Power grid energy scheduling method and system based on multi-source data fusion

Abstract

The application discloses a kind of power grid energy scheduling method and system based on multi-source data fusion, it is related to power grid energy scheduling technical field.The method comprises: S1, obtains the water and electricity environment data, photovoltaic environment data and load data of target power grid area;S2, generates water and electricity output constraint condition and photovoltaic efficiency correction condition;S3, time alignment is carried out to the water and electricity output data and photovoltaic output data after pre-processing by dynamic time warping algorithm, generates the fusion feature vector containing environmental constraint;S4, with the fusion feature vector as input, combined with the water and electricity output constraint condition and photovoltaic efficiency correction condition, construct multi-objective optimization model, obtain the scheduling scheme containing water and electricity output strategy, photovoltaic output strategy and thermal power peak shaving strategy by using rolling optimization algorithm.The present application is aimed at the winter dry season power grid dispatching problem in high-altitude mountainous area, through multi-source data fusion and dynamic coupling model, the precision of energy scheduling under environmental constraint is significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid energy scheduling, in particular to a power grid energy scheduling method and system based on multi-source data fusion. BACKGROUND

[0002] In the high-altitude mountainous power grid, the winter dry season faces unique energy scheduling challenges: river ice jam causes the water passing capacity of hydropower stations to decrease significantly, and the light transmittance of photovoltaic components is reduced due to morning fog weather, both of which superimpose to cause the mismatch between energy supply side fluctuation and load side heating load pulse demand. Such power grid has the characteristics of high proportion of hydropower and access of distributed photovoltaic, and the winter heating load is affected by climate and presents high-frequency fluctuation characteristics.

[0003] The prior art does not establish a real-time coupling mechanism of river ice jam degree and hydropower output, and morning fog parameters and photovoltaic efficiency, resulting in that the scheduling model cannot accurately reflect the actual energy supply capacity under environmental constraints. Specifically, when the hydropower output suddenly drops due to ice jam, the model cannot dynamically adjust the available capacity boundary; when the photovoltaic efficiency is reduced due to the influence of morning fog, there is a lack of real-time correction of its actual output capacity. The above defects cause the system to rely on frequent peak regulation of thermal power to balance supply and demand, thereby causing problems such as frequent start-stop of thermal power equipment, over-limit of power grid voltage fluctuation, and seriously affecting the economy and reliability of power grid operation. SUMMARY

[0004] The purpose of the present application is to provide a power grid energy scheduling method and system based on multi-source data fusion to solve the problems raised in the background art.

[0005] To achieve the above purpose, the present application provides the following technical solution: a power grid energy scheduling method based on multi-source data fusion, comprising the following steps:

[0006] S1, obtaining hydropower environmental data, photovoltaic environmental data and load data of a target power grid area, the hydropower environmental data including an ice jam degree parameter for reflecting the degree of river ice jam, the photovoltaic environmental data including morning fog light transmittance and morning fog duration for reflecting the influence of morning fog on photovoltaic components, and the load data including a load pulse signal of electric heating equipment;

[0007] S2, constructing a hydropower output attenuation model based on the ice jam degree parameter, constructing a photovoltaic efficiency inhibition model based on the morning fog light transmittance and the morning fog duration, and generating a hydropower output constraint condition and a photovoltaic efficiency correction condition, respectively;

[0008] S3, the water power output data and photovoltaic output data are preprocessed by applying the water power output constraint condition and photovoltaic efficiency correction condition, and the preprocessed water power output data and photovoltaic output data are time-aligned by a dynamic time warping algorithm to generate a fusion feature vector containing environmental constraints;

[0009] S4, taking the fusion feature vector as input, combining the water power output constraint condition and photovoltaic efficiency correction condition to construct a multi-objective optimization model, and solving to obtain a scheduling scheme containing water power output strategy, photovoltaic output strategy and thermal power peak shaving strategy by using a rolling optimization algorithm.

[0010] Preferably, the water environment data acquisition in step S1 comprises:

[0011] The ice jam thickness is obtained by deploying an ultrasonic sensor network in the river channel, and the ice jam degree parameter is determined based on the ice jam thickness;

[0012] The acquisition of the photovoltaic environmental data comprises:

[0013] The morning mist transmittance and duration are obtained by sensors on the surface of the photovoltaic module;

[0014] The load data acquisition comprises:

[0015] The start-stop signal and power fluctuation data of the electric heating equipment are obtained by the intelligent electric meter in the transformer area.

