical bits and interaction terms between physical bits; submit a task request through the task scheduling interface of the quantum processor, where the task type is a quantum annealing optimization task, and the task parameters include annealing period, sampling times and temperature simulation value; then package the quantum computable encoding data into task load according to the input format required by the quantum processor, and assign a unique task identifier; write the packaged task load to the task execution queue buffer of the quantum processor; finally, after the task scheduler confirms that the resources are available, generate a quantum optimization task instance.

[0055] S4.2, load the quantum optimization task instance to the quantum processor for annealing evolution, and perform quantum state evolution in the potential field constructed by the flood control target and the power generation target, and output the original solution set bit stream.

[0056] Specifically, the submitted quantum optimization task instance is obtained through a task execution interface of the quantum processor, and the encapsulated spin interaction term and interaction term data are read; an annealing evolution process is configured according to the task parameters, including setting the annealing cycle duration to a preset time length, the sampling number to a specified value, and the simulation temperature parameter to a set value; after the annealing starts, the quantum processor constructs a potential energy field according to the spin interaction term and the interaction term, and the potential energy field is composed of the positive interaction term corresponding to the flood control target and the negative interaction term corresponding to the power generation target; then, a quantum state evolution process is performed in the potential energy field, the state change of each physical bit is controlled through a superconducting quantum interference device, and the initial superposition state gradually evolves to the vicinity of the ground state with the lowest energy; the quantum bit state is measured at the end of each annealing cycle, and a solution set bit vector composed of zeros and ones is output; the annealing and measurement operations are repeatedly performed for a specified number of times, and finally an original solution set bit stream containing multiple solution set bit vectors is output.

[0057] It should also be noted that the specific process of setting the annealing cycle duration, the sampling number and the simulation temperature parameter: according to the recommended value range provided by the hardware technical manual of the quantum processor, the annealing cycle duration is selected as, for example, twenty microseconds; based on the variable scale and the complexity of the solution space in the optimization problem, the sampling number is determined as, for example, one thousand times, to ensure that the output solution set bit stream has statistical representativeness; in combination with the energy scale distribution characteristics of the flood control target and the power generation target in the potential energy field, the simulation temperature parameter is set to, for example, one hundredth of a kelvin, to balance the quantum tunneling effect and the thermal disturbance effect; the parameter configuration interface of the quantum processor is used to write the task control register, and the physical bit control module reads and executes when the annealing evolution starts.

[0058] S4.3, Pareto front screening is performed on the original solution set bit stream, dominated solutions are filtered out, and a non-dominated solution set is retained, and an optimization solution feature set is output.

[0059] Specifically, each solution set bit vector in the original solution set bit stream is converted into a corresponding flood control target value sequence and a power generation target value sequence, where each solution set bit vector corresponds to a calculation result of a group of flood control target function values and power generation target function values; the target function values corresponding to each solution set bit vector are compared in turn according to the Pareto dominance rule, and if a solution set bit vector is not lower than another solution set bit vector in the flood control target function value while being higher than the other solution set bit vector in the power generation target function value, the solution set bit vector is determined to be a non-dominated solution; after traversing all solution set bit vectors in the original solution set bit stream, all solution set bit vectors determined to be non-dominated solutions are retained to form a non-dominated solution set; finally, the optimization variable values corresponding to each solution set bit vector in the non-dominated solution set are extracted according to the original bit structure to form an optimization solution feature set.

[0060] It should be noted that the data source of the Pareto dominance rule is based on the flood control objective function value and the power generation objective function value corresponding to each solution bit vector.

[0061] S4.4, the optimal solution feature set is converted into a numerical vector of floodgate opening degree and generator unit output through linear weighted aggregation, and an optimal scheduling strategy vector is output.

[0062] Specifically, the bit structure corresponding to each optimal solution is extracted from the optimal solution feature set, and the bit segments representing the floodgate control variables and the generator unit control variables are identified; for each control variable, a weighted sum operation is performed on the corresponding bit, for example, the high bit has a weight of 2 raised to the power of a certain number, and the low bit has a weight that decreases sequentially, and the bit is converted to a decimal number; then the converted numerical value is normalized to the floodgate opening degree range [0, 100%] and the generator unit output range [0, rated power] according to the mapping relationship, for example, through a linear interpolation method for scale transformation; finally, all the floodgate opening degree values and the generator unit output values are combined into a fixed-dimensional numerical vector, and the output is an optimal scheduling strategy vector.

