A Method and System for Predicting Hydrological Trends in Power Stations Based on Big Data Analysis

Abstract

This invention provides a method and system for predicting hydrological trends in power plants based on big data analysis, relating to the field of hydrological prediction technology. First, multi-source hydrological correlation data within the power plant's jurisdiction is acquired. Then, correlation and evolution analysis is performed on hydrological, meteorological, and equipment evolution units to generate a multi-dimensional hydrological evolution correlation map showing the dynamic interaction relationships between different evolution units. Next, key evolutionary paths influencing hydrological trends are extracted from the multi-dimensional hydrological evolution correlation map. Based on these key evolutionary paths, a dynamic hydrological trend inference model containing action transmission rules is constructed. Finally, the current hydrological, meteorological, and equipment evolution unit data are input into the dynamic hydrological trend inference model, and trend inference operations are performed through action transmission rules to generate a power plant hydrological trend prediction sequence, effectively improving the accuracy and reliability of power plant hydrological trend prediction.

Description

Technical Field

[0001] This invention relates to the field of hydrological prediction technology for power plants, and more specifically, to a method and system for predicting hydrological trends in power plants based on big data analysis. Background Technology

[0002] In the field of power plant operation and management, accurate prediction of hydrological trends is of paramount importance for ensuring the safe and stable operation of power plants, rationally allocating water resources, and preventing floods. Traditional power plant hydrological prediction methods are often limited to a single type of data, such as inferring water level changes based solely on hydrological records from monitoring points, or simply relying on regional meteorological records to predict the impact of rainfall on hydrological conditions.

[0003] However, hydrological conditions at power stations are complex and dynamic systems influenced by the interactions of multiple factors. While monitoring point hydrological records reflect the real-time state of local water areas, they cannot fully capture the combined effects of meteorological changes and power station equipment operation on hydrological conditions. Regional meteorological records, though providing information on the atmospheric environment, struggle to accurately assess the direct and indirect impacts of power station equipment operation on hydrological conditions. Power station equipment operation records document equipment operation and parameters, but lack analysis of the dynamic correlation between these parameters and hydrological and meteorological data. Existing forecasting methods, lacking in-depth analysis of the complex evolutionary relationships among multi-source data, struggle to accurately capture the inherent patterns and trends of hydrological changes, resulting in low accuracy and reliability of forecasts that fail to meet the needs of refined power station management and scientific decision-making. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, the present invention provides a method for predicting hydrological trends in power plants based on big data analysis, the method comprising:

[0005] Acquire multi-source hydrological correlation data within the power station's jurisdiction. The multi-source hydrological correlation data includes hydrological records from monitoring points, regional meteorological records, and power station equipment operation records. The hydrological records from monitoring points correspond to hydrological evolution units, the regional meteorological records correspond to meteorological evolution units, and the power station equipment operation records correspond to equipment evolution units.

[0006] A correlation evolution analysis is performed on the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit to generate a multi-dimensional hydrological evolution correlation map. The multi-dimensional hydrological evolution correlation map is used to present the dynamic interaction relationship between different evolution units.

[0007] From the multi-dimensional evolutionary correlation map of water conditions, key evolutionary paths that influence water condition trends are extracted. These key evolutionary paths are correlation sequences that continuously interact between different evolutionary units and have a dominant influence on water condition changes.

[0008] A dynamic model for predicting hydrological trends is constructed based on the key evolution paths, and the dynamic model for predicting hydrological trends includes the rules for the transmission of the effects of the key evolution paths.

[0009] The current hydrological evolution unit data, current meteorological evolution unit data, and current equipment evolution unit data are input into the hydrological trend dynamic extrapolation model. The trend extrapolation operation is executed through the action transmission rules to generate the hydrological trend prediction sequence of the power station.

[0010] In another aspect, the present invention also provides a power station hydrological trend prediction system based on big data analysis, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the machine-readable storage medium to implement the above-mentioned method.

[0011] Based on the above, by acquiring multi-source hydrological data, including hydrological records from monitoring points within the power station's jurisdiction, regional meteorological records, and power station equipment operation records, we conduct correlation and evolution analysis on hydrological evolution units, meteorological evolution units, and equipment evolution units. This generates a multi-dimensional hydrological evolution correlation map, which can intuitively present the dynamic interaction relationships between different evolution units and deeply reveal the complex mechanisms of hydrological changes. From this map, we extract key evolutionary paths influencing hydrological trends, accurately locating the dominant correlation sequences affecting hydrological changes. A dynamic hydrological trend extrapolation model built based on these key evolutionary paths includes the rules governing the transmission of the effects of these paths, enabling dynamic simulation and extrapolation of the hydrological change process. By inputting current data from various evolution units into the model to perform trend extrapolation operations, we generate a power station hydrological trend prediction sequence, effectively improving the accuracy and reliability of power station hydrological trend prediction. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the execution flow of the power station hydrological trend prediction method based on big data analysis provided in the embodiments of the present invention.

[0013] Figure 2 This is a schematic diagram of exemplary hardware and software components of a power station hydrological trend prediction system based on big data analysis provided in an embodiment of the present invention. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a power station hydrological trend prediction method based on big data analysis, provided in one embodiment of the present invention. The following is a detailed description of this power station hydrological trend prediction method based on big data analysis.

[0015] Step S110: Obtain multi-source hydrological correlation data within the power station's jurisdiction. The multi-source hydrological correlation data includes monitoring point hydrological records, regional meteorological records, and power station equipment operation records. The monitoring point hydrological records correspond to hydrological evolution units, the regional meteorological records correspond to meteorological evolution units, and the power station equipment operation records correspond to equipment evolution units.

[0016] In this embodiment, a hydropower station is used as an application scenario for explanation. Multiple hydrological monitoring stations are distributed along the river within the jurisdiction of the hydropower station. Each monitoring station is equipped with devices such as water level sensors and flow velocity sensors to continuously collect water level data, water flow velocity data, etc., at its location. This collection of raw data arranged in a time series constitutes the hydrological record of the monitoring point. The change in hydrological status over time reflected in this hydrological record corresponds to a hydrological evolution unit.

[0017] Meanwhile, meteorological observation stations are set up around the hydropower station to collect meteorological data such as rainfall, temperature, air pressure, relative humidity, wind direction, and wind speed in the area. These meteorological data form regional meteorological records in chronological order, and the meteorological conditions reflected by these records correspond to meteorological evolution units over time.

[0018] In addition, the main equipment in the hydropower station, such as turbines, generators, and gate control systems, are all equipped with sensors and data acquisition modules to record the operating parameters of the equipment in real time, such as the inlet pressure, outlet pressure, and speed of the turbines, the active power, reactive power, stator current, and stator voltage of the generators, and the opening degree and opening and closing time of the gates. These equipment operating parameters are organized over time to form the power station equipment operation record, which reflects the change of equipment operating status over time and corresponds to the equipment evolution unit.

[0019] During the data collection process, for data involving geographic location or other potentially private or sensitive information, data anonymization is employed. The geographic location information of specific monitoring stations is replaced with standardized regional codes, retaining only the relevance and validity of the data to prevent the leakage of sensitive information.

[0020] Step S120: Perform correlation evolution analysis on the hydrological evolution unit, the meteorological evolution unit and the equipment evolution unit to generate a multi-dimensional hydrological evolution correlation map, which is used to present the dynamic interaction relationship between different evolution units.

[0021] Step S121: Extract time series data of water evolution units from the water condition records of the monitoring points, extract time series data of meteorological evolution units from the regional meteorological records, and extract time series data of equipment evolution units from the power station equipment operation records.

[0022] In the aforementioned hydropower station scenario, water level and flow velocity data sequences are extracted from the hydrological records of each monitoring station in chronological order. These data sequences are then combined to form time-series data for hydrological evolution units. The time interval between data points is determined by the acquisition frequency; for example, if data is acquired every 15 minutes, the time interval between two adjacent data points in the time-series data is 15 minutes. Similarly, rainfall, temperature, and wind speed data sequences are extracted from regional meteorological records in chronological order and combined to form time-series data for meteorological evolution units, with time intervals consistent with those for hydrological evolution units. Finally, turbine speed, generator active power, and gate opening data sequences are extracted from the power station equipment operation records in chronological order and combined to form time-series data for equipment evolution units, with time intervals also consistent with the aforementioned two types of time-series data.

[0023] Step S122: Divide the time series data of the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit into multiple evolution period units according to the same time granularity.

[0024] All three types of time series data are divided using the same time granularity. For example, if one hour is chosen as a time granularity unit, then all time series data are divided into evolutionary period units with a duration of one hour. Specifically, starting from the start time of the time series data, every hour of continuous data is considered as an evolutionary period unit. If the remaining data is less than one hour, it is also processed as an independent evolutionary period unit, ensuring that all data are divided into the corresponding evolutionary period units.

[0025] Step S123: Capture features of hydrological evolution unit data within each evolution time period unit to generate hydrological time period features; capture features of meteorological evolution unit data within each evolution time period unit to generate meteorological time period features; capture features of equipment evolution unit data within each evolution time period unit to generate equipment time period features.

[0026] Step S1231: Extract basic hydrological parameters from the hydrological evolution unit data within the evolution period unit. The basic hydrological parameters include water level records and flow records of monitoring points.

[0027] For each evolutionary time period, water level and flow records from all monitoring stations are extracted from the hydrological evolution unit data within that unit. These data constitute the basic hydrological parameters for that evolutionary time period. For example, if an evolutionary time period corresponds to 1 hour of data, and a monitoring station has 4 water level records (one every 15 minutes) and 4 flow records during that time period, then these water level and flow records together constitute the basic hydrological parameters for that evolutionary time period.

[0028] Step S1232: Calculate the mean of the basic hydrological parameters within the evolution period unit, wherein the mean is the arithmetic mean of all basic hydrological parameter data within the evolution period unit.

[0029] For the water level record data within the aforementioned evolution period unit, sum the water level record data from all monitoring stations within that period, and then divide by the total number of water level record data to obtain the mean of the water level record data; using the same method, calculate the arithmetic mean of the flow record data to obtain the mean of the flow record data.

[0030] Step S1233: Calculate the range of variation of basic hydrological parameters within the evolution period unit, where the range of variation is the difference between the maximum and minimum values ​​of the basic hydrological parameters within the evolution period unit.