[0016] Preferably, in step S2, the water power output attenuation model is represented as:

[0017]

[0018] In the formula, is the ice jam degree parameter, is a function fitted based on the fluid mechanics characteristics of the river channel, used to represent the constraint of the ice jam degree on the upper limit of the water power available output;

[0019] The photovoltaic efficiency inhibition model is represented as:

[0020]

[0021] In the formula, is the morning mist transmittance, is the morning mist duration, and is a function established based on light scattering theory and module temperature rise characteristics, used to represent the correction of the morning mist on the photovoltaic power generation efficiency.

[0022] Preferably, in step S3, the preprocessing of the water power output data is specifically:

[0023]

[0024] wherein, is the original water power output data, is the upper limit of water power available output output by the water power output attenuation model;

[0025] The preprocessing of the photovoltaic output data specifically includes:

[0026]

[0027] wherein, is the photovoltaic output prediction data, is the photovoltaic efficiency correction factor output by the photovoltaic efficiency suppression model.

[0028] Preferably, the dynamic time warping algorithm in the step S3 specifically includes:

[0029] A space-time cost function is constructed:

[0030]

[0031] wherein, is the time deviation weight, is the power deviation weight, are respectively the sampling times of the water power output data and the photovoltaic output data, are respectively the preprocessed water power output data and the photovoltaic output data;

[0032] Based on the space-time cost function, an optimal alignment path of the water power output data and the photovoltaic output data is solved, so that the time error does not exceed a preset threshold, and a time-aligned joint output sequence is generated;

[0033] The joint output sequence is spliced with the load data to obtain the fusion feature vector.

[0034] Preferably, the multi-objective optimization model in the step S4 takes the thermal power peak shaving cost and the grid voltage deviation as the optimization objectives, and the objective function is expressed as:

[0035] The thermal power peak shaving cost and the weighted voltage deviation square sum are minimized, that is:

[0036]

[0037] wherein, is the thermal power peak shaving cost, is the voltage quality weight, is the node voltage deviation;

[0038] The constraint conditions of the multi-objective optimization model include:

[0039]

[0040] In the formula, is the water power output, is the photovoltaic power output, is the photovoltaic rated power output, is the thermal power output, is the real-time load, is the rated load, is the power balance allowable error.

[0041] Preferably, further comprising:

[0042] Comparing the actual execution data of the scheduling scheme with the predicted data, if the deviation exceeds a preset threshold, the parameters of the water power output attenuation model and the photovoltaic efficiency suppression model are adaptively corrected;

[0043] The parameters include a river channel characteristic coefficient and an atmospheric attenuation coefficient.

[0044] The application further provides a power grid energy scheduling system based on multi-source data fusion, comprising:

[0045] An edge computing layer is configured to acquire water environment data, photovoltaic environment data and load data of a target power grid area, the water environment data comprising an ice jam degree parameter, the photovoltaic environment data comprising a morning mist transmittance and a morning mist duration, and the load data comprising a load pulse signal of an electric heating device;

[0046] The edge computing layer is further configured to generate a water power output constraint condition and a photovoltaic efficiency correction condition based on the ice jam degree parameter and the morning mist parameter, to preprocess and time-align water power output data and photovoltaic power output data, and to generate a fusion feature vector;

[0047] A cloud decision layer is configured to receive the fusion feature vector, to construct a multi-objective optimization model comprising the water power output constraint condition and the photovoltaic efficiency correction condition, and to solve the scheduling scheme by using a rolling optimization algorithm.

[0048] A communication control layer is configured to realize data interaction between the edge computing layer and the cloud decision layer, and to send the scheduling scheme to a power generation device for execution.

[0049] The application further provides an electronic device, which is a physical device, and comprises:

[0050] A processor and a memory are communicatively connected to each other;

[0051] The memory is configured to store executable instructions executed by the processor, and the processor is configured to execute the executable instructions to implement the power grid energy scheduling method and system based on multi-source data fusion as described above.

[0052] The application further provides a computer readable storage medium, wherein a computer program is stored in the computer readable storage medium, and the computer program is executed by a processor to realize the power grid energy scheduling method and system based on multi-source data fusion.