[0063] S5, based on the real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology graph, the floodgate opening degree values and the generator unit output values in the optimal scheduling strategy vector are extracted to generate preliminary control instructions.

[0064] S5.1, based on the real-time attributes of the water conservancy facility nodes, the floodgate opening degree values and the generator unit output values in the optimal scheduling strategy vector are extracted through the unique facility identifier to generate a facility-control parameter key-value pair.

[0065] Specifically, based on the real-time attributes of the water conservancy facility nodes, each row of records contains a unique facility identifier, a facility type, and a current operating state field; nodes belonging to floodgates or generator units are filtered according to the facility type, and the corresponding control parameter index position in the optimal scheduling strategy vector is found according to the unique facility identifier; for example, if the unique facility identifier of the floodgate "Gate 001" corresponds to the first value in the optimal scheduling strategy vector, it is taken as the floodgate opening degree value; if the unique facility identifier of the generator unit "Unit 003" corresponds to the fifth value in the optimal scheduling strategy vector, it is taken as the generator unit output value; finally, the unique facility identifier and the corresponding floodgate opening degree value or generator unit output value are combined into a key-value pair to generate a facility-control parameter key-value pair.

[0066] S5.2, the real-time spatial coordinates are extracted from the water conservancy facility node attributes in the dynamic hydraulic topology graph, and the standardized node path identifier is generated through the OPC UA address mapping rule to output the OPC UA node address.

[0067] Specifically, according to the water conservancy facility node attribute in the dynamic hydraulic topology map, the unique facility identifier, the device type and the real-time spatial coordinate field of each node are contained; according to the preset OPC UA address mapping rule, the device type is converted into the object category in the OPC UA naming rule, for example, the “flood discharge gate” is converted into “DischargeGate”; then the unique facility identifier is taken as the instance name, and combined with the X, Y, Z values in the real-time spatial coordinates, the node path identifier is generated by splicing according to the format, and finally the node path identifier is taken as the OPC UA node address.

[0068] It should also be noted that the specific steps of the preset OPC UA address mapping rule are: establishing the correspondence between the device type and the OPC UA object category, and converting the device types such as “flood discharge gate” and “generator set” into standard object category names. Then use the unique facility identifier as the instance name to ensure the uniqueness of each node address. Extract the real-time spatial coordinates as position information, and generate the standardized node path identifier according to the structure format of “ / field / object category / instance name / spatial coordinates”. The generated node path identifier is taken as the OPC UA node address output, which is used for data addressing and communication in the industrial control system.

[0069] S5.3, binding the facility-control parameter key-value pair with the OPC UA node address to generate a preliminary control instruction.

[0070] Specifically, the unique facility identifier and the corresponding flood discharge gate opening value or generator set output value in each record in the facility-control parameter key-value pair set are read; at the same time, the node address matched with the unique facility identifier is found from the OPC UA node address data; through the unique facility identifier as the association field, the one-to-one correspondence between the control parameter and the OPC UA node address is realized; according to the data structure format, the flood discharge gate opening value or the generator set output value is taken as the target control value field, and the OPC UA node address is taken as the target address field, and a preliminary control instruction record is formed by combination; the above binding operation is performed on all unique facility identifiers in turn to generate complete preliminary control instructions.

[0071] S6, based on the power generation gain coefficient, the generator set output value of the preliminary control instruction is dynamically scaled, and the standard control instruction is generated.

[0072] S6.1, calculating the normalized scaling factor based on the power generation gain coefficient and the preset rated gain coefficient.

[0073] Specifically, the power generation gain coefficient corresponding to the current water conservancy facility node is extracted from the multidimensional constraint characteristic matrix, and the power generation gain coefficient is stored in a specified column field of the multidimensional constraint characteristic matrix; the preset rated gain coefficient is read from the control parameter configuration table, and the preset rated gain coefficient is a value pre-set and stored in a parameter database at a fixed address; the power generation gain coefficient is ratioed with the preset rated gain coefficient to obtain a proportional factor, and the proportional factor data item is normalized. The expression is: ; in, Indicates the The normalized scale factor corresponding to each water conservancy facility node, Represents the first The value of the power generation gain coefficient field is recorded, and T represents the corresponding preset rated gain coefficient value in the control parameter configuration table.

