An integrated photovoltaic and storage intelligent dispatching system

Abstract

The application relates to the technical field of light storage intelligent scheduling, and discloses a light storage integrated intelligent scheduling system. The system comprises a photovoltaic and energy storage data monitoring module, a scheduling stability evaluation module, a scheduling strategy generation module and a scheduling execution and feedback module. The photovoltaic and energy storage data monitoring module captures the power generation of photovoltaic power generation units and energy storage units, the energy storage state and environmental parameter data sequences in a preset historical period in real time through a sensor array; the scheduling stability evaluation module globally traverses the data sequences to evaluate data volatility and determine corresponding stability indexes; the scheduling strategy generation module takes the stability indexes as input indexes, densely searches in a scheduling strategy configuration library to identify an optimal scheduling strategy; and the scheduling execution and feedback module implements integrated scheduling control based on the optimal strategy, generates a periodic scheduling data set and analyzes and outputs final results. The system can realize efficient collaborative operation of a light storage system.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic and storage intelligent scheduling, and in particular to an integrated photovoltaic and storage intelligent scheduling system. Background Art

[0002] As renewable energy continues to increase its share in the energy mix, photovoltaic power generation is gaining widespread adoption due to its clean and renewable nature. However, photovoltaic power generation is significantly affected by natural factors such as light intensity, temperature, and weather, resulting in significant intermittent and fluctuating output power. This unstable generation, if directly connected to the power grid, can impact the safe and stable operation of the grid, leading to problems such as voltage fluctuations and frequency offsets.

[0003] To mitigate the instability of photovoltaic power generation, energy storage technology has been introduced to form integrated photovoltaic and energy storage systems. Energy storage units can store excess photovoltaic power and release it when power generation is insufficient, thereby smoothing power fluctuations. However, the scheduling and management of current photovoltaic and energy storage systems still face many challenges. Traditional scheduling methods often rely on empirical rules or simple threshold control, lacking comprehensive monitoring and analysis of photovoltaic power generation units, energy storage units, and environmental parameters.

[0004] At the data monitoring level, existing systems often focus only on a few key parameters. Data collection lacks real-time and completeness, making it difficult to capture dynamic changes in system operation. The use of historical data is also limited, making it impossible to predict future system behavior by analyzing past operating conditions.

[0005] The generation of dispatch strategies lacks scientificity and flexibility. Due to a failure to effectively assess the stability of power generation, energy storage status, and environmental parameters, dispatch strategy development often lags behind and cannot be adjusted promptly to changes in system operating conditions. Furthermore, the inadequate retrieval mechanism within the dispatch strategy configuration library makes it difficult to quickly and accurately match the optimal strategy for the current operating conditions.

[0006] The existing system suffers from low control accuracy in scheduling execution and feedback, lacking an effective tracking and adjustment mechanism for scheduling task execution. The system's ability to analyze and process periodic scheduling data is weak, failing to provide valuable insights for subsequent scheduling optimization. This makes it difficult to improve the overall operational efficiency of solar-plus-storage systems, hindering the further development and application of integrated solar-plus-storage technology. Summary of the Invention

[0007] The purpose of the present invention is to provide a photovoltaic-storage integrated intelligent scheduling system to solve the problems raised in the above background technology.

[0008] To achieve the above objectives, the present invention provides an integrated photovoltaic and energy storage intelligent scheduling system, the system comprising:

[0009] Photovoltaic and energy storage data monitoring module, used to perform real-time data capture of the sensor array, and obtain the power generation data series, energy storage status data series, and environmental parameter data series for the photovoltaic power generation unit and energy storage unit within a preset historical period;

[0010] a scheduling stability evaluation module, configured to globally traverse the power generation data sequence, the energy storage state data sequence, and the environmental parameter data sequence to perform data volatility evaluation and determine a power generation stability index, an energy storage state stability index, and an environmental parameter stability index;

[0011] A scheduling strategy generation module is used to perform an intensive search operation in a preset scheduling strategy configuration library using the power generation stability index, energy storage state stability index, and environmental parameter stability index as input indexes to identify the optimal scheduling strategy;

[0012] The scheduling execution and feedback module is used to implement integrated scheduling control of photovoltaic power generation units and energy storage units according to a preset scheduling task set based on the optimal scheduling strategy, generate a periodic scheduling data set, parse and process the periodic scheduling data set, and output the final scheduling result.

[0013] Preferably, the scheduling stability evaluation module is further used to:

[0014] Performing data cleaning on the power generation data sequence using outlier detection technology to generate a cleaned power generation data set;

[0015] The outlier detection technology identifies and removes data points that fall outside a preset fluctuation range, which is dynamically adjusted based on the statistical characteristics of the data distribution.

[0016] performing a standard deviation calculation on the cleaning power generation data set to obtain the power generation stability index;

[0017] Similar volatility assessment processing is performed on the energy storage state data sequence and the environmental parameter data sequence to obtain the energy storage state stability index and the environmental parameter stability index respectively.

[0018] Preferably, the scheduling stability evaluation module is further used to:

[0019] Rearranging the generated power data sequence in ascending order of values, and selecting a data point at an intermediate value as a baseline reference line;

[0020] Calculating the upper and lower quartiles of the power generation data sequence;

[0021] The upper and lower quartiles are used as boundary constraints and combined with the baseline reference line to construct a data distribution framework.