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

[0054] By quantifying the physical correlation between the ice plug degree, the water and electricity output, the morning fog parameter and the photovoltaic efficiency, the problem of insufficient response to regional characteristics of the traditional model is solved; by using the cross-time scale data alignment technology, the synchronization deviation between the minute-level water and electricity data and the second-level photovoltaic data is eliminated, so that the multi-objective optimization model can reflect the actual supply capacity in real time; by embedding the rolling optimization algorithm of the environmental constraint, the frequent start-stop of the thermal power caused by the load fluctuation is effectively reduced, the equipment loss is reduced and the power quality is improved; the closed-loop feedback mechanism continuously optimizes the model parameters, forms adaptive scheduling capability, and overall enhances the reliability and economy of the power grid operation in special scenarios, the application aims at the power grid scheduling problem in the winter dry season of high-altitude mountainous areas, and through the multi-source data fusion and dynamic coupling model, the energy scheduling precision under the environmental constraint is significantly improved. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A main flowchart of the power grid energy scheduling method based on multi-source data fusion provided by the embodiment of the application is provided.

[0056] Figure 2 A structural schematic diagram of the power grid energy scheduling system based on multi-source data fusion provided by the embodiment of the application is provided.

[0057] Figure 3 A structural schematic diagram of an electronic device provided by the embodiment of the application is provided. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0059] The execution subject of the method in the embodiment is a terminal, which can be a mobile phone, a tablet computer, a palm computer PDA, a notebook computer or a desktop computer, and of course, can also be other devices with similar functions, which are not limited in the embodiment.

[0060] Please refer to Figure 1The application provides a power grid energy scheduling method and system based on multi-source data fusion.

[0061] In step S1, water and electricity environment data, photovoltaic environment data and load data of a target power grid region are acquired, the water and electricity environment data include an ice jam degree parameter for reflecting the degree of river ice jam, the photovoltaic environment data include morning mist light transmittance and morning mist duration for reflecting the influence of morning mist on photovoltaic components, and the load data include a load pulse signal of an electric heating device.

[0062] Specifically, in step S1, a three-dimensional data system with time-space coupling characteristics is constructed based on a geographic information system (GIS), a meteorological monitoring network and an intelligent electric meter acquisition system. Firstly, spatial data such as topography, power transmission line direction and distribution of transformer substations are accurately collected by using the GIS system to establish a high-precision geographic space model. The meteorological monitoring network in the coverage area is simultaneously accessed to acquire meteorological parameters such as air temperature, humidity, wind speed and sunshine duration in real time, and high-frequency monitoring is performed on extreme weather such as freezing rain and heavy fog that easily occurs in winter. Meanwhile, user-side power consumption data are collected by relying on the intelligent electric meter acquisition system, and information such as start / closure time of the electric heating device and load fluctuation curve is captured.

[0063] After the original data are acquired, statistical methods and machine learning algorithms are used for data cleaning to remove abnormal values and repeated data. Through technical means such as wavelet transform and principal component analysis (PCA), multi-dimensional data are subjected to feature extraction and dimension reduction processing, and finally, core indexes such as the ice jam degree parameter (based on comprehensive evaluation of meteorological and geographic data to evaluate the icing risk of the power transmission line), the morning mist light transmittance and duration (a prediction model is constructed by combining meteorological monitoring and visibility data) and the electric heating load pulse signal (the load peak value characteristics are identified by time series analysis of user power consumption data) are output as basic data sources for subsequent physical effect quantification. The specific action mechanism of each dimension data is as follows:

[0064] The water and electricity environment dimension: through fusion of satellite remote sensing images and river channel sensor data, an ice jam degree quantification model is constructed. The model analyzes the ice jam distribution characteristics by using a gray level co-occurrence matrix, calculates the river flow capacity attenuation coefficient by combining fluid mechanics equations, and directly inputs the ice jam degree parameter as a key driving variable into a water and electricity output attenuation model to realize dynamic evaluation of the influence of river icing in winter on the water and electricity generation capacity.

[0065] Photovoltaic environmental dimension: Based on the monitoring data of visibility, humidity and solar radiation intensity of meteorological stations, a spatio-temporal prediction model of morning mist generation and dissipation is established. The frequency domain characteristics of morning mist transmittance are analyzed by Fourier transform. Combined with the light-to-electricity conversion efficiency curve of photovoltaic modules, an efficiency inhibition model under low light conditions is constructed. The morning mist parameters are not only used for photovoltaic power prediction, but also used as feedback variables for model parameter optimization to improve prediction accuracy.

[0066] Load dimension: Using minute-level load data collected by smart meters, the pulse signal characteristics of electric heating equipment are extracted by wavelet transform, and a load prediction model based on long short-term memory network (LSTM) is established. The temperature-load response curve is analyzed to identify the load fluctuation law in peak and valley periods. These characteristic data are used as dynamic boundary conditions for power balance constraints to support the source-load collaborative scheduling optimization model.