[0031] In the basic hydrological parameters of this evolution period unit, find the maximum and minimum values ​​in the water level record data, and subtract the minimum value from the maximum value to obtain the range of variation of the water level record data; similarly, find the maximum and minimum values ​​in the flow record data, calculate the difference between the two, and obtain the range of variation of the flow record data.

[0032] Step S1234: Analyze the direction of change of basic hydrological parameters within the evolution period unit, wherein the direction of change is the upward or downward trend of basic hydrological parameters within the evolution period unit from the start time to the end time.

[0033] The water level records within the evolutionary time unit are arranged chronologically, and the water level values ​​at the beginning and end are compared. If the water level at the end is greater than the water level at the beginning, the trend is considered upward; if the water level at the end is less than the water level at the beginning, the trend is considered downward; if they are equal, the trend is considered stable. The same method is used to analyze the trend of change in flow records.

[0034] Step S1235: Extract the extreme points of the basic hydrological parameters within the evolution period unit, where the extreme points are the maximum and minimum points of the basic hydrological parameters within the evolution period unit.

[0035] In the water level record data sequence of this evolution period unit, the data point with the largest value is identified as the maximum value point, and the data point with the smallest value is identified as the minimum value point; in the water flow record data sequence, the maximum value point and the minimum value point are also identified.

[0036] Step S1236: Calculate the number of times the extreme point appears within the evolution period unit, and count the number of times the maximum value point appears and the number of times the minimum value point appears.

[0037] The number of times the maximum value point appears in the water level record data within the evolution period unit is counted. For example, if a certain water level value appears twice in this period and both times it is the maximum value, then the maximum value point appears twice. Similarly, the number of times the minimum value point appears in the water level record data is counted. The method for counting the number of extreme points in the flow record data is the same.

[0038] Step S1237: Construct a set of feature terms based on the mean, range of variation, direction of variation, and frequency of extreme points of the basic hydrological parameters.

[0039] The mean, range, direction of change, number of occurrences of maximum and minimum values ​​of the water level record data obtained above, as well as the mean, range, direction of change, number of occurrences of maximum and minimum values ​​of the water flow record data, are combined to form a set of feature terms.

[0040] Step S1238: Quantize each feature in the feature set, converting the direction of change into a quantized value, with an upward trend corresponding to a positive value and a downward trend corresponding to a negative value.

[0041] For the characteristic of direction of change, an upward trend is converted to 1, a downward trend to -1, and a stable trend to 0; other characteristics such as mean, range of change, and number of extreme points are numerical data and their original values ​​are retained.

[0042] Step S1239: Arrange the quantized feature items sequentially according to the preset feature sorting rules to form an ordered feature value sequence.

[0043] The preset feature sorting rule is as follows: first, arrange the water level-related features, then arrange the water flow-related features; among the water level-related features, arrange them in the order of mean, range of variation, direction of variation, number of occurrences of the maximum value, and number of occurrences of the minimum value; the water flow-related features are arranged in the same order. According to this rule, the quantified features are arranged sequentially to form an ordered sequence of feature values.

[0044] Step S12310: Normalize the feature value sequence and determine the normalized feature value sequence as the hydrological time period feature.

[0045] The min-max normalization method is used to map each value in the feature value sequence to the interval [0, 1]. Specifically, for each value in the sequence, the minimum value of that feature item across all evolution time periods is subtracted, and then divided by the difference between the maximum and minimum values ​​of that feature item across all evolution time periods to obtain the normalized value. The feature value sequence after normalization is the hydrological time period feature for that evolution time period unit. The generation process for meteorological and equipment time period features is similar to that for hydrological time period features. For meteorological and equipment evolution unit data, corresponding basic parameters are extracted, and features such as mean and range of variation are calculated. After quantification and normalization, the meteorological and equipment time period features are obtained.

[0046] Step S124: Calculate the interaction strength between the hydrological time period characteristics and the meteorological time period characteristics within the same evolution time period unit; calculate the interaction strength between the hydrological time period characteristics and the equipment time period characteristics within the same evolution time period unit; calculate the interaction strength between the meteorological time period characteristics and the equipment time period characteristics within the same evolution time period unit.

[0047] Within the same evolutionary time unit, for hydrological time-period characteristics and meteorological time-period characteristics, they are treated as two vectors, and their influence intensity is measured by calculating the cosine similarity between them. The calculation process for cosine similarity is as follows: calculate the dot product of the two vectors, then calculate the magnitude of each vector, and divide the dot product by the product of the two magnitudes. The result is the influence intensity, which ranges from [-1, 1]. The closer the value is to 1, the greater the influence intensity. Using the same method, the cosine similarity between hydrological time-period characteristics and equipment time-period characteristics is calculated as their influence intensity, as is the cosine similarity between meteorological time-period characteristics and equipment time-period characteristics.

[0048] Step S125: Based on the changes in the intensity of interaction within units at different evolutionary time periods, identify the interaction change patterns between evolutionary units. The interaction change patterns include interaction enhancement patterns, interaction weakening patterns, and interaction stabilization patterns.

[0049] For any two evolutionary units (such as a hydraulic evolutionary unit and a meteorological evolutionary unit), their influence intensity data are extracted over multiple consecutive evolutionary time periods to form an influence intensity sequence. Trend analysis is performed on this influence intensity sequence, calculating the difference in influence intensity between adjacent evolutionary time periods. If multiple consecutive differences are positive and show an increasing trend, it is identified as an intensifying influence pattern; if multiple consecutive differences are negative and their absolute values ​​show an increasing trend, it is identified as a weakening influence pattern; if multiple consecutive differences have relatively small absolute values ​​and fluctuate within a stable range, it is identified as a stable influence pattern.

[0050] Step S126: Using the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit as graph nodes, the intensity of action within the unit during different evolution periods as the attribute of the association edge between nodes, and the action change mode as the dynamic identifier of the association edge.

[0051] The hydrological evolution unit, meteorological evolution unit, and equipment evolution unit are abstracted into three independent nodes. Within each evolution period unit, if there is an interaction strength between two evolution units (i.e., the interaction strength calculated in step S124), an association edge is established between the corresponding two nodes. The attribute value of the association edge is set to the interaction strength within the evolution period unit, and the dynamic identifier of the association edge is set to the interaction change pattern (such as the interaction enhancement pattern) identified within the evolution period unit.

[0052] Step S127: According to the time sequence of the evolution period units, the nodes, associated edge attributes and dynamic identifiers are mapped to the graph structure in sequence to form an initial evolutionary association graph containing the time dimension.

[0053] Starting with the first evolutionary period unit, the nodes within that period and the associated edges (including attributes and dynamic identifiers) are added to an initially empty graph structure. Then, the second evolutionary period unit is processed, and its nodes and associated edges (including attributes and dynamic identifiers) are added to the graph structure. A connection is established between the graph portion of the first evolutionary period unit and the graph portion of the first evolutionary period unit through the time axis. This process continues until all evolutionary period units are mapped into the graph structure, forming an initial evolutionary association graph that dynamically changes over time.

[0054] Step S128: Optimize the structure of the initial evolutionary association map by removing association edges with an effect strength lower than a preset effect threshold within the evolutionary time period unit, and retaining association edges with an effect strength higher than the preset effect threshold and their corresponding dynamic identifiers.

[0055] A preset threshold is established, determined based on the distribution of influence intensity in historical data. For example, the median of the influence intensity of all associated edges within all evolutionary time periods is taken as the preset threshold. For each evolutionary time period in the initial evolutionary correlation graph, the influence intensity of all associated edges within that time period is checked. If the influence intensity of an associated edge is lower than the preset threshold, the associated edge is deleted from the graph portion of that evolutionary time period; if the influence intensity is higher than or equal to the preset threshold, the associated edge and its corresponding dynamic identifier are retained.

[0056] Step S129: Integrate the optimized nodes, associated edge attributes, and dynamic identifiers to generate a multi-dimensional evolution association map of hydrological conditions.

[0057] After the structural optimization in step S128, the nodes, associated edge attributes, and dynamic identifiers retained in all evolutionary time period units are integrated in chronological order to form a compact, time-dimensional hydrological multi-dimensional evolutionary correlation map. This hydrological multi-dimensional evolutionary correlation map can clearly show the interaction relationships and changes between different evolutionary units at different time stages.

[0058] Step S130: Extract key evolutionary paths that influence the trend of water conditions from the multidimensional evolutionary correlation map of water conditions. The key evolutionary paths are correlation sequences that continuously interact between different evolutionary units and have a dominant influence on changes in water conditions.

[0059] Step S131: Analyze the node connection structure of the multidimensional evolution correlation map of the hydrological situation, and determine the direct and indirect correlation relationships between the evolutionary unit corresponding to each node and the evolutionary units corresponding to other nodes.

[0060] The structure of the multi-dimensional evolutionary correlation graph of hydrological conditions is analyzed by traversing all nodes and correlation edges in the graph. A direct correlation refers to a direct connection between two nodes via a single edge. For example, if a hydrological evolution unit node and a meteorological evolution unit node have a direct correlation edge, then they have a direct correlation. An indirect correlation refers to a connection between two nodes through other nodes. For example, if there is no direct correlation between a hydrological evolution unit node and an equipment evolution unit node, but there is a direct correlation between a hydrological evolution unit node and a meteorological evolution unit node, and vice versa, then the hydrological evolution unit node and the equipment evolution unit node have an indirect correlation.

[0061] Step S132: Extract the attribute information of all associated edges in the multidimensional evolution correlation map of the water situation. The attribute information includes the intensity of action and the mode of action change within different evolution period units.

[0062] Traverse each correlation edge in the multidimensional evolution correlation map of water conditions, record the influence intensity (i.e., the attribute value of the correlation edge) and the influence change pattern (i.e., the dynamic identifier of the correlation edge) of the correlation edge in each evolution period unit, and organize the above information according to the time order of the evolution period unit to form the attribute information set of each correlation edge.

[0063] Step S133: Perform time series analysis on the influence strength of each associated edge, and calculate the cumulative influence strength of the associated edge within the continuous evolution time period unit. The cumulative influence strength is the result of the superposition of influence strength within the continuous evolution time period unit.

[0064] For each associated edge, select the influence intensity data of N consecutive evolution time periods from its attribute information set (N is the number of consecutive time periods set according to actual needs), sum the influence intensity data, and the sum is the cumulative influence intensity value of the associated edge in these N consecutive evolution time periods.