[0022] Preferably, the scheduling strategy generation module is further used to:

[0023] Acquire multiple sample power generation stability indicators, multiple sample energy storage state stability indicators, multiple sample environmental parameter stability indicators, and corresponding multiple sample scheduling strategies as training data;

[0024] Constructing a multidimensional feature space, wherein the coordinate origin of the multidimensional feature space is set as a reference origin, the first axis represents the power generation stability index dimension, the second axis represents the energy storage state stability index dimension, and the third axis represents the environmental parameter stability index dimension;

[0025] The training data is mapped into the multidimensional feature space to form a plurality of sample space coordinate points, and the plurality of sample space coordinate points are marked using the plurality of sample scheduling strategies to generate the scheduling strategy configuration library.

[0026] Preferably, the scheduling strategy generation module is further used to:

[0027] A first reference plane passing through the power generation stability indicator and parallel to a plane formed by the second axis and the third axis is defined in the scheduling strategy configuration library;

[0028] A second reference plane is defined in the scheduling strategy configuration library, passing through the energy storage state stability index and being parallel to a plane formed by the first axis and the third axis;

[0029] A third reference plane is defined in the scheduling strategy configuration library, which passes through the environmental parameter stability index and is parallel to the plane formed by the first axis and the second axis;

[0030] The area defined by the plane formed by the second axis and the third axis, the plane formed by the first axis and the third axis, the plane formed by the first axis and the second axis, and the first reference plane, the second reference plane, and the third reference plane is used as a strategy subspace, wherein the strategy subspace includes a plurality of strategy sample points;

[0031] An intensive search operation is performed on the multiple policy sample points to locate a target policy sample point, and the sample scheduling policy corresponding to the target policy sample point is used as the optimal scheduling policy.

[0032] Preferably, the scheduling strategy generation module is further used to:

[0033] Extracting a central strategy sample point of the strategy subspace, taking the central strategy sample point as a starting position, and constructing a central neighborhood range according to a preset dense search radius;

[0034] Counting the total number of strategy sample points within the central neighborhood, and dividing the total number by the volume of the central neighborhood to obtain a central neighborhood density value;

[0035] Randomly select a strategic sample point from the edge of the central neighborhood range as the first retrieval point, and calculate the neighborhood density value of the first retrieval point;

[0036] Determine whether the neighborhood density value of the first search point is greater than or equal to the center neighborhood density value. If so, update the first search point as the starting position, continue to perform intensive search operations until a preset number of search iterations is reached, and use the final search point as the target strategy sample point.

[0037] Preferably, the scheduling strategy generation module is further used to:

[0038] If the neighborhood density value of the first search point is less than the center neighborhood density value, the search failure counter whose initial value is zero is incremented, and another strategic sample point is randomly selected from the edge of the center neighborhood range as a new first search point for intensive search analysis;

[0039] When the value of the retrieval failure counter exceeds a preset maximum failure threshold, the central strategy sample point is used as the target strategy sample point.

[0040] Preferably, the system further comprises:

[0041] A trend prediction module is used to perform time series prediction processing on the power generation stability index, the energy storage state stability index and the environmental parameter stability index to generate a predicted power generation stability index, a predicted energy storage state stability index and a predicted environmental parameter stability index;

[0042] A scheduling compensation module is used to perform compensation optimization on the optimal scheduling strategy based on the predicted power generation stability index, the predicted energy storage state stability index, and the predicted environmental parameter stability index to generate an enhanced scheduling strategy;

[0043] The scheduling execution and feedback module is further configured to implement scheduling control based on the enhanced scheduling strategy.

[0044] Preferably, the scheduling compensation module is further used to:

[0045] Using the predicted power generation stability index, the predicted energy storage state stability index, and the predicted environmental parameter stability index as input indexes, performing an association matching operation in the scheduling strategy configuration library to generate a second calibration strategy library;

[0046] Identifying a strategy feature triggering interval in the second calibration strategy library to obtain a second strategy feature space;

[0047] Iteratively generate a strategy based on the preset fitness constraint and the second strategy feature space to form a second scheduling strategy solution;

[0048] The second scheduling strategy solution is integrated and optimized with the optimal scheduling strategy to obtain the enhanced scheduling strategy.

[0049] Preferably, the scheduling execution and feedback module is further used to:

[0050] Transmitting the periodically scheduled data set to a distributed storage port;

[0051] The distributed storage port synchronously calls the operating status parameters of each storage node for analysis and processing, and stores the periodically scheduled data set partitions to the target storage node according to the analysis results;

[0052] The operating status parameters include memory utilization parameters, transmission delay parameters, and data read and write rate parameters of the storage node.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] By setting up a photovoltaic and energy storage data monitoring module, the sensor array can capture real-time data, obtaining multiple data sequences of photovoltaic power generation units and energy storage units within a preset historical period. This enables the system to fully understand the operating status of the photovoltaic storage system, including changes in power generation, fluctuations in energy storage status, and the impact of environmental parameters, providing rich and detailed basic information for subsequent scheduling.

[0055] The Dispatch Stability Assessment module performs a global traversal of various data sequences to assess data volatility and determine corresponding stability indicators. This comprehensive assessment method provides in-depth analysis of the stability of power generation, energy storage status, and environmental parameters, clearly presenting potential fluctuations in system operation. By understanding these indicators, the system can more accurately understand the operating characteristics of the solar-energy storage system and identify key factors that may affect its stable operation.

[0056] The scheduling strategy generation module uses stability indicators as input indexes and performs intensive searches within a pre-set scheduling strategy configuration library to identify the optimal scheduling strategy. This multi-metric search approach breaks away from the limitations of traditional empirical scheduling and can quickly find a strategy that matches the current system state from a rich strategy library. This allows scheduling strategies to be formulated more closely aligned with actual operational needs, enhancing their adaptability and pertinence.