[0067] Step S2, constructing a water and electricity output attenuation model based on the ice plug degree parameter, constructing a photovoltaic efficiency inhibition model based on the morning mist transmittance and the morning mist duration, and generating water and electricity output constraint conditions and photovoltaic efficiency correction conditions respectively.

[0068] Specifically, in the step S2, the water and electricity output attenuation model is represented as:

[0069]

[0070] In the formula, is the ice plug degree parameter, is a function fitted based on the fluid mechanics characteristics of the river channel, used to represent the constraint of the ice plug degree on the upper limit of the available output of the water and electricity;

[0071] Optionally, a high-precision ultrasonic sensor network deployed at key sections of the river channel is used to collect the ice plug degree parameter (unit: mm) in real time. The sensor network is distributedly deployed, with one monitoring point set every 500 meters upstream of the hydropower station, and each monitoring point equipped with 3 groups of ultrasonic probes at different angles. The triangular positioning algorithm is used to eliminate the measurement blind area, ensuring the accuracy and timeliness of the ice plug thickness data collection.

[0072] Further, in the modeling of hydraulic characteristics, first, based on the open channel flow theory in fluid mechanics, a nonlinear relationship model of ice plug thickness and water area is established , where is the initial water area of the river channel without ice plug, which is accurately obtained through historical hydrological data combined with laser radar terrain scanning data; is the river channel characteristic coefficient, which is determined by curve fitting of historical ice conditions data using the least squares method according to parameters such as river slope and roughness;

[0073] In the output constraint derivation, the water area change is converted into the available hydropower output constraint by combining the water turbine efficiency characteristic curve with the Bernoulli equation, considering the influence of ice jam on water flow velocity and head loss, and introducing a correction coefficient Finally, the upper limit of hydropower output derivation formula is established:

[0074]

[0075] In the formula, is the water turbine efficiency, is the water density, is the acceleration of gravity, is the average flow velocity, is a correction factor that changes dynamically with the thickness of the ice jam, calibrated by laboratory simulation and field measurement data;

[0076] Further, through the above calculation process, the ice jam constraint upper limit sequence of hydropower output containing time stamp is generated This sequence not only considers the immediate impact of the current ice jam state, but also predicts the development trend of the ice jam within the next 3 hours through a rolling prediction model. The output constraint boundary data is updated at 15-minute intervals, providing dynamic constraint conditions for the hydropower data preprocessing in step S3, ensuring the safety and economy of hydropower dispatching decisions.

[0077] Specifically, in step S2, in the morning mist environment, the power generation efficiency of photovoltaic modules will fluctuate significantly due to light intensity attenuation and component surface temperature variation. Therefore, a photovoltaic efficiency suppression model is constructed, which is expressed as:

[0078]

[0079] In the formula, is the transmittance of the morning mist, is the duration of the morning mist, which is a function established based on light scattering theory and component temperature rise characteristics, is used to represent the correction of the morning mist on the photovoltaic power generation efficiency.

[0080] Optionally, by deploying high-precision sensors on the surface of photovoltaic modules, the transmittance of the morning mist (unit: %) and the duration (unit: hours) are collected in real time, where the transmittance is obtained by measuring the ratio of atmospheric transmitted light intensity to standard light intensity with a multispectral radiometer, and the duration adopts an event triggering mechanism, which starts timing when the transmittance continuously falls below 90% and stops when it rises above the threshold.

[0081] Further, based on Mie scattering theory, a dynamic light intensity attenuation model of morning fog particles is established wherein, is the standard light intensity (1000W / m2), , as the atmospheric attenuation coefficient, is calibrated by combining historical meteorological data and actual power generation data (typical value range 0.05-0.15 ). At the same time, considering the effect of temperature drop of component surface caused by morning fog condensation, a temperature rise delay compensation formula is introduced: each millimeter of fog droplet thickness will cause a linear decrease of 0.5% in photovoltaic efficiency, and this coefficient is verified by laboratory fog chamber environment test. Finally, by combining the light intensity attenuation model and the temperature drop correction formula, the photovoltaic efficiency correction factor is obtained by iterative calculation.

[0082] Further, the photovoltaic efficiency correction factor with morning fog constraint is output , which is embedded as a dynamic constraint condition in the photovoltaic data preprocessing module of step S3. When calculating the real-time power generation of photovoltaic modules, the power limitation is realized by , wherein is the nominal power in the fog-free environment, and this correction factor can effectively reduce the power prediction error caused by morning fog to within 3%.