[0065] Step S134: Sort different associated edges connected to the same node according to the cumulative value of the effect strength of the associated edges, and select the associated edges with the highest cumulative value of effect strength as the core associated edges.

[0066] For each node, collect all the associated edges connected to that node, sort the associated edges in descending order of their cumulative strength, select the top M associated edges (M is the preset number of filters), and determine these associated edges as the core associated edges.

[0067] Step S135: Construct an initial evolution path based on the core associative edge, wherein the initial evolution path is a path segment connecting two nodes through the core associative edge.

[0068] Each core associative edge connects two nodes. Combining these two nodes with the core associative edge connecting them forms a simple path segment, which is the initial evolutionary path. For example, if the core associative edge connects a hydrological evolution unit node and a meteorological evolution unit node, the initial evolutionary path is "hydrological evolution unit node - meteorological evolution unit node".

[0069] Step S136: Analyze the correlation between the initial evolutionary paths, and splice together the initial evolutionary paths that have common nodes and consistent change patterns to form a longer evolutionary path sequence.

[0070] Examine all initial evolution paths. If two initial evolution paths contain a common node (i.e., the endpoint of one path is the starting point of the other), and the influence change patterns of the edges connected to the common node in both paths are the same (e.g., both are influence enhancement patterns), then these two initial evolution paths are concatenated to form a longer evolution path sequence. For example, if the initial evolution paths "hydrological evolution unit node - meteorological evolution unit node" and "meteorological evolution unit node - equipment evolution unit node" have a common node "meteorological evolution unit node", and the influence change patterns of the edges connected to this common node in both paths are influence enhancement patterns, then they are concatenated into the evolution path sequence "hydrological evolution unit node - meteorological evolution unit node - equipment evolution unit node".

[0071] Step S137: Perform an action transmission continuity analysis on each evolution path sequence, and calculate the number of action transmission interruptions in the evolution path sequence within a continuous evolution time period. The number of action transmission interruptions is the number of times the action strength of the associated edge in the evolution path sequence is lower than a preset transmission threshold.

[0072] For example, step S1371: Determine the number of associated edges contained in the evolution path sequence and the corresponding evolution time period unit range.

[0073] For each evolutionary path sequence, all its nodes and associated edges are traversed, and the total number of associated edges is counted. Simultaneously, based on the evolutionary time units corresponding to the associated edges in the evolutionary path sequence, the start and end times of the evolutionary time units covered by the path sequence are determined, thus clarifying the range of the evolutionary time units. For example, if an evolutionary path sequence consists of 3 associated edges, corresponding to evolutionary time units 1-5, 2-6, and 3-7 respectively, then the path sequence contains 3 associated edges, and the corresponding evolutionary time unit range is 1-7.

[0074] Step S1372: Extract the influence strength data of each associated edge in the evolution path sequence within the corresponding evolution time period unit.

[0075] Based on the identifier of each associated edge in the evolution path sequence, the influence intensity data of the associated edge within its corresponding evolution time period is extracted from the associated edge attribute information of the multi-dimensional hydrological evolution association map. For example, if an associated edge corresponds to evolution time period 1-5, then the influence intensity data of the associated edge in evolution time period 1, 2, 3, 4, and 5 are extracted.

[0076] Step S1373: Set an action propagation threshold, which is determined based on the average action strength of all associated edges.

[0077] Calculate the average effect strength of all retained correlation edges (i.e., the correlation edges optimized in step S128) in the multidimensional evolution correlation map of hydrological conditions, and set 70% of this average value as the effect transmission threshold. For example, if the average effect strength of all correlation edges is 0.5, then the effect transmission threshold is set to 0.35.

[0078] Step S1374: Traverse each associated edge in the evolution path sequence and check whether its influence strength is lower than the influence propagation threshold.

[0079] For each associated edge in the evolutionary path sequence, check whether its influence strength data in each corresponding evolutionary time unit is less than the influence propagation threshold. For example, if the influence strength data of an associated edge in evolutionary time units 1-5 are 0.4, 0.3, 0.5, 0.32, and 0.45 respectively, and the influence propagation threshold is 0.35, then the influence strength of this associated edge in evolutionary time units 2 and 4 is less than the threshold.

[0080] Step S1375: If the influence strength of any associated edge is lower than the influence propagation threshold, record an influence propagation interruption.

[0081] During the inspection, if the influence strength of an associated edge is found to be lower than the influence propagation threshold within a certain evolutionary time unit, it is determined that an influence propagation interruption has occurred within that evolutionary time unit, and this is counted. For example, if the influence strength of the aforementioned associated edge is lower than the threshold for two evolutionary time units, then two influence propagation interruptions are recorded.

[0082] Step S1376: Count the total number of times the influence strength of all associated edges in the evolution path sequence is lower than the influence propagation threshold within the corresponding evolution time period unit.

[0083] The total number of interruptions in the action transmission of an evolutionary path sequence within its corresponding evolutionary time unit is obtained by summing the number of interruptions occurring for each associated edge in the evolutionary path sequence. For example, if an evolutionary path sequence contains 3 associated edges and experiences 2, 1, and 0 interruptions respectively, the total number of interruptions is 3.

[0084] Step S1377: Verify the validity of the total number of statistics, and exclude cases where the effect strength is abnormally low due to missing data collection. If the effect strength of any associated edge is low due to missing data collection, it will not be included in the number of effect transmission interruptions.

[0085] By examining the acquisition logs of the influence strength data of associated edges, it can be determined whether the low influence strength is caused by data loss due to data acquisition equipment failure, communication interruption, or other reasons. If there is a missing data record at the acquisition time corresponding to a certain influence strength data, then the low influence strength will not be counted in the influence transmission interruption count. For example, if the influence strength of an associated edge in evolutionary period unit 3 is 0.2 (below the threshold of 0.35), but there is a missing data acquisition record at that time, then this interruption will not be counted in the total number of interruptions.

[0086] Step S1378: Recount the number of action transmission interruptions after excluding abnormal situations, and use it as the final number of action transmission interruptions for this evolutionary path sequence.

[0087] The corrected number of interruptions due to missing data collection is obtained by subtracting the total number of interruptions from the total number of interruptions. For example, if the original total number of interruptions was 3, and 1 of them was due to missing data, then the final number of interruptions would be 2.

[0088] Step S1379: Record the number of interruptions in the final effect transmission of different evolutionary path sequences to form a statistical list of interruption counts.

[0089] The identifiers of each evolutionary path sequence and the number of interruptions in their final transmission are matched one-to-one and compiled into a list of interruption counts to facilitate the subsequent screening of candidate key evolutionary paths.

[0090] Step S13710: Match the interruption count list with the evolution path sequence one by one.

[0091] Ensure that each record in the interruption count statistics list has a clear association with the corresponding evolution path sequence in the data structure, such as binding them through the same sequence identifier, so as to facilitate the filtering of evolution path sequences based on the number of interruptions in step S138.

[0092] Step S138: Select evolution path sequences with fewer than a preset interruption threshold as candidate key evolution paths.

[0093] A preset interruption threshold is established, which is set according to the requirement for continuity of action transmission. The number of action transmission interruptions in the evolutionary path sequence is compared with the preset interruption threshold. If the number of action transmission interruptions is less than the preset interruption threshold, the evolutionary path sequence is identified as a candidate critical evolutionary path.

[0094] Step S139: Assess the hydrological impact of candidate critical evolution paths, and calculate the change range of hydrological evolution unit data corresponding to each candidate critical evolution path. The change range of hydrological evolution unit data is the range of change of hydrological records at monitoring points under the influence of candidate critical evolution paths.

[0095] For each candidate critical evolution path, extract the hydrological data of the monitoring points during the period of the path's action (i.e., the evolution period unit covered by the path sequence), find the maximum and minimum values ​​of the hydrological data of the monitoring points during this period, and calculate the difference between the maximum and minimum values. This difference is the data change range of the hydrological evolution unit corresponding to the candidate critical evolution path.

[0096] Step S1310: Select candidate critical evolution paths whose data change amplitude in hydrological evolution units exceeds a preset change threshold, and determine them as critical evolution paths that affect hydrological trends.

[0097] A preset change threshold is established, which is determined based on the normal fluctuation range in historical hydrological data. The change amplitude of hydrological evolution unit data of candidate critical evolution paths is compared with the preset change threshold. If the change amplitude is greater than the preset change threshold, the candidate critical evolution path is determined as a critical evolution path.

[0098] Step S140: Construct a dynamic simulation model of hydrological trends based on the key evolution paths. The dynamic simulation model of hydrological trends includes the rules for the transmission of the effects of the key evolution paths.

[0099] Step S141: Extract the node types and the variation law of the influence intensity of the associated edges between nodes in each key evolution path. The node types include hydrological evolution unit nodes, meteorological evolution unit nodes and equipment evolution unit nodes.

[0100] For each critical evolution path, the evolution unit type (hydrological evolution unit node, meteorological evolution unit node, or equipment evolution unit node) corresponding to each node in the path is identified, and the connection order between nodes is recorded. Simultaneously, the influence intensity data of each associated edge in the path is extracted within different evolutionary time periods. By performing trend fitting on the above data, a curve showing the change in influence intensity over time is obtained. The change pattern reflected by this curve is the change pattern of the influence intensity of the associated edges between nodes.

[0101] Step S142: Determine the data input type of different nodes in the simulation process according to the node type. The hydrological evolution unit node corresponds to the hydrological data input type, the meteorological evolution unit node corresponds to the meteorological data input type, and the equipment evolution unit node corresponds to the equipment operation data input type.

[0102] The data input type received by the hydrological evolution unit node during the simulation process is set as hydrological data, specifically including water level data, water flow velocity data, etc.; the data input type received by the meteorological evolution unit node is meteorological data, specifically including rainfall data, temperature data, wind speed data, etc.; the data input type received by the equipment evolution unit node is equipment operation data, specifically including turbine speed data, generator active power data, gate opening data, etc.

[0103] Step S143: Based on the variation law of the influence strength of the associated edges in the critical evolution path, establish an influence transfer function. The influence transfer function is used to describe the data transformation relationship when the data of the previous node is transferred to the next node through the associated edge.

[0104] Step S1431: Extract the influence intensity data of the associated edges in the key evolution path within different evolution time units to form an influence intensity time series.

[0105] For a specific associated edge in a key evolutionary path, extract the intensity data of all evolutionary time periods from the attribute information set of that associated edge, and arrange them in chronological order of the evolutionary time periods to form an intensity time series.