[0057] The dispatch execution and feedback module implements integrated dispatch control based on the optimal dispatch strategy and parses and processes periodic dispatch data sets to output the final results. Integrated dispatch control enables the coordinated operation of photovoltaic power generation units and energy storage units, allowing them to cooperate with each other under different operating conditions and fully utilize their respective capabilities. The parsing and processing of dispatch data can clearly demonstrate the dispatch results, facilitate the timely identification of problems in the dispatch process, and enable the system to continuously adjust and optimize the dispatch method based on actual conditions, allowing the photovoltaic storage system to maintain a good operating state under various environmental conditions and improve the overall level of coordinated operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a working principle diagram of the photovoltaic and energy storage integrated intelligent scheduling system of the present invention;

[0059] Figure 2 Flowchart of data cleaning and indicator calculation for the scheduling stability evaluation module;

[0060] Figure 3 Flowchart built for the scheduling policy configuration library;

[0061] Figure 4 Flowchart for handling retrieval failure;

[0062] Figure 5 Flowchart of trend prediction and scheduling compensation. DETAILED DESCRIPTION

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0064] See also Figure 1 The present invention provides an integrated photovoltaic and energy storage intelligent dispatching system, which includes: a photovoltaic and energy storage data monitoring module, a dispatching stability assessment module, a dispatching strategy generation module, and a dispatching execution and feedback module. The specific implementation is as follows:

[0065] The photovoltaic and energy storage data monitoring module captures real-time data from photovoltaic power generation units and energy storage units through a sensor array. The module captures data from the sensor array over a preset historical period, generating data series on power generation, energy storage status, and environmental parameters.

[0066] The dispatch stability evaluation module performs global traversal on the power generation data sequence, the energy storage state data sequence and the environmental parameter data sequence, and starts a data volatility evaluation process. The evaluation process calculates the power generation stability index, the energy storage state stability index and the environmental parameter stability index.

[0067] The dispatch strategy generation module receives the power generation stability index, the energy storage state stability index and the environmental parameter stability index as input indexes. The module performs intensive retrieval operation in the preset dispatch strategy configuration library by using the input indexes, and identifies the optimal dispatch strategy.

[0068] The dispatch execution and feedback module drives the photovoltaic power generation unit and the energy storage unit to implement integrated dispatch control according to the preset dispatch task set based on the optimal dispatch strategy.

[0069] The module generates a periodic dispatch data set, analyzes and processes it, and outputs the final dispatch result.

[0070] Embodiment 1: refer to Figure 2 After the photovoltaic and energy storage data monitoring module captures the power generation data sequence, the energy storage state data sequence and the environmental parameter data sequence, the dispatch stability evaluation module starts data processing operation. The module performs data cleaning process by using outlier detection technology on the power generation data sequence. The outlier detection technology dynamically sets the preset fluctuation range boundary value based on the statistical characteristics of the data sequence itself. The calculation of the boundary value depends on the discrete degree of the data point distribution, and the boundary threshold value is updated in real time by continuously monitoring the change trend of the data sequence. Specifically, the module scans all data points of the power generation data sequence, and analyzes the numerical deviation of each data point from the adjacent data points one by one. If the numerical deviation of a data point exceeds the current preset fluctuation range, the point is marked as an outlier. The system removes all identified outliers to generate a cleaned power generation data set. The set only retains valid data points that meet the fluctuation range constraint, and excludes the interference of abnormal jumps or measurement noise.

[0071] After the data cleaning is completed, the dispatch stability evaluation module performs standard deviation calculation on the cleaned power generation data set. The standard deviation calculation uses a fixed algorithm to measure the discrete degree of the data points around the average value. The module first calculates the arithmetic mean of the cleaned power generation data set as the center reference of the data distribution. Then, the square value of the difference between each data point and the center reference is calculated, and the total square difference value is divided by the total number of data points to obtain the variance value. The square root operation is performed on the variance value to output the power generation stability index. The index quantifies the overall fluctuation amplitude of the power generation data, and the higher the value, the more significant the volatility.

[0072] For both the energy storage state data sequence and the environmental parameter data sequence, the dispatch stability evaluation module adopts the same processing logic to perform data cleaning and index calculation. The module independently applies an outlier detection technique to the energy storage state data sequence. The technique sets a fluctuation range threshold according to the numerical characteristics of the energy storage state data sequence, independently identifies and removes outliers in the sequence, and generates a cleaned energy storage state data set. Subsequently, a separate standard deviation calculation process is performed to output the energy storage state stability index. The processing process of the environmental parameter data sequence repeats the above operation, and the environmental parameter fluctuation range threshold of the data cleaning process is independent of other sequences, and the output environmental parameter stability index independently reflects the fluctuation characteristics of the environmental parameters.

[0073] The dispatch stability evaluation module performs a data distribution framework construction operation on the power generation data sequence. The module reorders all data points of the original power generation data sequence from small to large in value to generate a monotonically increasing ordered data sequence. A data point at the middle position in the ordered sequence is selected. If the total number of data points is odd, the middle point is the median point in the ordered sequence; if it is even, the middle point is the arithmetic mean of the two median points. The data point at the middle position is designated as the reference line, representing the central tendency of the data distribution.

[0074] The module continues to calculate the upper and lower quartiles. The upper quartile corresponds to the data point value at the 75% position in the ordered sequence, and the lower quartile corresponds to the data point value at the 25% position. The calculation process uses linear interpolation: if the target quantile is between two data points, a weighted average value is calculated in proportion. For example, the lower quartile position index value is (N+1) x 0.25 (N is the total number of data points). If the value is an integer, the corresponding sequence number data point value is directly taken; if it is a decimal, the weighted average value of the adjacent two points is taken. The upper quartile position index value is (N+1) x 0.75, and the same calculation logic is used.