[0083] In this embodiment, the generated and constitute the core constraint conditions of step S3 feature preprocessing, so that the original energy output data is converted into an effective supply capacity sequence considering the actual environmental constraints, providing a physically meaningful boundary limit for cross-time scale data coupling.

[0084] Step S3, the hydroelectric power output data and the photovoltaic output data are preprocessed by applying the hydroelectric power output constraint condition and the photovoltaic efficiency correction condition, and the preprocessed hydroelectric power output data and photovoltaic output data are time-aligned by the dynamic time warping algorithm to generate a fusion feature vector containing environmental constraints.

[0085] Specifically, in step S3, the actual output of hydroelectric power and photovoltaic power is affected by multiple factors such as natural conditions, equipment parameters and grid operation constraints, so the original data needs to be corrected to improve the accuracy of the scheduling model.

[0086] wherein, the minute-level hydroelectric power output raw data represents the theoretical generating capacity, but in actual operation, constraints such as the maximum flow rate of the water turbine, the reservoir water level limit, etc. need to be considered. By applying constraints, the original data is corrected to the actual feasible output. The mathematical expression is:

[0087]

[0088] In the formula, is the original data of water and electricity output, is the upper limit of water and electricity available output output by the water and electricity output attenuation model, which is determined by the technical parameters of the water and electricity unit, the prediction of the basin inflow, and the stable operation requirements of the power grid. This operation can effectively avoid the failure of the dispatching scheme caused by overestimating the water and electricity supply.

[0089] Further, the photovoltaic output prediction data Based on the ideal illumination condition calculation, the actual power generation is affected by factors such as cloud cover, component conversion efficiency attenuation, etc. Introducing as a correction coefficient, the coefficient includes a temperature reduction coefficient, a dust loss coefficient, and a device aging factor. Data calibration is realized through the following formula:

[0090]

[0091] In the formula, is the photovoltaic output prediction data, is the photovoltaic efficiency correction factor output by the photovoltaic efficiency suppression model. This operation can convert the theoretical prediction power into more actual dispatchable power.

[0092] In this embodiment, by constructing an environmental constraint condition model, the water and electricity and photovoltaic theoretical output data are dynamically corrected and converted into actual available power that meets the real-time environmental restrictions. This method effectively avoids the overestimation risk of the dispatching model to the energy supply capacity, significantly improves the feasibility and safety of the power grid energy dispatching scheme, and provides protection for reliable power supply under complex environment.

[0093] Specifically, in the process of power grid multi-source data fusion, due to the differences in data collection frequency and characteristics of water and electricity and photovoltaic, a quantitative index that can measure the temporal and spatial differences of data needs to be established. For this purpose, a time-space cost function is constructed:

[0094]

[0095] In the formula, is the time deviation weight, is the power deviation weight. These weight parameters are not fixed values, but are determined by training a large amount of historical ice plug-morning fog scene data. Special weather conditions such as ice plug and morning fog can have a significant impact on water and electricity and photovoltaic power generation. Training the weight using such scene data can make the cost function more suitable for the data alignment needs under complex working conditions, are the sampling times of the water and electricity output data and the photovoltaic output data, respectively, These are the preprocessed hydropower output data and photovoltaic output data, respectively. The cost function, through weighted summation, comprehensively considers the differences in the time and power dimensions, providing a quantitative basis for subsequent data alignment.

[0096] Furthermore, an improved DTW (Dynamic Time Warping) algorithm is adopted:

[0097] Considering that the data acquisition interval for hydropower is 1 minute while that for photovoltaic data is only 10 seconds, this significant difference in sampling frequency can lead to data asynchrony, severely affecting the subsequent generation and analysis of joint features. To address this issue, an improved DTW algorithm is adopted. This algorithm, based on the traditional DTW, allows for an offset of ±3 time steps to accommodate the difference in the two data acquisition frequencies. By solving for the optimal alignment path between hydropower and photovoltaic data using this algorithm, the time error can be controlled within ≤90 seconds, while reducing the power deviation variance by 40%. This optimization effectively solves the data asynchrony problem caused by the difference in sampling frequencies of different sensors, providing accurate time-aligned foundational data for subsequent joint feature generation, and ensuring reliable data support for joint analysis and scheduling decisions based on hydropower and photovoltaic data.