[0074] It should also be explained that the preset rated gain coefficient is a fixed value pre-configured and stored in the parameter database, which corresponds to the target gain reference value of the power generation equipment under standard operating conditions; the parameter database is deployed in the local server of the industrial control equipment, and the preset rated gain coefficient is stored in the configuration table under the specified path according to the preset data organization structure. The storage address is defined by the OPC UA address mapping rules to ensure that it can be read through the standardized data access interface.

[0075] S6.2. Parse the generator set output value and the standardized node path identifier in the preliminary control instruction and output the output sequence.

[0076] Specifically, each record in the preliminary control instruction is read to extract the generator set output value and the corresponding standardized node path identifier; the generator set output value is classified and sorted according to the unique facility identifier contained in the standardized node path identifier; then the generator set output value is arranged in sequence according to the sorting rule of the unique facility identifier to generate an output sequence organized in node order.

[0077] It should also be noted that the data source of the sorting rule of the facility identifier is based on the unique facility identifier field of the water conservancy facility node in the dynamic hydraulic topology graph, which is assigned according to the preset coding specification when the facility is deployed, and contains the facility type prefix and the sequence number; for example, "Gate 001" represents the first floodgate, and "Unit 003" represents the third generator unit. The sorting rule is based on the character structure of the facility identifier, which is first sorted by the facility type prefix, and the same type is sorted in ascending order according to the numerical value of the sequence number. The sorting rule is stored in the control parameter configuration table as part of the standardized data processing process, ensuring consistent sorting logic for each parsing operation.

[0078] S6.3, scaling operation of output sequence and scale factor and generator unit output value to generate dynamic scaling output value sequence.

[0079] Specifically, the parsed output sequence is read, which is composed of multiple generator unit output values, each corresponding to a unique facility identifier. At the same time, the calculated normalized scale factor is read, which is based on the ratio of the generator gain coefficient to the preset rated gain coefficient. According to the sorting rule of the facility identifier, each generator unit output value is retrieved in turn and adjusted in proportion to the normalized scale factor to generate the corresponding dynamic scaling output value. Finally, all dynamic scaling output values are combined in order to form a dynamic scaling output value sequence.

[0080] S6.4, merge the dynamic scaling output value and the floodgate opening value to form a control parameter set, and package the control parameter set through the OPC UA binary encoder to generate a standard control instruction.

[0081] Specifically, the generated dynamic scaling output value and the corresponding floodgate opening value are read in turn, and combined into a control parameter set containing power generation and flood control parameters according to the sorting rule of the facility identifier. The predefined OPC UA binary encoding rule is obtained, which specifies the field length, data type and arrangement order. Then, using the OPCUA binary encoder, each field in the control parameter set is executed byte-by-byte encoding operation according to the encoding rule to generate a binary format standard control instruction.

[0082] It should also be noted that the specific steps of the predefined OPC UA binary encoding rule: determine the field length and data type, specify the fixed or variable length of each field in the control parameter set according to the OPC UA standard specification, and specify the data type (such as integer, floating point, string, etc.); Then define the arrangement order, set the arrangement order of each field in the message body according to the actual needs of the control instruction, ensure that the parsing end can correctly identify and process; Use a specific byte alignment method, pad or truncate the bytes according to the OPC UA protocol requirements, ensure the consistency and compatibility of the data structure; Finally, apply the encoding algorithm, use the binary encoding method specified by OPC UA to convert the data of each field into binary format, and form a complete binary encoding rule.

[0083] The embodiment also provides a water storage project optimal scheduling system, comprising: a graph modeling module, used for collecting a multi-source heterogeneous industrial dataset, performing graph modeling processing, and constructing a dynamic water power topology graph; a feature decoupling module, used for performing water power physical feature extraction and coupling constraint decoupling calculation on the dynamic water power topology graph, and generating a multi-dimensional constraint feature matrix; a Hamiltonian compiling module, used for mapping a flood control risk weight and a power generation gain coefficient in the multi-dimensional constraint feature matrix into positive diagonal items and negative diagonal items of a QUBO Hamiltonian matrix respectively, mapping a water power correlation strength into a non-diagonal coupling item, and performing analytical distribution to generate quantum computable coding data; an optimal solution module, used for loading the quantum computable coding data to a quantum processor for annealing optimization, generating a non-dominated solution set through quantum parallel search, and outputting an optimal scheduling strategy vector; an instruction conversion module, used for extracting a floodgate opening degree value and a generator unit output value of the optimal scheduling strategy vector based on real-time attributes of water conservancy facility nodes of the dynamic water power topology graph, and generating a preliminary control instruction; and a dynamic optimization module, used for dynamically scaling the generator unit output value of the preliminary control instruction based on the power generation gain coefficient, and encapsulating and generating a standard control instruction.