[0075] The dispatch stability evaluation module constructs a three-dimensional data distribution framework using the upper and lower quartiles as boundary constraints and combining the reference line. The upper and lower quartiles form the horizontal fluctuation range boundary, and the reference line serves as the vertical axis. This framework forms a columnar space structure: with the reference line as the central axis, the upper and lower quartiles as the boundary coordinates of the top and bottom surfaces. Data points outside this framework space are considered abnormal fluctuations. The framework construction uses data space mapping technology to convert one-dimensional data sequences into a three-dimensional distribution model. The boundaries of this model are dynamically adjusted according to the data of the current historical period, and the framework volume automatically scales with the distance between the quartiles. The module simultaneously performs the same sorting, quartile calculation, and framework construction operations on the energy storage state data sequence and the environmental parameter data sequence. The reference line of the energy storage state data sequence is independently calculated as the middle position point of its own ordered sequence, and the upper and lower quartiles are calculated separately according to its data distribution; the environmental parameter sequence uses the same logic to generate an independent framework. Each sequence framework is constructed in parallel without data cross interference.

[0076] Throughout the implementation process, anomalous data points eliminated during the outlier detection phase are recorded in a separate log, standard deviation calculations are written to a real-time database, and the constructed data distribution framework generates a 3D coordinate model file. The module automatically refreshes all calculation parameters during the next monitoring cycle and iterates using new historical input data. All intermediate data and final results are synchronized to the input port of the scheduling strategy generation module via a network transmission protocol.

[0077] Example 2: See Figure 3 Before initiating the intensive search operation, the scheduling strategy generation module constructs a preset scheduling strategy configuration library. The module first retrieves multiple sets of sample data from the historical database, including sample power generation stability indicator values, sample energy storage state stability indicator values, and sample environmental parameter stability indicator values. Each set of sample indicator values ​​corresponds to a verified valid sample scheduling strategy record. This sample data forms a complete training dataset, covering the state combinations of photovoltaic power generation units and energy storage units under different operating scenarios.

[0078] After the training dataset is prepared, the scheduling strategy generation module initiates the multidimensional feature space construction process. This feature space uses a three-dimensional Euclidean coordinate system. The origin of the coordinate system is set to the zero reference point, which serves as the starting point for all dimensional calculations. The first axis of the space represents the dimension of the power generation stability indicator, and the coordinate scale expands linearly according to the indicator value. The second axis represents the dimension of the energy storage state stability indicator, and the scaling rules are the same as the first axis. The third axis independently represents the dimension of the environmental parameter stability indicator, and its scaling range is independently defined based on the physical properties of the environmental parameter. The three dimensional axes are mutually orthogonal to form a complete spatial framework.

[0079] The module initiates the spatial mapping program and reads each record in the training dataset one by one. For each record, the scheduling strategy generation module extracts the sample power generation stability index value and converts it into a first-axis coordinate position. The sample energy storage state stability index value is independently converted into a second-axis coordinate position, and the sample environmental parameter stability index value is converted into a third-axis coordinate position. After the three-dimensional coordinate values ​​are determined, a sample spatial coordinate point is generated in space. This coordinate point is tagged with an attribute that records its corresponding original sample scheduling strategy parameter set. After performing the same mapping operation on all records in the training dataset, multiple sample spatial coordinate points are formed in space, each carrying a corresponding strategy tag. The spatial point set and the strategy tag set together constitute the scheduling strategy configuration library entity. This configuration library is stored as a distributed database and supports high-speed coordinate query operations.

[0080] After the dispatch stability assessment module outputs the current power generation stability index, energy storage state stability index, and environmental parameter stability index, the dispatch strategy generation module uses them as input indexes for parsing. The module extracts the current power generation stability index value and converts it into first-axis coordinate values. The current energy storage state stability index value is converted into second-axis coordinate values, and the current environmental parameter stability index value is converted into third-axis coordinate values. This set of converted coordinates serves as the search anchor coordinates, initiating the spatial definition process in the dispatch strategy configuration library.

[0081] The scheduling strategy generation module constructs the reference plane based on the anchor point coordinates. First, the first reference plane is constructed: the plane is perpendicular to the first axis in space, and the intercept is the coordinate value of the current power generation stability index. The plane extension direction is parallel to the two-dimensional plane formed by the second axis and the third axis, covering the entire space. Secondly, the second reference plane is constructed: the plane is perpendicular to the second axis, the intercept is the coordinate value of the current energy storage state stability index, and the extension direction is parallel to the plane formed by the first axis and the third axis. Finally, the third reference plane is constructed: the plane is perpendicular to the third axis, the intercept is the coordinate value of the current environmental parameter stability index, and the extension direction is parallel to the plane formed by the first axis and the second axis. The construction process of the three planes adopts the principles of spatial analytic geometry, and its equations are automatically generated by the intercept.

[0082] The three reference planes and the spatial reference plane work together to define the strategy subspace. The strategy subspace is a closed region bounded by the following boundaries: the first reference plane in the power generation dimension; the second reference plane in the energy storage state dimension; and the third reference plane in the environmental parameter dimension. On the negative coordinate side, the planes formed by the first and second axes, the first and third axes, and the second and third axes serve as the basic boundaries. The strategy subspace forms an irregular hexahedron, containing some sample space coordinate points. The scheduling strategy generation module initiates an indexing process to retrieve all sample space coordinate points within this subspace and generate a set of strategy candidate points.