[0106] Step S1432: Perform trend fitting on the time series of action intensity to obtain a fitting curve of action intensity changing with time, wherein the fitting curve is used to reflect the changing trend of action intensity.

[0107] The moving average method is used to process the time series of action intensity to smooth out noise in the data. Then, based on the smoothed data points, a continuous fitting curve is obtained using a polynomial fitting method (such as quadratic polynomial fitting). This fitting curve can intuitively reflect the overall trend of action intensity over time.

[0108] Step S1433: Analyze the variation characteristics of the fitted curve, determine the inflection point of the change in action intensity and the action intensity value at the inflection point, wherein the inflection point is the evolution period unit in which the trend of action intensity change changes.

[0109] Differentiate the fitted curve to obtain the derivative curve. The zero point of the derivative curve corresponds to the inflection point of the fitted curve. Based on the derivative curve, find all points where the derivative is zero. The evolutionary time period corresponding to these points on the time axis is the inflection point position. Substitute the inflection point position into the fitted curve to calculate the intensity value at the inflection point.

[0110] Step S1434: Divide the time series of action intensity into multiple action stages according to the inflection point position, and keep the trend of action intensity change consistent within each action stage.

[0111] Using the inflection point as the dividing point, the time series of the effect intensity is divided into several continuous subsequences. Each subsequence corresponds to an effect stage. Within each effect stage, the trend of the effect intensity (such as rising, falling, or remaining stable) remains consistent.

[0112] Step S1435: Perform correlation analysis on the intensity data of each action phase and the previous node data and the next node data in the corresponding evolution period unit. First, convert the previous node data and the next node data into dimensionless standardized data to obtain standardized previous node data and standardized next node data. Then, determine the correspondence between the standardized previous node data and the product of the action intensity and the standardized next node data.

[0113] For each stage of action, the preceding node data, following node data, and action intensity data for all evolutionary time units within that stage are extracted. The z-score normalization method is used to convert the preceding node data into standardized preceding node data (specifically, subtracting the mean of the preceding node data across all evolutionary time units from each data point, and then dividing by the standard deviation of the preceding node data across all evolutionary time units); the same method is used to convert the following node data into standardized following node data. Then, the product of the standardized preceding node data and the action intensity within each evolutionary time unit is calculated, and the correlation between this product and the standardized following node data is analyzed. For example, by observing the distribution pattern of the scatter plot, it can be determined whether there is a linear or non-linear correspondence between the two.

[0114] Step S1436: Construct a local transfer function for each action stage based on the correspondence. The input of the local transfer function is the data of the previous node and the action intensity, and the output is the predicted value of the data of the next node.

[0115] Based on the correspondence determined in step S1435, if there is a linear relationship between the product of the standardized previous node data and the action intensity and the standardized next node data, a linear local transfer function is constructed, in the form of the predicted value of the standardized next node data = a * (standardized previous node data * action intensity) + b, where a and b are parameters to be determined. If there is a nonlinear relationship, a nonlinear local transfer function is constructed, such as a quadratic function. Then, using historical data within the action phase (standardized previous node data, action intensity, and standardized next node data), the parameters in the function are estimated using methods such as the least squares method to determine the specific form of the local transfer function.

[0116] Step S1437: Determine the transition conditions between different action stages, wherein the transition conditions are set based on the change range of the previous node data and the change range of the action intensity.

[0117] Calculate the change amplitude of the preceding node data (i.e., the absolute value of the difference between the preceding node data of the first evolutionary period unit of the next stage and the preceding node data of the last evolutionary period unit of the previous stage) and the change amplitude of the action intensity (i.e., the absolute value of the difference between the action intensity of the first evolutionary period unit of the next stage and the action intensity of the last evolutionary period unit of the previous stage) at the boundary between two adjacent action stages. Compare these two change amplitudes with preset preceding node data transition thresholds and action intensity transition thresholds, respectively. If the preceding node data change amplitude is less than or equal to the preceding node data transition threshold and the action intensity change amplitude is less than or equal to the action intensity transition threshold, the transition condition is met, allowing the switch from the current action stage to the next action stage.

[0118] Step S1438: Connect multiple local transfer functions in series according to the time sequence of their action phases, and establish an action transfer function in combination with transition conditions. The action transfer function can output the corresponding predicted value of the next node data based on the input previous node data and the action intensity of the unit in the current evolution period.

[0119] The local transfer functions for each action stage are arranged sequentially according to their order of action. During the simulation, based on the action stage of the current evolutionary period unit, the corresponding local transfer function is selected. The data from the previous node and the action intensity of the current evolutionary period unit are input to calculate the predicted value of the next node. When the evolutionary period unit enters the next action stage, it is checked whether the transition condition is met. If it is, the local transfer function corresponding to the next action stage is switched.

[0120] Step S1439: Perform error verification on the action transfer function. Input the previous node data and action intensity within the historical evolution period cell into the action transfer function, and compare the difference between the predicted value of the next node data and the actual next node data.

[0121] Multiple evolutionary time periods are selected from historical data. The previous node data and the intensity of the action within the above time period are input into the constructed action transfer function to obtain the predicted value of the next node data. Then, the predicted value is compared with the actual next node data within the evolutionary time period, and the absolute error and relative error between the two are calculated.

[0122] Step S14310: Adjust the parameters of the action transfer function according to the error verification results to reduce the difference between the predicted value and the actual data until the difference is less than the preset error threshold.

[0123] If the absolute error or relative error in the error verification result is greater than the preset error threshold, then return to step S1436, adjust the parameters in the local transfer function (such as a and b in the linear function), recalculate the predicted value and perform error verification, repeat this process until the difference between the predicted value and the actual data is less than the preset error threshold.

[0124] Step S144: Assign a projection weight to each critical evolution path. The projection weight is determined based on the change range of the hydrological evolution unit data corresponding to the critical evolution path. The greater the change range of the hydrological evolution unit data, the higher the projection weight.

[0125] The variation range of hydrological evolution unit data for all key evolution paths is normalized so that the normalized values ​​are within the range of [0, 1]. The normalized values ​​are used as the inference weights for the corresponding key evolution paths. That is, the greater the variation range of hydrological evolution unit data, the greater the normalized value and the higher the inference weight.

[0126] Step S145: Construct a path fusion module for the dynamic inference model of hydrological trends. The path fusion module is used to receive the inference results of multiple key evolution paths and to fuse the inference results according to the inference weights.

[0127] Step S1451: Determine the input data type of the path fusion module. The input data type is the intermediate hydrological prediction result output by each key evolution path. The intermediate hydrological prediction result includes the hydrological prediction value of the corresponding evolution period unit.

[0128] The data received by the path fusion module is the intermediate hydrological prediction result output by each key evolution path at each simulation time step (corresponding to the evolution period unit). The intermediate hydrological prediction result is specifically represented by a numerical value, namely the hydrological prediction value of that evolution period unit.

[0129] Step S1452: Assign weight coefficients to the intermediate results of hydrological prediction for each critical evolution path, wherein the weight coefficients are consistent with the extrapolation weights of the critical evolution path.

[0130] The inference weight of each key evolution path is directly used as the weight coefficient of the intermediate results of hydrological prediction for that path.

[0131] Step S1453: Establish a weighted fusion algorithm, which is used to multiply the intermediate results of hydrological prediction for each key evolution path with the corresponding weight coefficient to obtain a weighted hydrological prediction value.

[0132] The specific operation of the weighted fusion algorithm is as follows: for each simulation time step, traverse all key evolution paths, multiply the intermediate hydrological prediction result (hydrological prediction value) of each path by its corresponding weight coefficient, and obtain the weighted hydrological prediction value of that path at that time step.

[0133] Step S1454: Sum all weighted hydrological prediction values ​​to obtain the initial fused prediction value.

[0134] Within each simulation time step, the weighted hydrological predictions of all key evolution paths are summed, and the sum is the initial fused prediction value for that time step.

[0135] Step S1455: Construct a fusion result correction mechanism for the path fusion module. The fusion result correction mechanism is used to adjust the initial fusion prediction value based on the deviation between the historical fusion results and the actual hydrological data.

[0136] Collect initial fusion prediction values ​​and corresponding actual hydrological data for multiple past simulation time steps, calculate the deviation for each time step (initial fusion prediction value minus actual hydrological data), and construct a fusion result correction mechanism based on these deviations.

[0137] Step S1456: Calculate the average deviation between the historical fusion results and the actual hydrological data, wherein the average deviation is the arithmetic mean of the deviations within multiple evolution period units.

[0138] Select the deviations of the K most recent evolutionary time periods (K is the preset historical data window size), sum the above deviations and divide by K to obtain the average deviation.

[0139] Step S1457: Determine the correction coefficient based on the average deviation. When the average deviation is positive, the correction coefficient is less than 1. When the average deviation is negative, the correction coefficient is greater than 1.

[0140] The correction factor is calculated as follows: Correction factor = 1 - (average deviation / average historical hydrological data). When the average deviation is positive, the correction factor is less than 1; when the average deviation is negative, the correction factor is greater than 1.

[0141] Step S1458: Multiply the initial fusion prediction value by the correction coefficient to obtain the corrected fusion prediction value.

[0142] Within each simulation time step, the initial fusion prediction value for that time step is multiplied by the correction coefficient determined according to step S1457 to obtain the corrected fusion prediction value.

[0143] Step S1459: Establish the output rules of the path fusion module. The output rules are used to arrange the corrected fusion prediction values ​​in the order of the simulation time step of the hydrological trend dynamic simulation model to form a phased fusion prediction sequence.

[0144] After completing the fusion processing for all extrapolation time steps, the path fusion module arranges the corrected fusion prediction values ​​for each time step in the order of the extrapolation time steps to form a phased fusion prediction sequence.

[0145] Step S14510: Integrate the input data type, weight coefficient, weight fusion algorithm, fusion result correction mechanism, correction coefficient calculation method and output rules to form a path fusion module.

[0146] The input data types, weight coefficient allocation methods, specific operations of the weight fusion algorithm, the composition of the fusion result correction mechanism, the calculation method of the correction coefficient, and the output rules determined in the above steps are integrated to form a complete path fusion module, which serves as part of the dynamic inference model of water conditions.

[0147] Step S146: Set the simulation time step of the hydrological trend dynamic simulation model, and the simulation time step is consistent with the time granularity of the evolution period unit in the hydrological multidimensional evolution correlation map.