[0084] The embodiment also provides a computer device suitable for the water storage project optimal scheduling method, comprising: a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the water storage project optimal scheduling method proposed in the above embodiment.

[0085] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is configured to perform wired or wireless communication with an external terminal. The wireless communication can be achieved by WIFI, an operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0086] The embodiment also provides a storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method for optimizing the operation of the water storage project according to the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic storage, a flash memory, a magnetic disk or an optical disk.

[0087] In summary, the application achieves accurate modeling of the complex water flow dynamics of the water storage project by vectorizing and graph structure encapsulation processing of the real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology graph and the dynamic weights of the water system connection edges. The adaptability to changes in actual operating conditions can be effectively improved, and the mutual influence between different facilities can be accurately captured, thereby enhancing the scientificity and accuracy of the optimization scheduling decision. The purpose of maximizing the power generation benefit while ensuring flood control safety is achieved, and the water resource utilization efficiency and the safety and stability of the operation of the hydropower station are improved.

[0088] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method for optimizing the scheduling of a water storage project, characterized by: include, Collect multi-source heterogeneous industrial data sets, perform graph modeling, and construct dynamic hydraulic topology maps; Extract hydraulic physical characteristics and perform coupling constraint decoupling calculation on the dynamic hydraulic topology map to generate a multi-dimensional constraint feature matrix; The flood control risk weight and power generation gain coefficient in the multidimensional constraint characteristic matrix are mapped into the positive diagonal terms and negative diagonal terms of the QUBO Hamiltonian matrix respectively. At the same time, the hydraulic correlation strength is mapped into the off-diagonal coupling term, and the analytical distribution is performed to generate quantum computable coded data. Loading quantum computable encoded data into the quantum processor for annealing optimization, generating a non-dominated solution set through quantum parallel search, and outputting the optimal scheduling strategy vector; Based on the real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology graph, the floodgate opening value and generator output value of the optimal scheduling strategy vector are extracted to generate preliminary control instructions; The generator set output value of the preliminary control instruction is dynamically scaled based on the power generation gain coefficient, and the standard control instruction is encapsulated and generated.

2. The method for optimizing the scheduling of a water storage project according to claim 1, wherein: The specific steps of constructing a dynamic hydraulic topology map are as follows: Collect raw real-time data streams from hydrological station flow meters, water plant PLCs, and gate sensors to generate multi-source heterogeneous industrial datasets; Perform timestamp synchronization and spatial positioning matching on multi-source heterogeneous industrial datasets, and output spatial-temporal aligned industrial data tables; Based on the reservoir hub topology rules, the space-time aligned industrial data table is automatically mapped into an initial topology graph containing water conservancy facility nodes and water system connection edges; The real-time attributes of the water conservancy facility nodes and the dynamic weights of the water system connection edges in the initial topology map are incrementally modified to generate a dynamic hydraulic topology map.

3. The method for optimizing the scheduling of a water storage project according to claim 1, wherein: The specific steps of generating a multi-dimensional constraint feature matrix are as follows: The real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology graph are vectorized and encapsulated into node feature tensors. The dynamic weights of the water system connection edges are simultaneously constructed into an adjacency matrix, and the graph structure is encapsulated to generate a graph structure data tensor. Perform graph convolution kernel calculation on the graph structure data tensor to generate node physical field feature vectors; The fluid-solid coupling intensity is decomposed on the node physical field eigenvectors, and the decoupled node characteristic matrix is ​​output and fused with the dynamic hydraulic topology map to generate a multi-dimensional constraint characteristic matrix.