[0083] The intensive search operation is performed on the set of strategy candidate points. The scheduling strategy generation module calculates the coordinates of the geometric center position of the strategy subspace and locates the sample space coordinate point closest to the center position as the central strategy sample point. The module sets a fixed radius value as the preset intensive search radius and establishes a spherical neighborhood space with the central strategy sample point as the center of the sphere. The total number of sample space coordinate points contained in the spherical space is counted, and the ratio of this total number to the volume of the spherical space is calculated and recorded as the central neighborhood density value. The system randomly selects a sample space coordinate point from the boundary of the spherical neighborhood as the initial search anchor point. The point density within the same radius of the anchor point is calculated as the current neighborhood density value. When the current neighborhood density value is greater than or equal to the central neighborhood density value, the anchor point becomes the new search center, and the density comparison operation is performed iteratively. When the number of iterations reaches the preset search iteration limit, the final anchor point is marked as the target strategy sample point. The corresponding sample scheduling strategy of the target point is output to the downstream module as the optimal scheduling strategy. If the density of the first searched point is lower than that of the central neighborhood, the system triggers a fault-tolerance mechanism: A new candidate point is randomly selected on the boundary and compared with the rule. If the number of consecutive selections exceeds a preset maximum failure threshold, the central strategy sample point is automatically identified as the target point. The search results are encapsulated in a standard instruction format and transmitted to the scheduling execution and feedback module via the data bus.

[0084] The configuration library construction process is completed during the system initialization phase, and the mapping data is permanently stored in non-volatile memory. During the retrieval operation, the module monitors the boundary overflow of the three-dimensional coordinate value in real time. When the input indicator exceeds the historical sample range, the system uses a spatial interpolation algorithm to expand the retrieval range. After each strategy call, the current input indicator and the final strategy are appended to the training dataset as new samples, periodically triggering incremental updates of the configuration library. The update cycle is automatically set based on the system's operating time. The coordinate mapping algorithm uses double-precision floating-point operations to ensure the accuracy of spatial position calculations. The strategy tag storage uses a distributed hash table to support fast strategy matching and retrieval. All spatial definition and calculation operations are performed by a dedicated spatial calculation engine, and data input and output use a serialized transmission protocol.

[0085] Example 3: See Figure 4 The scheduling strategy generation module initiates a dense search operation in the strategy subspace. The module identifies the geometric center coordinate position of the strategy subspace. This coordinate is obtained by calculating the arithmetic mean of the coordinates of all strategy sample points in the subspace. The strategy sample point corresponding to the center position is defined as the initial center point. The module loads the preset dense search radius parameter, which is set to a fixed scalar value and the initial value is set based on the historical statistics of the sample point distribution density in the strategy configuration library. With the initial center point as the center of the sphere and the preset dense search radius value as the radius, a spherical neighborhood range is delineated in the three-dimensional feature space. This range belongs to the closed sphere area in Euclidean space.

[0086] The module performs a spherical neighborhood density measurement operation. It counts the total number of strategic sample points that fall within the spherical neighborhood. It obtains the neighborhood space volume data, which is automatically generated using the spherical volume calculation rule. The system calculates the central neighborhood density value, which is defined as:

[0087] ;

[0088] in, represents the central neighborhood density value, Represents the total number of strategy sample points in the neighborhood, Represents the volume of a spherical neighborhood. The volume is calculated using standard geometric formulas, with the radius being the preset dense search radius input.

[0089] The retrieval process enters the edge sampling phase. The module locates the surface boundary surface of the spherical neighborhood. A spatial point position is randomly selected from the boundary surface coordinate set. This random selection uses a uniform distribution sampling algorithm to ensure that each surface area has an equal probability of being selected. After determining the coordinates of the sampling point, the strategy sample point closest to the coordinate is found as the first retrieval point. With the first retrieval point as the center, a spherical neighborhood of equal radius is constructed. The module independently calculates the total number of content strategy sample points and the volume ratio within the neighborhood range, and outputs the neighborhood density value of the first retrieval point. .

[0090] The system starts the density comparison process. The density value of the neighborhood of the first retrieval point and the central neighborhood density value Perform numerical comparison. The value is greater than or equal to When the value of is reached, the search anchor position is updated: the first search point is used as the new center point. Based on the updated center point, a new spherical neighborhood is constructed using the same preset dense search radius. The density calculation, edge sampling and density comparison process are re-executed. This process is iterative, and the center point position and corresponding density value data are updated in each iteration. The number of iterations is recorded in the system counter, and the counter is automatically increased by one after each iteration. When the counter value reaches the preset search iteration threshold , terminate the iteration process. The strategy sample point corresponding to the current center point coordinate is selected as the target strategy sample point. The marked scheduling strategy of the target point is used as the final output.

[0091] When the first comparison was found When the system activates the fault-tolerant processing protocol, the retrieval failure counter is initialized. , assign the initial state zero value. Failure counter increment operation: The system re-performs uniform random sampling from the original center neighborhood boundary and selects a new surface boundary space point. The strategic sample point closest to the new position is determined as the substitute retrieval point. A spherical neighborhood is constructed with the substitute retrieval point as the center, and the neighborhood density value is independently calculated. . Perform density comparison again: If , then the substitute search point becomes the new center point, and the main iteration process continues to advance; if , the failure counter is incremented again and the sampling of the replacement point continues.

[0092] The fault tolerance protocol includes an interrupt constraint mechanism. When the retrieval failure counter The value exceeds the preset maximum failure threshold , the system forces the retrieval process to terminate. The original initial center point is directly identified as the target strategy sample point, and its corresponding scheduling strategy is used as the output solution. Failure threshold Set as an independent parameter, the value range is dynamically adjusted according to the point density characteristics of the strategy subspace: the threshold is automatically increased in sparse areas and decreased in dense areas.

[0093] At the operational level, the neighborhood boundary point sampling is generated using the Monte Carlo method. The spherical surface uniform sampling algorithm is based on the three-dimensional spherical coordinate transformation, the angle parameter and Each density comparison follows a uniform distribution in the range [0, 2π]. The radius value is fixed to a preset dense search radius constant. Neighbor searches for policy sample points are accelerated using a kd-tree spatial index structure, which is pre-generated when the policy subspace is loaded. Density score calculations are performed using a double-precision floating-point unit to avoid roundoff error accumulation. Each density comparison result is written to a real-time status register to drive the process control state machine.