[0148] The time step of the dynamic simulation model of hydrological trend is set to be the same as the time granularity of the evolution period unit in the multidimensional evolution correlation map of hydrological conditions. For example, if the time granularity of the evolution period unit is 1 hour, then the simulation time step is also set to 1 hour.

[0149] Step S147: Establish a simulation state update mechanism for the dynamic simulation model of water conditions. The simulation state update mechanism is used to adjust the parameters of the action transfer function in the next time step based on the simulation results of the current time step.

[0150] The simulation state update mechanism specifically includes: after each simulation time step, comparing the simulation result (fusion prediction value) of that time step with the actual hydrological data (if any), calculating the deviation between the two, and adjusting the parameters of the action transfer function in the next time step (such as the local transfer function parameters a and b in step S1436) according to the magnitude and direction of the deviation, so that the output of the action transfer function is closer to the actual situation.

[0151] Step S148: Integrate the data input type, action transfer function, inference weight, path fusion module, inference time step and inference status update mechanism to form the core framework of the dynamic inference model of hydrological trend.

[0152] The core framework of the dynamic hydrological trend simulation model is formed by integrating the data input type determined in step S142, the action transfer function established in step S143, the simulation weights allocated in step S144, the path fusion module constructed in step S145, the simulation time step set in step S146, and the simulation status update mechanism established in step S147.

[0153] Step S149: Logically verify the core framework of the dynamic inference model of water conditions. Through logical comparison, verify the logical consistency between the action transfer functions of different key evolution paths. If there is a logical conflict, adjust the parameters of the action transfer function. Through rule verification, verify the matching of the weight fusion rule of the path fusion module with the inference logic. If they do not match, modify the weight fusion rule.

[0154] Logical comparison operation: Check whether the output trends of the transfer functions of different key evolutionary paths are consistent when processing the same type of input data. If there are opposite trends, a logical conflict is determined, and the process returns to step S143 to adjust the parameters of the relevant transfer functions. Rule verification operation: Check whether the weight fusion rules of the path fusion module (such as the allocation of weight coefficients, weighted summation, etc.) are consistent with the model's inference logic (such as inference based on the importance of key evolutionary paths). If they are inconsistent (such as the weight coefficients being unrelated to the inferred weights), the weight fusion rules are modified.

[0155] Step S1410: After completing the logic verification, determine the final structure of the dynamic simulation model of the water situation trend.

[0156] After completing logical verification and resolving all logical conflicts and rule mismatches, the core framework of the dynamic simulation model of water situation trends is determined, and the final structure of the model is formed.

[0157] Step S150: Input the current hydrological evolution unit data, the current meteorological evolution unit data, and the current equipment evolution unit data into the hydrological trend dynamic extrapolation model, and perform trend extrapolation operations through the action transmission rules to generate a hydrological trend prediction sequence for the power station.

[0158] Step S151: Collect the current hydrological data of the monitoring points within the power station's jurisdiction and convert it into current hydrological evolution unit data; collect the current regional meteorological data and convert it into current meteorological evolution unit data; collect the current power station equipment operation data and convert it into current equipment evolution unit data.

[0159] The hydropower station's monitoring system collects real-time hydrological data (water level, flow velocity, etc.) from monitoring points. Following the same data format and processing method as the time-series data extracted in step S121, this data is converted into current hydrological evolution unit data. Similarly, real-time collected regional meteorological data is converted into current meteorological evolution unit data, and real-time collected power station equipment operation data is converted into current equipment evolution unit data.

[0160] Step S152: Determine the start and end times of the dynamic simulation model of hydrological trends, and calculate the total number of simulation steps based on the simulation time step of the dynamic simulation model of hydrological trends.

[0161] Set the start time of the simulation to the current moment and the end time of the simulation to T hours after the current moment (T is the preset simulation duration). Calculate the total number of simulation steps by dividing T by the simulation time step (if T is 24 hours and the simulation time step is 1 hour, then the total number of simulation steps is 24).

[0162] Step S153: Input the current hydrological evolution unit data, current meteorological evolution unit data, and current equipment evolution unit data into the corresponding nodes of the hydrological trend dynamic simulation model as the initial input data for the simulation start time.

[0163] Input the current hydrological evolution unit data into the corresponding hydrological evolution unit node in the hydrological trend dynamic simulation model, input the current meteorological evolution unit data into the meteorological evolution unit node, and input the current equipment evolution unit data into the equipment evolution unit node. These data serve as the initial input data for the simulation start time (first step simulation).

[0164] Step S154: Call the action transfer function in the dynamic simulation model of hydrological trend to perform action transfer calculation on the initial input data, obtain the intermediate result of hydrological prediction for the first simulation time step, and input the intermediate result of hydrological prediction for the first simulation time step into the path fusion module of the dynamic simulation model of hydrological trend, perform fusion processing according to the path fusion rules, and obtain the fused prediction value of hydrological trend for the first simulation time step.

[0165] For example, step S1541: extract the fluctuation characteristics of the current hydrological evolution unit data, the change gradient of the current meteorological evolution unit data, and the operational stability index of the current equipment evolution unit data from the initial input data. The fluctuation characteristics are the extreme value differences of the current hydrological evolution unit data within a preset time window, the change gradient is the rate of change of the values ​​of the current meteorological evolution unit data at adjacent acquisition times, and the operational stability index is the deviation ratio between the current equipment evolution unit data and the rated operating parameters.

[0166] A preset time window is defined (e.g., the past 5 data acquisition times including the current time). Water level data within this time window is extracted from the current hydrological evolution unit data. The maximum and minimum values ​​are identified, and the difference between the maximum and minimum values ​​is calculated as a fluctuation characteristic. For the current meteorological evolution unit data, such as rainfall data, the difference in rainfall between two adjacent acquisition times is calculated and then divided by the time interval to obtain the change gradient. For the current equipment evolution unit data, such as generator active power data, the difference between the active power data and the rated active power is calculated and then divided by the rated active power to obtain the deviation ratio, which is used as an operational stability indicator.

[0167] Step S1542: Standardize the fluctuation characteristics, change gradient, and operational stability index respectively to obtain standardized fluctuation characteristics, change gradient, and operational stability index; based on the correlation edge attributes between nodes in the key evolution path, establish a coupling relationship matrix between the current hydrological evolution unit, the current meteorological evolution unit, and the current equipment evolution unit, where the element values ​​of the coupling relationship matrix correspond to the degree of mutual influence between data from different evolution units.

[0168] Using the same min-max normalization method as in step S12310, the fluctuation characteristics, change gradient, and operational stability index are standardized to the [0, 1] interval. Based on the influence strength of the correlation edges between nodes in the key evolution path (i.e., correlation edge attributes), a 3x3 coupling relationship matrix is ​​constructed. The rows and columns of the matrix correspond to the hydrological evolution unit, meteorological evolution unit, and equipment evolution unit, respectively. The matrix element values ​​are the influence strength of the correlation edges between the corresponding two evolution units, representing the degree of mutual influence between them.

[0169] Step S1543: Substitute the standardized fluctuation characteristics, change gradient and operational stability index into the coupling relationship matrix, and generate the coupling effect value of multi-unit data through matrix operation. The coupling effect value is used to quantify the joint impact of different evolution unit data on hydrological prediction.

[0170] The standardized fluctuation characteristics, change gradients, and operational stability indicators are combined into a column vector. This column vector is then multiplied by the coupling relationship matrix to obtain a new column vector. The elements in this column vector are the coupling effect values ​​of the multi-cell data.

[0171] Step S1544: Call the action transfer function in the dynamic simulation model of water situation trend, take the coupling action value as the function input, and calculate the first stage transfer result by combining the change law of the action intensity of the associated edge corresponding to the action transfer function.

[0172] Based on the node connection order in the critical evolution path, determine the action transfer function that needs to be called now, input the coupling action value into the action transfer function, and combine the change law of the action strength of the associated edge corresponding to the function (such as the change curve of action strength over time) to obtain the first stage transfer result through function calculation.

[0173] Step S1545: Extract historical action transmission records that are similar to the current coupling action value during the historical simulation process, and obtain the transmission deviation coefficient in the historical records. The transmission deviation coefficient is the ratio of the difference between the historical transmission result and the actual hydrological data to the actual hydrological data.

[0174] In the historical simulation records of the model, find records where the coupling effect value is close to the current coupling effect value (e.g., the difference is within a preset range), and extract the transmission deviation coefficient from these records. The transmission deviation coefficient is calculated as (historical transmission result - actual hydrological data) / actual hydrological data.

[0175] Step S1546: Multiply the first-stage transmission result with the transmission deviation coefficient to obtain the corrected transmission result.

[0176] Multiply the first-stage transmission result by (1 + the average value of the transmission deviation coefficients) to obtain the corrected transmission result, where the average value of the transmission deviation coefficients is the arithmetic mean of all transmission deviation coefficients extracted in step S1545.

[0177] Step S1547: Based on the node type order of the key evolution path, determine whether the corrected transmission result needs to be converted across units. If the current node type is a meteorological evolution unit node and the next node type is a hydrological evolution unit node, then the preset meteorological-hydrological conversion coefficient is used to adjust the value of the corrected transmission result.

[0178] Check the types of the current node and the next node in the critical evolution path. If the current node is a meteorological evolution unit node and the next node is a hydrological evolution unit node, then call the preset meteorological-hydrological conversion coefficient from the model parameters, multiply the corrected transmission result by the conversion coefficient, and obtain the adjusted transmission result. If no cross-unit conversion is required, the adjusted transmission result is the corrected transmission result.

[0179] Step S1548: Use the adjusted transmission result as the intermediate result of the hydrological prediction for the first simulation time step, and record the coupling effect value, transmission deviation coefficient and conversion coefficient corresponding to the intermediate result of the hydrological prediction.

[0180] The adjusted transmission result obtained in step S1547 is determined as the intermediate result of the hydrological prediction for the first simulation time step, and the corresponding coupling value, transmission deviation coefficient, and conversion coefficient are stored in the model's cache for subsequent analysis and verification.

[0181] For example, step S1549: Extract the predicted values, key evolution path identifiers, and associated coupling values ​​corresponding to each key evolution path from the intermediate results of hydrological prediction for the first simulation time step.