[0098] Furthermore, based on the spatiotemporal cost function, the optimal alignment path between hydropower output data and photovoltaic output data is solved to ensure that the time error does not exceed a preset threshold, thereby generating a time-aligned joint output sequence; the joint output sequence is then concatenated with the load data to obtain the fused feature vector.

[0099] Optionally, in power grid energy dispatch, the impact of environmental factors on energy supply and demand cannot be ignored. The time-aligned hydropower-photovoltaic adjustment sequence is then deeply integrated with load characteristics (including the frequency and amplitude of pulse signals characterizing electric heating electricity consumption). Specifically, a fused feature vector is constructed using data stitching technology.

[0100]

[0101] in, This represents the adjustment of hydropower output after environmental constraints have been corrected, and its numerical change reflects the limitations of hydrological environment such as river ice blockage on hydropower supply. The photovoltaic output adjustment sequence, which takes into account the impact of meteorological conditions such as morning fog, reflects the dynamic effect of changes in the solar environment on new energy power generation. By using the frequency and amplitude of the electric heating pulse signal, the fluctuation characteristics of electricity load caused by residential heating on the demand side can be accurately characterized.

[0102] It can be understood that the output of the fusion feature vector realizes the organic integration of supply side constraints and demand side fluctuations, forming a complete input data set. The data set can be directly used as the state variable of the optimization model in step S4, providing a comprehensive and accurate data basis for the subsequent development of scheduling strategies.

[0103] Further, in the data preprocessing stage, the environmental constraint information extracted in step S2 is introduced to systematically correct the original hydroelectric, photovoltaic and load data. This correction mechanism can effectively filter out abnormal data caused by environmental factors, ensuring that the data entering the coupling algorithm truly reflects the actual operating environment constraints. The generated fusion feature vector serves as the core input of the optimization model in step S4, taking into account the dynamic influence of environmental factors such as ice jam and morning fog from the data source. Compared to traditional scheduling models that have decision-making bias problems due to insufficient data processing, this method can significantly improve the scientificity and accuracy of the scheduling strategy, making the power grid energy scheduling more in line with the actual operating scenario, and enhancing the reliability and stability of the scheduling decision.

[0104] In this embodiment, by time-aligning and power-correcting hydroelectric and photovoltaic data, the problem of data asynchronization and prediction deviation is effectively solved. On this basis, the characteristics of electric heating load are integrated, which not only comprehensively reflects the environmental constraints of the supply side, but also accurately captures the dynamic changes of the demand side. The construction of this fusion feature vector provides complete and reliable input data for the subsequent rolling optimization scheduling model, enabling the scheduling strategy to achieve precise regulation and control of source-load collaboration under complex and variable environmental conditions.

[0105] Step S4, taking the fusion feature vector as input, combining the hydroelectric power output constraint condition and photovoltaic efficiency correction condition to construct a multi-objective optimization model, and using a rolling optimization algorithm to obtain a scheduling scheme containing hydroelectric power output strategy, photovoltaic power output strategy and thermal power peak shaving strategy.

[0106] Specifically, the multi-objective optimization model in step S4 takes the thermal power peak shaving cost and grid voltage deviation as the optimization objective. This model uses a rolling optimization mechanism to dynamically adjust the scheduling scheme at fixed time intervals, ensuring the continuous output of optimal scheduling strategies in complex and variable grid operating environments. The specific objective function is represented as:

[0107] Minimize the thermal power peak shaving cost and the weighted sum of voltage deviation squares, i.e.

[0108]

[0109] In the formula, For the cost of thermal power peak regulation, the single start-stop penalty term is directly related to the frequency characteristics of the load pulse signal, that is, when the load fluctuation is frequent, the number of start-stop of the thermal power equipment increases, and the penalty coefficient will be significantly improved to quantify the equipment wear cost, for example, in the period of extremely cold weather leading to the surge of electric heating load, if the thermal power equipment is frequently started and stopped;

[0110] For the weight for balancing the cost of thermal power and voltage deviation, it can be dynamically adjusted according to the actual demand of power grid operation;

[0111] For the node voltage deviation, the voltage deviation is calculated based on the power data in the fusion feature vector of step S3, combined with the grid node parameters (such as line impedance, transformer ratio, etc.), and the voltage deviation is minimized It can effectively avoid the voltage out-of-limit problem caused by uneven power distribution, and ensure the safe and stable operation of the power grid;

[0112] The constraint conditions of the multi-objective optimization model include:

[0113]