[0094] Counter Parameters ( , ) Load the initial value through the configuration interface. The runtime monitoring system records the actual number of iterations and the number of failures. When the main iteration process is completed, the coordinates and attribute tags of the target strategy sample point are transmitted to the strategy output port through serialized data packets. In the fault-tolerant path, the spatial coordinates of the initial center point automatically trigger the coordinate remapping module to ensure that the output data format is compliant. After each retrieval operation, the system resets all counters and status flags and clears the data in the temporary neighborhood buffer. The strategy index result is encapsulated as a structure object and transmitted to the scheduling execution control end via the PCIe bus. The port response time is constrained by the system interrupt delay, and the buffer is designed as an 8KB circular queue. The mathematical basis of the entire process is high-dimensional space density clustering theory, and the operation process does not rely on gradient calculation.

[0095] Example 4: See Figure 5After the dispatch strategy generation module outputs the optimal dispatch strategy for the site, the trend prediction module starts running. This module inputs the power generation stability index, energy storage state stability index, and environmental parameter stability index data for the current cycle. Taking environmental parameters as an example, the system obtains an environmental parameter stability index value of 0.15, which corresponds to the standard deviation measurement of fluctuations in parameters such as temperature and irradiance over a historical period. The trend prediction module loads a time series database and retrieves the change trajectory of similar indicators over the past 72 hours. The module processes the input indicators using an adaptive weighted sliding window algorithm, with a window width set to 8 sampling periods. After data processing, it outputs a sequence of predicted values: the indicator changes over the next three operation cycles are calculated in increasing time steps. The final predicted values ​​for the power generation stability index are 0.18, the energy storage state stability index is 0.22, and the environmental parameter stability index is 0.19, forming a complete set of predicted indicators. All predicted values ​​are temporarily stored in a buffer register for subsequent module calls.

[0096] The dispatch compensation module receives the predicted indicator set and simultaneously obtains access to the dispatch policy configuration library. The module uses the predicted power stability indicator (0.18), the predicted energy storage state stability indicator (0.22), and the predicted environmental parameter stability indicator (0.19) as secondary input indexes. The dispatch policy configuration library activates the spatial matching engine to locate the predicted indicator coordinates within the three-dimensional feature space. This matching process generates a second calibration policy library entity, whose data is organized as follows:

[0097] ;

[0098] This table shows partial data from the second calibration strategy library, including four sample points. The coordinates of point P-1029 (0.18, 0.22, 0.19) exactly match the input prediction metric. The matching engine selects all sample points within a spatial distance threshold of 0.05 to form the strategy feature trigger interval. The interval is shaped like an ellipsoid, with the major axis radius scaled by the dimensional importance factor. The set of coordinate points within this interval forms the second strategy feature space, whose feature vector dimensions match those of the original space.

[0099] The scheduling compensation module loads a set of preset fitness constraints, including the maximum charge and discharge rate limit, the PV output fluctuation tolerance range, and the energy storage unit SOC safety threshold. The module executes an iterative strategy generation algorithm in the second strategy feature space:

[0100] Initialize the strategy population: All strategy solutions from SPS-881 to SPS-884 are used as the parent

[0101] Crossover operation: Randomly select two strategies to exchange control parameter sequence segments

[0102] Mutation operation: Modify the charge and discharge trigger threshold in the strategy with a 5% probability

[0103] Fitness evaluation: Check the degree to which each derivative solution satisfies the constraints

[0104] Bad solution elimination: remove policy variants that violate safety thresholds

[0105] After six rounds of iterative optimization, the final stable second scheduling strategy, SPS-883-R2, was developed. Key parameters for this strategy include a ±12% tolerance window for PV output fluctuations, a maximum charge rate for energy storage units adjusted to 0.8C, and a depth of discharge limit of 85% of rated capacity. The solution's data structure uses binary encoding, with a total length of 128 bits.

[0106] The module initiates the policy fusion process. The initial optimal scheduling strategy SPS-770 (output from Example 2) and SPS-883-R2 are parsed in parallel to extract the control instruction sets for both schemes. The fusion process utilizes a weighted parameter hybrid mechanism: the charging rate parameter is weighted 40% by SPS-770 and 60% by SPS-883-R2, calculated as: final value = 0.4 × original value + 0.6 × new value. A priority retention rule is applied to the photovoltaic fluctuation window parameters, selecting configuration items with a wider tolerance range. At the protocol stack level, the communication frame structure of the two strategies is reorganized into a unified data packet, with the packet header marked with a policy fusion identifier. The resulting enhanced scheduling strategy SPS-F901 enters the ready state and contains 34 control fields and 16 monitoring flags. All operation logs are written to non-volatile memory with millisecond-level timestamp accuracy.

[0107] The final scheduling execution and feedback module loads the SPS-F901 strategy package. The parser deconstructs the control instructions, the photovoltaic power generation unit receives the output curve adjustment instructions, and the energy storage unit simultaneously obtains the charge and discharge schedule. Field equipment executes the instructions according to the sequence, generating a periodic scheduling data set and feeding it back to the data center. The entire process is completed within a single scheduling cycle, with a total time of less than 500 milliseconds from forecast initiation to strategy execution. The system reinitializes the strategy parameters during the next acquisition cycle and clears the temporary buffer data. The enhanced strategy's performance parameters are appended to the historical strategy library for subsequent strategy generation and recall. The second calibrated strategy library is automatically archived after the strategy is output, with a storage period set to 30 days.