[0182] For the first extrapolation time step, each critical evolution path outputs an intermediate hydrological prediction result (predicted value), and at the same time, extracts the unique identifier of each path (critical evolution path identifier) ​​and the coupling effect value of that path at that time step.

[0183] Step S15410: Based on the key evolution path identifier, call the historical prediction deviation sequence of the key evolution path stored in the path fusion module. The historical prediction deviation sequence is the set of differences between the predicted values ​​and the actual hydrological data of the key evolution path in the past extrapolation time steps.

[0184] Based on the key evolution path identifier, the historical prediction deviation sequence of the path is read from the storage unit of the path fusion module. The historical prediction deviation sequence contains the difference between the predicted value and the actual hydrological data at each time step in the past multiple simulation time steps.

[0185] Step S15411: Calculate the volatility variance of the historical prediction deviation sequence, which is used to reflect the stability of the prediction results of the key evolution path.

[0186] The variance of all deviation values ​​in the historical prediction deviation sequence is calculated. The larger the fluctuation variance, the worse the stability of the prediction results of the key evolution path; the smaller the fluctuation variance, the better the stability.

[0187] Step S15412: Determine the dynamic contribution of the key evolution path at the current time step based on the ratio of fluctuation variance to coupling effect value. The smaller the fluctuation variance and the larger the coupling effect value, the higher the dynamic contribution.

[0188] Calculate the ratio of the fluctuation variance to the coupling effect value of the key evolution path, normalize the ratio (mapped to the [0, 1] interval), and set the dynamic contribution to 1 minus the normalized ratio. That is, the smaller the fluctuation variance, the larger the coupling effect value, the smaller the ratio, the smaller the normalized ratio, and the higher the dynamic contribution.

[0189] Step S15413: Weight the intermediate hydrological prediction results of each key evolution path with the corresponding dynamic contribution to obtain the weighted prediction value of that path.

[0190] Multiply the intermediate hydrological forecast results (predicted values) of each key evolution path by its corresponding dynamic contribution to obtain the weighted forecast value of that path at the current simulation time step.

[0191] Step S15414: Collect the weighted predictions of all key evolutionary paths and calculate the mean of these weighted predictions as the initial fusion baseline.

[0192] The weighted predicted values ​​of all critical evolution paths are summed, and then divided by the number of critical evolution paths. The average value obtained is the initial fusion baseline value.

[0193] Step S15415: Extract the trend terms from the variation law of the influence intensity of the associated edge corresponding to each key evolution path, and determine the consistency of the direction of each trend term. If the trend terms exceeding the preset proportion are all upward trends, then it is determined that the current water situation prediction is generally upward.

[0194] For each critical evolution path, extract trend terms (such as the slope of the fitted curve) from the variation pattern of the influence intensity of its associated edges (extracted in step S141). If the slope is positive, the trend term is upward; if the slope is negative, it is downward; if the slope is zero, it is a stable trend. Count the proportion of upward trend terms to the total number of trend terms. If this proportion exceeds a preset proportion (e.g., 60%), the current hydrological forecast is determined to be generally upward; if the proportion of downward trend terms exceeds a preset proportion, the overall trend is determined to be downward; otherwise, it is determined to be a stable trend.

[0195] Step S15416: Adjust the initial fusion benchmark value according to the overall trend. If the trend is upward, add a trend correction amount to the initial fusion benchmark value. The trend correction amount is the product of the average rate of change of all upward trend terms and the initial fusion benchmark value.

[0196] If the overall trend is upward, calculate the average rate of change of all upward trend terms (the average slope of the trend terms), and the trend correction amount = average rate of change * initial fusion baseline value. The adjusted initial fusion baseline value = initial fusion baseline value + trend correction amount. If the trend is downward, the trend correction amount = average rate of change (which is negative in this case) * initial fusion baseline value. The adjusted initial fusion baseline value = initial fusion baseline value + trend correction amount. If the trend is stable, the adjusted initial fusion baseline value = initial fusion baseline value.

[0197] Step S15417: Cross-validate the adjusted initial fusion baseline value with the weighted prediction values ​​of each key evolution path, and calculate the absolute value of the deviation between the adjusted initial fusion baseline value and each weighted prediction value.

[0198] For each key evolution path, the weighted prediction value is calculated, and the absolute value of the difference between this value and the adjusted initial fusion baseline value is obtained.

[0199] Step S15418: Count the number of weighted prediction values ​​whose absolute value of the statistical deviation is less than the preset deviation threshold. If the proportion of the number of weighted prediction values ​​to the total number of weighted prediction values ​​exceeds the preset proportion threshold, then the adjusted initial fusion benchmark value is used as the hydrological fusion prediction value for the first simulation time step.

[0200] Preset a deviation threshold and a proportion threshold (e.g., the deviation threshold is 0.1 and the proportion threshold is 70%). Count the number of weighted prediction values ​​whose absolute deviation value is less than the deviation threshold, and calculate the proportion of this number to the total number of weighted prediction values. If the proportion exceeds the proportion threshold, the adjusted initial fusion benchmark value is the hydrological fusion prediction value for the first simulation time step.

[0201] Step S15419: If the proportion of the weighted predicted value does not reach the preset proportion threshold, the dynamic contribution is recalculated. In this calculation process, the weight of recent data from the historical prediction deviation sequence is added, and the weighting operation, initial fusion benchmark value calculation and tendency adjustment steps are repeated until the hydrological fusion predicted value that meets the cross-validation requirements is obtained.

[0202] If the proportion does not reach the preset proportion threshold, when calculating the variance of the historical prediction deviation sequence, the recent deviation value is given a higher weight (such as using the weighted moving average method to calculate the variance), the dynamic contribution is recalculated, and then steps S15413 to S15418 are repeated until the hydrological fusion prediction value that meets the cross-validation requirements (i.e. the proportion exceeds the proportion threshold) is obtained.

[0203] Step S15420: Record the dynamic contribution set, trend correction amount, and cross-validation results corresponding to the final hydrological fusion prediction value.

[0204] The hydrological fusion prediction value of the first simulation time step and the dynamic contribution set of all key evolution paths, the tendency correction amount, and the cross-validation results (such as the proportion of the absolute value of the deviation being less than the deviation threshold) are recorded in the simulation result storage unit of the model.

[0205] Step S155: Based on the dynamic simulation model of hydrological trend and the simulation state update mechanism, adjust the parameters of the action transfer function in the second simulation time step based on the hydrological fusion prediction value of the first simulation time step.

[0206] Step S1551: Extract the initial parameters of the action transfer function in the first simulation time step, wherein the initial parameters include the coefficients and constant terms in the action transfer function.

[0207] Read the initial parameters of the action transfer function used in the first simulation time step from the dynamic simulation model of water situation trend, such as the coefficient a and constant term b of the local transfer function in step S1436.

[0208] Step S1552: Calculate the deviation between the hydrological fusion prediction value and the current actual hydrological evolution unit data for the first simulation time step. The deviation value is the result of subtracting the current actual hydrological evolution unit data from the hydrological fusion prediction value.

[0209] Obtain the hydrological fusion prediction value and the current actual hydrological evolution unit data for the first simulation time step (if the end time of the first simulation time step has passed, the actual data at that time can be obtained; if it has not passed, the actual data at the current time can be used as a reference), and calculate the deviation value = hydrological fusion prediction value - current actual hydrological evolution unit data.

[0210] Step S1553: Analyze the positive and negative attributes and absolute value of the deviation value. When the deviation value is positive and the absolute value is greater than the preset deviation threshold, it is determined that the output of the action transfer function is overestimated. When the deviation value is negative and the absolute value is greater than the preset deviation threshold, it is determined that the output of the action transfer function is underestimated.

[0211] A preset deviation threshold is set. If the deviation value is positive and its absolute value is greater than the deviation threshold, it means that the output of the action transfer function (hydrological fusion prediction value) is higher than the actual situation and there is an overestimation. If the deviation value is negative and its absolute value is greater than the deviation threshold, it means that the output is lower than the actual situation and there is an underestimation. If the absolute value of the deviation value is less than or equal to the deviation threshold, the output is judged to be normal and no parameter adjustment is required.

[0212] Step S1554: Determine the direction of parameter adjustment based on the determination result of the deviation value. If there is an overestimation, adjust the coefficient of the action transfer function to decrease it. If there is an underestimation, adjust the coefficient of the action transfer function to increase it.

[0213] If the value is determined to be overestimated, the coefficients of the transfer function (such as a) should be adjusted to decrease; if the value is determined to be underestimated, the coefficients should be adjusted to increase; the adjustment direction of the constant term b is the same as that of the coefficients.

[0214] Step S1555: Calculate the parameter adjustment range, which is proportional to the absolute value of the deviation value. The larger the absolute value of the deviation value, the larger the parameter adjustment range.

[0215] The parameter adjustment range is calculated as follows: Adjustment range = absolute value of deviation * adjustment coefficient, where the adjustment coefficient is a preset proportional constant. The larger the absolute value of the deviation, the larger the adjustment range.

[0216] Step S1556: Based on the parameter adjustment direction and parameter adjustment magnitude, modify the initial parameters of the action transfer function to obtain the adjusted parameters.

[0217] Depending on the direction of parameter adjustment, the initial parameters (coefficient a and constant term b) are added to or subtracted by the adjustment magnitude to obtain the adjusted parameters. For example, if coefficient a needs to be reduced and the adjustment magnitude is Δa, then the adjusted coefficient a' = a - Δa.

[0218] Step S1557: Substitute the adjusted parameters into the action transfer function, recalculate the input data for the first extrapolation time step, and obtain the verification prediction value.

[0219] Substitute the adjusted parameters into the action transfer function, and recalculate the action transfer using the input data of the first extrapolation time step (initial input data) to obtain the verification prediction value.

[0220] Step S1558: Calculate the verification deviation between the predicted value and the current actual hydrological evolution unit data. If the verification deviation is less than the preset deviation threshold, then determine the adjusted parameters as the parameters to be used for the second simulation time step.

[0221] Calculate the verification deviation value = verification prediction value - current actual hydrological evolution unit data. If the absolute value of the verification deviation value is less than the preset deviation threshold, the adjusted parameter is valid and is determined as the parameter used in the action transfer function in the second simulation time step.

[0222] Step S1559: If the verification deviation value is still greater than the preset deviation threshold, repeat the parameter adjustment direction determination, parameter adjustment range calculation and parameter modification operations until the verification deviation value is less than the preset deviation threshold.