4. The method for optimizing the scheduling of a water storage project according to claim 1, wherein: The specific steps of generating quantum computable coded data are as follows: Extract flood control risk weights and power generation gain coefficients from the multidimensional constraint feature matrix, convert flood control risk weights into positive target items and power generation gain coefficients into negative target items through positive and negative correction coefficients, and output a quantitative target value sequence; The triangular area data are screened from the hydraulic correlation intensity of the multi-dimensional constraint characteristic matrix, and the facility interaction intensity matrix is ​​generated after scaling and adjustment by the proportional coefficient; The quantitative target value sequence is used as the diagonal term, the facility interaction intensity matrix is ​​used as the off-diagonal coupling term, and they are assembled into a QUBO Hamiltonian matrix to output the optimized control matrix; The optimized control matrix is ​​parsed and assigned to the corresponding physical bits of the quantum processor through the quantum hardware address mapper, and compiled to generate quantum computable coded data.

5. The water storage project optimization scheduling method according to claim 1, characterized in that: The specific steps of outputting the optimal scheduling strategy vector are as follows: Load quantum computable encoded data into the task execution queue of the quantum processor and output quantum optimization task instances; The quantum optimization task instance is loaded into the quantum processor for annealing evolution, quantum state evolution is carried out in the potential energy field jointly constructed by the flood control goal and the power generation goal, and the original solution set bit stream is output; Perform Pareto front screening on the original solution set bit stream, filter out the dominated solutions and retain the non-dominated solution set, and output the optimized solution feature set; The optimized solution feature set is converted into a numerical vector of the flood discharge gate opening and the generator output through linear weighted aggregation, and the optimal scheduling strategy vector is output.

6. The method for optimizing the scheduling of a water storage project according to claim 1, wherein: The specific steps of generating preliminary control instructions are as follows: Based on the real-time attributes of water conservancy facility nodes, the unique facility identifier is used to locate the node, extract the floodgate opening value and generator output value in the optimal scheduling strategy vector, and generate facility-control parameter key-value pairs; Extract real-time spatial coordinates from the water conservancy facility node attributes in the dynamic hydraulic topology map, generate standardized node path identifiers through OPC UA address mapping rules, and output OPC UA node addresses; Field-bind the facility-control parameter key-value pairs with the OPC UA node addresses to generate preliminary control instructions.

7. The method for optimizing the scheduling of a water storage project according to claim 1, wherein: The encapsulation generates standard control instructions, and the specific steps are as follows: Calculating a normalized proportional factor based on the power generation gain coefficient and the preset rated gain coefficient; Parse the generator set output value and standardized node path identifier in the preliminary control instruction and output the output sequence; Perform scaling operations on the output sequence and the proportional factor with the generator set output value to generate a dynamic scaling output value sequence; The dynamic scaling output value is combined with the flood gate opening value to form a control parameter set, which is then encapsulated through the OPC UA binary encoder to generate standard control instructions.

8. A water storage project optimization scheduling system based on the water storage project optimization scheduling method according to any one of claims 1 to 7, characterized in that: include, The graph modeling module is used to collect multi-source heterogeneous industrial data sets, perform graph modeling processing, and construct a dynamic hydraulic topology map; Feature decoupling module, used to extract hydraulic physical features and perform coupling constraint decoupling calculations on the dynamic hydraulic topology map to generate a multi-dimensional constraint feature matrix; The Hamiltonian compilation module is used to map the flood control risk weights and power generation gain coefficients in the multidimensional constraint characteristic matrix into the positive and negative diagonal terms of the QUBO Hamiltonian matrix, respectively, and to map the hydraulic correlation strength into off-diagonal coupling terms, and perform analytical distribution to generate quantum computable encoded data; The optimization solution module is used to load quantum computable encoded data into the quantum processor for annealing optimization, generate non-dominated solution sets through quantum parallel search, and output the optimal scheduling strategy vector; The instruction conversion module is used to extract the floodgate opening value and generator output value of the optimal scheduling strategy vector based on the real-time attributes of the water conservancy facility nodes in the dynamic hydraulic topology map, and generate preliminary control instructions; The dynamic optimization module is used to dynamically scale the generator set output value of the preliminary control instruction based on the power generation gain coefficient, and encapsulate and generate standard control instructions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the water storage project optimization scheduling method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the water storage project optimization scheduling method according to any one of claims 1 to 7 are implemented.

Images