[0108] Example 5: After the scheduling execution and feedback module completes the integrated scheduling control, a periodic scheduling data set is generated. The data set contains photovoltaic power generation unit output trajectory, energy storage unit charging and discharging state sequence, scheduling instruction execution log and device response timestamp. The module starts the data transmission protocol, encapsulates the complete data set as a data packet stream, and transmits it to the distributed storage port through the gigabit Ethernet interface. The data packet uses a block transmission mechanism, each data block is attached with a CRC check code and a serial number identifier, and the port receiving end performs real-time verification and sequence recombination.

[0109] The distributed storage port activates the node state monitoring program and synchronously calls all connected storage nodes to run the state parameters. The memory utilization rate parameter is collected by the node agent program, and the value is refreshed in real time. The current used memory of each node accounts for the percentage of the total allocated memory. The transmission delay parameter is measured by the port test subsystem: the port sends a standard probe packet to each node, records the difference between the sending timestamp and the response receiving timestamp, and calculates the average delay value of 20 consecutive probes. The data read / write rate parameter is provided by the node-embedded performance monitor, which counts the number of data bytes processed per second in the last 5 minutes. All parameters are sampled and updated every 200 milliseconds to form a dynamic parameter table residing in the port cache area.

[0110] The port analysis engine loads the parameter table data and starts the weighted evaluation operation. The memory utilization rate parameter is normalized to the [0, 1] interval value, and the conversion coefficient is set to 0.45. The transmission delay parameter is processed using an inverse proportional function. The original delay value is first taken as the inverse and then multiplied by the scaling factor 100. The data read / write rate parameter directly uses the MB / s unit value, and a linear scaler is used to match the evaluation system. The three parameters are multiplied by independent weight coefficients: the memory utilization rate weight value is 0.4, the transmission delay weight value is 0.3, and the data read / write rate weight value is 0.3. The comprehensive performance score of each storage node is calculated by the weighted sum formula, and the score range is standardized to the percentage system. The evaluation operation is performed before each batch of data set transmission, and a score ranking list of all nodes is generated.

[0111] According to the score ranking result, the data storage strategy is executed. The top three nodes with the highest comprehensive scores are marked as the high-performance node group, and are assigned the data core block storage task. The five nodes in the middle segment are listed as the standard node group, and store the benchmark monitoring data. The remaining low-score nodes are used as the disaster recovery backup group, and only store data check copies. The partition storage logic is implemented at the data packet level: the original data set is divided by the port parser according to the data structure, the photovoltaic output trajectory is assigned to the high-performance node, the device response log is assigned to the standard node, and the check copy is distributed to all nodes. The target node assignment instruction is sent to each node controller through the instruction channel.

[0112] After receiving the storage command, the node controller mounts the data storage partition. High-performance nodes enable RAID-5 disk array mode and initiate parallel write channels upon receiving PV output trajectory data packets. Data blocks are split into 128KB fragments and written synchronously to multiple physical disks. Standard nodes use a conventional file system. Device response log data packets are written to the end of the log file in append mode, and the file system records the updated index table. The nodes in the disaster recovery group perform data mirroring, compressing the checksum replica data blocks and storing them in dedicated backup sectors using the LZ77 dictionary encoding algorithm. After all nodes complete the write, they return an operation status code to the storage port. The status code includes a storage address pointer and a data block checksum.

[0113] The distributed storage port aggregates the status information returned by each node to generate a global storage mapping table. This table records key information items: the set of node address pointers corresponding to the PV output trajectory data block, the node path address of the device response log, and the root node value of the checksum replica hash tree. The mapping table is synchronized to the scheduling execution and feedback module via the data bus and written to the port's local non-volatile memory. The module calls the data parser to process the raw scheduling data set and outputs a final scheduling result report with a storage location index. The data access interface supports fast retrieval by index entry, and the retrieval process directly accesses the physical storage location through node address pointers.

[0114] The transmission link layer establishes a persistent connection mechanism. Heartbeat detection is maintained between the port master controller and the storage node, and status data packets are exchanged every 1.5 seconds. When a node is offline for more than three consecutive heartbeat cycles, the storage redistribution process is automatically triggered: the original storage data block of the node is redistributed to the online node after hash verification. The node performance score is reset to zero every 24 hours, and the historical records are stored in an independent monitoring database for system auditing. The data packet transmission rate dynamically adapts to the network status, and the transmission window size is automatically adjusted as the network delay changes. The upper limit of the completion time of the entire storage process is set to 800 milliseconds, and the node that does not respond after the timeout is marked as an abnormal state. The storage operation log records detailed process data, including parameters such as data packet size, node selection timestamp, and write duration. The log file is regularly transferred to the remote backup center.