[0223] If the absolute value of the verification deviation is still greater than the preset deviation threshold, return to step S1553, re-analyze the positive and negative attributes and absolute value of the verification deviation, determine the new parameter adjustment direction and adjustment range, modify the parameters and perform verification again until the absolute value of the verification deviation is less than the preset deviation threshold.

[0224] Step S15510: Use the final determined parameters as the parameters for the action transfer function in the second derivation time step.

[0225] After completing the parameter adjustment and passing the verification, the final adjusted parameters are set as the parameters used in the action transfer function in the second simulation time step.

[0226] Step S156: Collect the current meteorological evolution unit data and the current equipment evolution unit data corresponding to the second simulation time step, and combine them with the hydrological fusion prediction value of the first simulation time step as the input data for the second simulation time step.

[0227] The monitoring system collects the current regional meteorological data and the current power station equipment operation data corresponding to the second simulation time step, converts them into current meteorological evolution unit data and current equipment evolution unit data, and uses the hydrological fusion prediction value of the first simulation time step as the current hydrological evolution unit data. All three are used as the input data for the second simulation time step.

[0228] Step S157: Repeat the action transfer calculation, path fusion processing and simulation parameter adjustment operations until the total number of simulation steps is calculated and the hydrological fusion prediction value for each simulation time step is obtained.

[0229] Following the operational procedures of steps S154 (including sub-steps) and S155 (including sub-steps), the input data for the second simulation time step is processed for action transfer calculation and path fusion to obtain the hydrological fusion prediction value for the second simulation time step. Based on this prediction value, the parameters of the action transfer function in the third simulation time step are adjusted. Then, the current meteorological evolution unit data and the current equipment evolution unit data corresponding to the third simulation time step are collected and combined with the hydrological fusion prediction value of the second simulation time step as input data for simulation calculation. This process is repeated until the calculation of all simulation steps (e.g., 24 steps) is completed, and the hydrological fusion prediction value for each simulation time step is obtained.

[0230] Step S158: Organize the hydrological fusion prediction values ​​for each simulation time step into a time series, arrange them in chronological order according to the simulation time steps, and form a hydrological trend prediction sequence.

[0231] All hydrological prediction values ​​for all time steps are arranged sequentially according to the order of the time steps (from the first to the last) to form a hydrological trend prediction sequence for the power station.

[0232] Figure 2 The illustration shows exemplary hardware and software components of a power plant hydrological trend prediction system 100 based on big data analysis, which can implement the ideas of this application, according to some embodiments of this application. For example, processor 120 can be used in the power plant hydrological trend prediction system 100 based on big data analysis and to perform the functions in this application.

[0233] The power plant hydrological trend prediction system 100 based on big data analysis can be a general-purpose server or a special-purpose server; both can be used to implement the power plant hydrological trend prediction method based on big data analysis of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0234] For example, a power plant hydrological trend prediction system 100 based on big data analytics may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the power plant hydrological trend prediction system 100 based on big data analytics may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The power plant hydrological trend prediction system 100 based on big data analytics also includes an I / O interface 150 between the computer and other input / output devices.

[0235] For ease of explanation, only one processor is described in the power plant hydrological trend prediction system 100 based on big data analysis. However, it should be noted that the power plant hydrological trend prediction system 100 based on big data analysis in this application may also include multiple processors. Therefore, the steps executed by one processor described in this application may also be executed jointly by multiple processors or individually. For example, if the processor of the power plant hydrological trend prediction system 100 based on big data analysis executes steps A and B, it should be understood that steps A and B may also be executed jointly by two different processors or individually by one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0236] Furthermore, this embodiment of the invention also provides a readable storage medium, which has computer-executable instructions pre-set in it. When the processor executes the computer-executable instructions, the above-mentioned method for predicting hydrological trends in power plants based on big data analysis is implemented.

[0237] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for predicting hydrological trends in power plants based on big data analysis, characterized in that, The method includes: Acquire multi-source hydrological correlation data within the power station's jurisdiction. The multi-source hydrological correlation data includes hydrological records from monitoring points, regional meteorological records, and power station equipment operation records. The hydrological records from monitoring points correspond to hydrological evolution units, the regional meteorological records correspond to meteorological evolution units, and the power station equipment operation records correspond to equipment evolution units. A correlation evolution analysis is performed on the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit to generate a multi-dimensional hydrological evolution correlation map. The multi-dimensional hydrological evolution correlation map is used to present the dynamic interaction relationship between different evolution units. From the multi-dimensional evolutionary correlation map of water conditions, key evolutionary paths that influence water condition trends are extracted. These key evolutionary paths are correlation sequences that continuously interact between different evolutionary units and have a dominant influence on water condition changes. A dynamic model for predicting hydrological trends is constructed based on the key evolution paths, and the dynamic model for predicting hydrological trends includes the rules for the transmission of the effects of the key evolution paths. The current hydrological evolution unit data, current meteorological evolution unit data, and current equipment evolution unit data are input into the hydrological trend dynamic extrapolation model. The trend extrapolation operation is executed through the action transmission rules to generate a hydrological trend prediction sequence for the power station. The step of performing correlation evolution analysis on the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit to generate a multi-dimensional hydrological evolution correlation map includes: Time series data of hydrological evolution units are extracted from the hydrological records of the monitoring points, time series data of meteorological evolution units are extracted from the regional meteorological records, and time series data of equipment evolution units are extracted from the power station equipment operation records. The time series data of the hydrological evolution unit, the meteorological evolution unit, and the equipment evolution unit are divided into multiple evolution period units according to the same time granularity; Feature capture is performed on the hydrological evolution unit data within each evolution period unit to generate hydrological period features; feature capture is performed on the meteorological evolution unit data within each evolution period unit to generate meteorological period features; feature capture is performed on the equipment evolution unit data within each evolution period unit to generate equipment period features. Calculate the interaction strength between the hydrological time period characteristics and the meteorological time period characteristics within the same evolution time period unit; calculate the interaction strength between the hydrological time period characteristics and the equipment time period characteristics within the same evolution time period unit; calculate the interaction strength between the meteorological time period characteristics and the equipment time period characteristics within the same evolution time period unit. Based on the changes in the intensity of interaction within units at different evolutionary stages, the interaction change patterns between evolutionary units are identified, including interaction enhancement patterns, interaction weakening patterns, and interaction stabilization patterns. The water situation evolution unit, the meteorological evolution unit, and the equipment evolution unit are used as graph nodes. The intensity of the effect within the unit in different evolution periods is used as the attribute of the association edge between nodes, and the effect change mode is used as the dynamic identifier of the association edge. According to the time sequence of the evolution period units, the nodes, associated edge attributes and dynamic identifiers are mapped to the graph structure in sequence to form an initial evolutionary association graph containing the time dimension. The initial evolutionary correlation graph is structurally optimized by removing correlation edges with an effect strength lower than a preset effect threshold within the evolutionary time period unit, and retaining correlation edges with an effect strength higher than the preset effect threshold and their corresponding dynamic identifiers. By integrating the optimized node, associated edge attributes, and dynamic identifiers, a multi-dimensional evolutionary association map of hydrological conditions is generated.

2. The method for predicting hydrological trends in power plants based on big data analysis according to claim 1, characterized in that, The key evolutionary paths influencing water situation trends, extracted from the multi-dimensional evolutionary correlation map of water conditions, include: The node connection structure of the multidimensional evolution correlation map of the hydrological situation is analyzed to determine the direct and indirect correlations between the evolutionary unit corresponding to each node and the evolutionary units corresponding to other nodes. Extract the attribute information of all associated edges in the multidimensional evolution correlation map of the hydrological situation. The attribute information includes the intensity and change pattern of the effect within different evolution time units. A time series analysis is performed on the influence strength of each associated edge to calculate the cumulative influence strength of the associated edge within a continuous evolution time period unit. The cumulative influence strength is the result of the superposition of influence strength within the continuous evolution time period unit. Based on the cumulative value of the influence strength of the associated edges, different associated edges connected to the same node are sorted, and the associated edges with the highest cumulative value of influence strength are selected as the core associated edges. An initial evolution path is constructed based on the core associative edge, wherein the initial evolution path is a path segment connecting two nodes through the core associative edge; By analyzing the correlation between initial evolutionary paths, initial evolutionary paths with common nodes and consistent change patterns are spliced ​​together to form a longer evolutionary path sequence. For each evolution path sequence, an action transmission continuity analysis is performed to calculate the number of action transmission interruptions in the evolution path sequence within a continuous evolution time period. The number of action transmission interruptions is the number of times the action strength of the associated edge in the evolution path sequence is lower than a preset transmission threshold. Evolutionary path sequences with fewer than a preset interruption threshold are selected as candidate key evolutionary paths. The hydrological impact of candidate critical evolution paths is assessed, and the change range of hydrological evolution unit data corresponding to each candidate critical evolution path is calculated. The change range of hydrological evolution unit data is the range of change of hydrological records at monitoring points under the influence of candidate critical evolution paths. Candidate critical evolution paths with data changes exceeding a preset threshold are selected and identified as critical evolution paths influencing hydrological trends.

3. The method for predicting hydrological trends in power plants based on big data analysis according to claim 1, characterized in that, The construction of the dynamic prediction model of hydrological trends based on the key evolution path includes: Extract the node types and the variation patterns of the influence intensity of the edges between nodes in each key evolution path. The node types include hydrological evolution unit nodes, meteorological evolution unit nodes, and equipment evolution unit nodes. The data input type of different nodes in the simulation process is determined according to the node type. The hydrological evolution unit node corresponds to the hydrological data input type, the meteorological evolution unit node corresponds to the meteorological data input type, and the equipment evolution unit node corresponds to the equipment operation data input type. Based on the variation law of the influence strength of the associated edges in the critical evolution path, an influence transfer function is established. The influence transfer function is used to describe the data transformation relationship when the data of the previous node is transferred to the next node through the associated edge. Each critical evolution path is assigned a projection weight, which is determined based on the magnitude of change in the hydrological evolution unit data corresponding to the critical evolution path. The greater the magnitude of change in the hydrological evolution unit data, the higher the projection weight. A path fusion module is used to construct a dynamic model of hydrological trend. The path fusion module is used to receive the simulation results of multiple key evolution paths and fuse the simulation results according to the simulation weights. Set the simulation time step of the dynamic simulation model of hydrological trend, and the simulation time step is consistent with the time granularity of the evolution period unit in the multidimensional evolution correlation map of hydrological conditions; A simulation state update mechanism is established for a dynamic simulation model of water conditions. The simulation state update mechanism is used to adjust the parameters of the action transfer function in the next time step based on the simulation results of the current time step. The core framework of the dynamic hydrological trend simulation model is formed by integrating data input types, action transfer functions, simulation weights, path fusion modules, simulation time steps, and simulation status update mechanisms. The core framework of the dynamic inference model for water conditions is logically verified. Through logical comparison, the logical consistency between the action transfer functions of different key evolution paths is verified. If logical conflicts exist, the parameters of the action transfer functions are adjusted. Through rule verification, the matching between the weight fusion rules of the path fusion module and the inference logic is verified. If they do not match, the weight fusion rules are modified. After completing the logical verification, the final structure of the dynamic simulation model of water situation trends is determined.