[0114] In the formula, is the hydropower output, is a hydropower output prediction model based on step S2, wherein represents the inflow, water level and other hydrological parameters, and this constraint ensures that the hydropower dispatching does not exceed the actual available capacity, avoiding damage to the power generation equipment or waste of water resources due to excessive dispatching;

[0115] is the photovoltaic output, is the rated photovoltaic output, is the thermal power output, is a photovoltaic efficiency correction function based on the morning mist transmittance and the morning mist duration time constructed in step S2, The value range is [0, 1], and the closer the value is to 1, the better the light transmittance; is the total duration of the morning mist from formation to dissipation, in hours, and this function is established by correlation analysis of historical meteorological data and photovoltaic power monitoring data, which can dynamically quantify the influence of morning mist on the power generation efficiency of photovoltaic power stations, and provide more accurate power prediction constraints for energy dispatching;

[0116] is the real-time load, is the rated load, is the power balance allowable error, which requires that the total output of hydropower, photovoltaic and thermal power must match the load demand, and the fluctuation range is controlled within the safety threshold, to ensure the balance of power supply and demand and the stable operation of the power grid.

[0117] Step S5, compare the actual execution data of the scheduling scheme with the predicted data, if the deviation exceeds the preset threshold, adaptively correct the parameters of the water power output attenuation model and the photovoltaic efficiency suppression model;

[0118] The parameters include a river channel characteristic coefficient and an atmospheric attenuation coefficient.

[0119] Specifically, a model predictive control (MPC) framework is adopted, with a 10-minute rolling window, and the following operations are performed every period:

[0120] Input: The feature vector generated by multi-source data fusion in step S3 As the core input, this vector integrates multi-dimensional information such as weather forecasts, load forecasts, and equipment status, and at the same time, the latest set of environmental constraint parameters is accessed synchronously , ensuring the timeliness and completeness of the model input data;

[0121] Processing: For complex optimization problems with nonlinear constraints, an improved particle swarm optimization (IPSO) algorithm is used for solution. This algorithm quickly locates the optimal solution in a multi-dimensional solution space by introducing dynamic inertia weight and adaptive learning factor. The specific outputs include:

[0122] Hydropower output interval According to the reservoir capacity, inflow flow, and power demand, the hydropower output interval is output, providing a flexible adjustment range for real-time scheduling;

[0123] Photovoltaic power limit Combined with cloud coverage prediction and inverter capacity, the photovoltaic power limit value is determined to avoid the risk of light abandonment and equipment overload;

[0124] Thermal power peak shaving strategy Based on system load fluctuations and environmental protection constraints, a thermal power peak shaving strategy is generated to optimize the unit start-stop plan and output curve;

[0125] Feedback: A "prediction-execution-feedback" double closed-loop control mechanism is established to compare the actual execution data (from the closed-loop control module of step 5) with the model prediction value point by point. When the deviation exceeds the 8% threshold, the parameter adaptive correction program is triggered. Taking water power scheduling as an example, the system automatically reversely corrects the fitting coefficient of the ice jam thickness and the water area river channel characteristic coefficient A parameter dynamic correction library is built through historical data mining to continuously optimize the precision of the physical effect model. Ultimately, a complete closed-loop control system of "data collection-constraint generation-intelligent scheduling-feedback correction" is formed.

[0126] It can be understood that the physical constraint of step S2 provides an environmental constraint boundary for the multi-objective optimization model, and the fusion feature of step S3 provides a real-time running state input for the multi-objective optimization model, so that the optimization scheduling result not only conforms to the actual supply capacity under the ice plug / fog environment, but also matches the high-frequency fluctuation characteristics of the electric heating load, and fundamentally solves the problem of frequent start-stop of thermal power caused by the "data disconnection" of the traditional model.

[0127] In the embodiment, by quantifying the physical correlation between the ice plug degree and the water and electricity output, and the fog parameter and the photovoltaic efficiency, the problem of insufficient response to regional characteristics of the traditional model is solved; by using the cross-time scale data alignment technology, the synchronization deviation between the minute-level water and electricity data and the second-level photovoltaic data is eliminated, so that the multi-objective optimization model can reflect the actual supply capacity in real time; by embedding the rolling optimization algorithm of environmental constraints, the frequent start-stop of thermal power caused by load fluctuation is effectively reduced, the equipment loss is reduced, and the power quality is improved; the closed-loop feedback mechanism continuously optimizes the model parameters, forms an adaptive scheduling capability, and overall enhances the reliability and economy of the power grid operation under special scenarios. The present application aims at the power grid scheduling problem in the winter dry season of high-altitude mountainous areas, and significantly improves the precision of energy scheduling under environmental constraints through multi-source data fusion and dynamic coupling model.