[0115] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0116] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An integrated photovoltaic and storage intelligent scheduling system, characterized in that: The system comprises: Photovoltaic and energy storage data monitoring module, used to perform real-time data capture of the sensor array, and obtain the power generation data series, energy storage status data series, and environmental parameter data series for the photovoltaic power generation unit and energy storage unit within a preset historical period; a scheduling stability evaluation module, configured to globally traverse the power generation data sequence, the energy storage state data sequence, and the environmental parameter data sequence to perform data volatility evaluation and determine a power generation stability index, an energy storage state stability index, and an environmental parameter stability index; A scheduling strategy generation module is used to perform an intensive search operation in a preset scheduling strategy configuration library using the power generation stability index, energy storage state stability index, and environmental parameter stability index as input indexes to identify the optimal scheduling strategy; A scheduling execution and feedback module is used to implement integrated scheduling control of photovoltaic power generation units and energy storage units according to a preset scheduling task set based on the optimal scheduling strategy, generate a periodic scheduling data set, parse and process the periodic scheduling data set, and output a final scheduling result; The system further comprises: A trend prediction module is used to perform time series prediction processing on the power generation stability index, the energy storage state stability index and the environmental parameter stability index to generate a predicted power generation stability index, a predicted energy storage state stability index and a predicted environmental parameter stability index; A scheduling compensation module is used to perform compensation optimization on the optimal scheduling strategy based on the predicted power generation stability index, the predicted energy storage state stability index, and the predicted environmental parameter stability index to generate an enhanced scheduling strategy; The scheduling execution and feedback module is further used to implement scheduling control based on the enhanced scheduling strategy; The scheduling compensation module is also used for: Using the predicted power generation stability index, the predicted energy storage state stability index, and the predicted environmental parameter stability index as input indexes, performing an association matching operation in the scheduling strategy configuration library to generate a second calibration strategy library; Identifying a strategy feature triggering interval in the second calibration strategy library to obtain a second strategy feature space; Iteratively generate a strategy based on the preset fitness constraint and the second strategy feature space to form a second scheduling strategy solution; The second scheduling strategy solution is integrated and optimized with the optimal scheduling strategy to obtain the enhanced scheduling strategy.

2. The photovoltaic-storage integrated intelligent scheduling system according to claim 1, characterized in that: The scheduling stability evaluation module is also used to: Performing data cleaning on the power generation data sequence using outlier detection technology to generate a cleaned power generation data set; The outlier detection technology identifies and removes data points that fall outside a preset fluctuation range, which is dynamically adjusted based on the statistical characteristics of the data distribution. performing a standard deviation calculation on the cleaning power generation data set to obtain the power generation stability index; Similar volatility assessment processing is performed on the energy storage state data sequence and the environmental parameter data sequence to obtain the energy storage state stability index and the environmental parameter stability index respectively.

3. The photovoltaic-storage integrated intelligent scheduling system according to claim 2, characterized in that: The scheduling stability evaluation module is also used to: Rearranging the generated power data sequence in ascending order of values, and selecting a data point at an intermediate value as a baseline reference line; Calculating the upper and lower quartiles of the power generation data sequence; The upper and lower quartiles are used as boundary constraints and combined with the baseline reference line to construct a data distribution framework.

4. The photovoltaic-storage integrated intelligent scheduling system according to claim 1, characterized in that: The scheduling strategy generation module is also used for: Acquire multiple sample power generation stability indicators, multiple sample energy storage state stability indicators, multiple sample environmental parameter stability indicators, and corresponding multiple sample scheduling strategies as training data; Constructing a multidimensional feature space, wherein the coordinate origin of the multidimensional feature space is set as a reference origin, the first axis represents the power generation stability index dimension, the second axis represents the energy storage state stability index dimension, and the third axis represents the environmental parameter stability index dimension; The training data is mapped into the multidimensional feature space to form a plurality of sample space coordinate points, and the plurality of sample space coordinate points are marked using the plurality of sample scheduling strategies to generate the scheduling strategy configuration library.

5. The photovoltaic-storage integrated intelligent scheduling system according to claim 4, characterized in that: The scheduling strategy generation module is also used for: A first reference plane passing through the power generation stability indicator and parallel to a plane formed by the second axis and the third axis is defined in the scheduling strategy configuration library; A second reference plane is defined in the scheduling strategy configuration library, passing through the energy storage state stability index and being parallel to a plane formed by the first axis and the third axis; A third reference plane is defined in the scheduling strategy configuration library, which passes through the environmental parameter stability index and is parallel to the plane formed by the first axis and the second axis; The area defined by the plane formed by the second axis and the third axis, the plane formed by the first axis and the third axis, the plane formed by the first axis and the second axis, and the first reference plane, the second reference plane, and the third reference plane is used as a strategy subspace, wherein the strategy subspace includes a plurality of strategy sample points; An intensive search operation is performed on the multiple policy sample points to locate a target policy sample point, and the sample scheduling policy corresponding to the target policy sample point is used as the optimal scheduling policy.

6. The photovoltaic-storage integrated intelligent scheduling system according to claim 5, characterized in that: The scheduling strategy generation module is also used for: Extracting a central strategy sample point of the strategy subspace, taking the central strategy sample point as a starting position, and constructing a central neighborhood range according to a preset dense search radius; Counting the total number of strategy sample points within the central neighborhood, and dividing the total number by the volume of the central neighborhood to obtain a central neighborhood density value; Randomly select a strategic sample point from the edge of the central neighborhood range as the first retrieval point, and calculate the neighborhood density value of the first retrieval point; Determine whether the neighborhood density value of the first search point is greater than or equal to the center neighborhood density value. If so, update the first search point as the starting position, continue to perform intensive search operations until a preset number of search iterations is reached, and use the final search point as the target strategy sample point.

7. The photovoltaic-storage integrated intelligent scheduling system according to claim 6, characterized in that: The scheduling strategy generation module is also used for: If the neighborhood density value of the first search point is less than the center neighborhood density value, the search failure counter whose initial value is zero is incremented, and another strategic sample point is randomly selected from the edge of the center neighborhood range as a new first search point for intensive search analysis; When the value of the retrieval failure counter exceeds a preset maximum failure threshold, the central strategy sample point is used as the target strategy sample point.

8. The photovoltaic-storage integrated intelligent dispatching system according to claim 1, characterized in that: The scheduling execution and feedback module is also used to: Transmitting the periodically scheduled data set to a distributed storage port; The distributed storage port synchronously calls the operating status parameters of each storage node for analysis and processing, and stores the periodically scheduled data set partitions to the target storage node according to the analysis results; The operating status parameters include memory utilization parameters, transmission delay parameters, and data read and write rate parameters of the storage node.

Images