4. The method for predicting hydrological trends in power plants based on big data analysis according to claim 3, characterized in that, The establishment of the action transfer function based on the variation law of the influence strength of associated edges in the key evolution path includes: Extract the influence intensity data of the associated edges in the key evolution path within different evolution time periods to form an influence intensity time series; Trend fitting is performed on the time series of action intensity to obtain a fitting curve of action intensity changing with time, and the fitting curve is used to reflect the changing trend of action intensity; Analyze the variation characteristics of the fitted curve to determine the inflection point of the change in action intensity and the action intensity value at the inflection point. The inflection point is the evolution period unit where the trend of action intensity change changes. Based on the inflection point, the time series of the effect intensity is divided into multiple effect stages, and the trend of effect intensity changes remains consistent within each effect stage. Correlation analysis was performed on the intensity data of each action phase and the previous node data and the next node data in the corresponding evolution period unit. First, the previous node data and the next node data were converted into dimensionless standardized data to obtain standardized previous node data and standardized next node data. Then, the correspondence between the standardized previous node data and the product of the action intensity and the standardized next node data was determined. Based on the correspondence, a local transfer function is constructed for each action stage. The input of the local transfer function is the data of the previous node and the action intensity, and the output is the predicted value of the data of the next node. Determine the transition conditions between different action stages, the transition conditions being set based on the change range of the previous node data and the change range of the action intensity; Multiple local transfer functions are connected in series according to the time sequence of their action phases, and an action transfer function is established by combining the transition conditions. The action transfer function can output the corresponding predicted value of the next node data based on the input previous node data and the action intensity of the unit in the current evolution period. Error verification is performed on the action transfer function by inputting the previous node data and action intensity within the historical evolution period unit into the action transfer function and comparing the difference between the predicted value of the next node data and the actual next node data. Adjust the parameters of the action transfer function based on the error verification results to reduce the difference between the predicted value and the actual data until the difference is less than the preset error threshold.

5. The method for predicting hydrological trends in power plants based on big data analysis according to claim 3, characterized in that, The path fusion module for constructing the dynamic inference model of hydrological trends includes: The input data type of the path fusion module is determined. The input data type is the intermediate hydrological prediction result output by each key evolution path. The intermediate hydrological prediction result includes the hydrological prediction value of the corresponding evolution time period unit. A weighting coefficient is assigned to the intermediate hydrological prediction results for each critical evolution path, and the weighting coefficient is consistent with the inference weight of the critical evolution path. A weighted fusion algorithm is established, which is used to multiply the intermediate hydrological prediction results of each key evolution path with the corresponding weight coefficient to obtain the weighted hydrological prediction value. The initial fused prediction value is obtained by summing all the weighted hydrological prediction values. A fusion result correction mechanism is constructed for the path fusion module. The fusion result correction mechanism is used to adjust the initial fusion prediction value based on the deviation between the historical fusion results and the actual hydrological data. The average deviation between the historical fusion results and the actual hydrological data is calculated, and the average deviation is the arithmetic mean of the deviations within multiple evolution period units; The correction factor is determined based on the average deviation. When the average deviation is positive, the correction factor is less than 1; when the average deviation is negative, the correction factor is greater than 1. Multiply the initial fusion prediction value by the correction coefficient to obtain the corrected fusion prediction value; Establish output rules for the path fusion module. These output rules are used to arrange the corrected fusion prediction values ​​according to the time step of the dynamic simulation model of the hydrological trend, forming a phased fusion prediction sequence. The input data types, weight coefficients, weight fusion algorithms, fusion result correction mechanisms, correction coefficient calculation methods, and output rules are integrated to form the path fusion module.

6. The method for predicting hydrological trends in power plants based on big data analysis according to claim 1, characterized in that, The process of inputting current hydrological evolution unit data, current meteorological evolution unit data, and current equipment evolution unit data into the hydrological trend dynamic extrapolation model, and executing trend extrapolation operations through the action transmission rules to generate a power station hydrological trend prediction sequence includes: Collect current hydrological data from monitoring points within the power station's jurisdiction and convert it into current hydrological evolution unit data; collect current regional meteorological data and convert it into current meteorological evolution unit data; collect current power station equipment operation data and convert it into current equipment evolution unit data. Determine the start and end times of the dynamic simulation model for hydrological trends, and calculate the total number of simulation steps based on the simulation time step of the dynamic simulation model for hydrological trends. Input the current hydrological evolution unit data, current meteorological evolution unit data, and current equipment evolution unit data into the corresponding nodes of the hydrological trend dynamic extrapolation model as the initial input data for the extrapolation start time; The action transfer function in the dynamic simulation model of hydrological trend is called to perform action transfer calculation on the initial input data to obtain the intermediate hydrological prediction result for the first simulation time step. The intermediate hydrological prediction result for the first simulation time step is then input into the path fusion module of the dynamic simulation model of hydrological trend and fused according to the path fusion rules to obtain the hydrological fusion prediction value for the first simulation time step. Based on the dynamic simulation model of hydrological trends and the simulation state update mechanism, the parameters of the action transfer function in the second simulation time step are adjusted based on the hydrological fusion prediction value of the first simulation time step. Collect the current meteorological evolution unit data and the current equipment evolution unit data corresponding to the second simulation time step, and combine them with the hydrological fusion prediction value of the first simulation time step as the input data for the second simulation time step; Repeatedly perform the action transfer calculation, path fusion processing and simulation parameter adjustment operations until the total number of simulation steps is calculated and the hydrological fusion prediction value for each simulation time step is obtained; The hydrological fusion prediction values ​​for each simulation time step are processed into a time series and arranged in chronological order according to the simulation time steps to form a hydrological trend prediction sequence.

7. The method for predicting hydrological trends in power plants based on big data analysis according to claim 6, characterized in that, The aforementioned simulation state update mechanism based on the dynamic simulation model of hydrological trends adjusts the parameters of the action transfer function in the second simulation time step based on the hydrological fusion prediction value of the first simulation time step, including: Extract the initial parameters of the action transfer function in the first simulation time step, the initial parameters including the coefficients and constants in the action transfer function; Calculate the deviation between the hydrological fusion prediction value and the current actual hydrological evolution unit data at the first simulation time step. The deviation value is the result of subtracting the current actual hydrological evolution unit data from the hydrological fusion prediction value. Analyze the positive and negative attributes and absolute value of the deviation value. When the deviation value is positive and the absolute value is greater than the preset deviation threshold, it is determined that the output of the action transfer function is overestimated. When the deviation value is negative and the absolute value is greater than the preset deviation threshold, it is determined that the output of the action transfer function is underestimated. The direction of parameter adjustment is determined based on the deviation value. If there is an overestimation, the coefficient of the action transfer function is adjusted to decrease it; if there is an underestimation, the coefficient of the action transfer function is adjusted to increase it. The parameter adjustment range is calculated, and the parameter adjustment range is proportional to the absolute value of the deviation value. The larger the absolute value of the deviation value, the larger the parameter adjustment range. Based on the direction and magnitude of parameter adjustment, the initial parameters of the action transfer function are modified to obtain the adjusted parameters; Substitute the adjusted parameters into the action transfer function and recalculate the input data for the first extrapolation time step to obtain the verification prediction value; Calculate the verification deviation between the predicted value and the current actual hydrological evolution unit data. If the verification deviation is less than the preset deviation threshold, then determine the adjusted parameters as the parameters to be used for the second simulation time step. If the verification deviation value is still greater than the preset deviation threshold, repeat the parameter adjustment direction determination, parameter adjustment range calculation and parameter modification operations until the verification deviation value is less than the preset deviation threshold. The finalized parameters will be used as the parameters for the action transfer function in the second simulation time step.

8. The method for predicting hydrological trends in power plants based on big data analysis according to claim 1, characterized in that, The step of capturing features from hydrological evolution unit data within each evolution period to generate hydrological time period features includes: Basic hydrological parameters are extracted from the hydrological evolution unit data within the evolution time period unit. These basic hydrological parameters include water level records and water flow records of monitoring points. Calculate the mean of the basic hydrological parameters within the evolution period unit, where the mean is the arithmetic mean of all basic hydrological parameter data within the evolution period unit; Calculate the range of variation of basic hydrological parameters within the evolution period unit, where the range of variation is the difference between the maximum and minimum values ​​of the basic hydrological parameters within the evolution period unit; The direction of change of basic hydrological parameters within the evolution period unit is analyzed, wherein the direction of change is the upward or downward trend of the basic hydrological parameters within the evolution period unit from the start time to the end time. Extract the extreme points of the basic hydrological parameters within the evolution period unit, where the extreme points are the maximum and minimum points of the basic hydrological parameters within the evolution period unit; Calculate the number of times extreme points occur within an evolutionary time unit, and count the number of times maximum and minimum points occur; A set of feature terms is constructed based on the mean, range of variation, direction of variation, and frequency of extreme points of basic hydrological parameters; Each feature in the feature set is quantized to convert the direction of change into a quantized value, with an upward trend corresponding to a positive value and a downward trend corresponding to a negative value. According to the preset feature sorting rules, the quantized feature items are arranged in sequence to form an ordered feature value sequence; The characteristic numerical sequence is normalized, and the normalized characteristic numerical sequence is determined as the hydrological time period characteristic.

9. A power station hydrological trend prediction system based on big data analysis, characterized in that, The power station hydrological trend prediction system based on big data analysis includes a processor and a memory, the memory and the processor are connected, the memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the power station hydrological trend prediction method based on big data analysis as described in any one of claims 1-8.

Images