[0128] On the basis of the above-mentioned embodiment, as shown in Figure 2 The present application also provides a power grid energy scheduling system based on multi-source data fusion, which is used to support the power grid energy scheduling method based on multi-source data fusion of the above-mentioned embodiment, and comprises:

[0129] An edge computing layer 100 is used to acquire water and electricity environment data, photovoltaic environment data and load data of a target power grid area, the water and electricity environment data comprises an ice plug degree parameter, the photovoltaic environment data comprises a fog transmittance and a fog duration, and the load data comprises a load pulse signal of an electric heating device;

[0130] The edge computing layer 100 is further used to generate a water and electricity output constraint condition and a photovoltaic efficiency correction condition based on the ice plug degree parameter and the fog parameter, to preprocess and time-align the water and electricity output data and the photovoltaic output data, and to generate a fusion feature vector;

[0131] A cloud decision layer 200 is used to receive the fusion feature vector, to construct a multi-objective optimization model comprising the water and electricity output constraint condition and the photovoltaic efficiency correction condition, and to solve the scheduling scheme by using a rolling optimization algorithm;

[0132] A communication control layer 300 is used to realize data interaction between the edge computing layer and the cloud decision layer, and to send the scheduling scheme to a power generation device for execution.

[0133] In the embodiment, by quantifying the physical correlation between the ice plug degree and the water and electricity output, the morning mist parameter and the photovoltaic efficiency, the problem of insufficient response to regional characteristics of the traditional model is solved; by using the cross-time scale data alignment technology, the synchronization deviation of the minute-level hydropower data and the second-level photovoltaic data is eliminated, so that the multi-objective optimization model can reflect the actual supply capacity in real time; by embedding the rolling optimization algorithm of environmental constraints, the frequent start-stop of thermal power caused by load fluctuation is effectively reduced, the equipment loss is reduced and the power quality is improved; the closed-loop feedback mechanism continuously optimizes the model parameters to form adaptive scheduling capability, which overall enhances the reliability and economy of power grid operation in special scenarios. The present application aims at the winter dry season power grid scheduling problem in high-altitude mountainous areas, and significantly improves the energy scheduling accuracy under environmental constraints through multi-source data fusion and dynamic coupling model.

[0134] Further, the power grid energy scheduling system based on multi-source data fusion can run the power grid energy scheduling method based on multi-source data fusion, and specific implementation can be referred to the method embodiments, which will not be described here.

[0135] On the basis of the above embodiments, as Figure 3 The electronic device includes:

[0136] At least one processor 22, at least one memory 21, a communication interface 23 and a communication bus 24, the processor 22 is in communication connection with the memory 21;

[0137] In the embodiment, the memory 21 can be implemented in any appropriate manner, for example: the memory 21 can be a read-only memory, a mechanical hard disk, a solid state disk or a U disk, etc.; the memory 21 is used to store executable instructions executed by the processor;

[0138] In the embodiment, the processor 22 can be implemented in any appropriate manner, for example, the processor 22 can take the form of, for example, a microprocessor or a processor and a computer readable medium storing computer readable program code (such as software or firmware) executable by the (micro) processor, logic gates, switches, application specific integrated circuits (ASIC), programmable logic controllers and embedded microcontrollers, etc.; the processor is used to execute the executable instructions to implement the power grid energy scheduling method based on multi-source data fusion as described above.

[0139] On the basis of the above embodiments, the present application further provides a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the power grid energy scheduling method based on multi-source data fusion as described above.

[0140] Those skilled in the art can clearly understand that the modules and method steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software manner depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0141] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described devices, apparatuses and modules can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0142] In several embodiments provided in the present application, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the above-described device embodiments are merely schematic, for example, the division of the modules is only a logical function division, and actual implementation can have another division manner, for example, a plurality of modules or units can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed ones can be indirect coupling or communication connection through some interfaces, devices or apparatuses, which can be electrical, mechanical or other forms.

[0143] The modules described as separate components can or can not be physically separated, and the components shown as modules can or can not be physical units, i.e. can be located in one place or distributed on a plurality of network units. Part or all of the units can be selected to achieve the purpose of the embodiment according to actual needs.

[0144] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0145] If the functions are realized in the form of software function modules and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the present application that essentially contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory server, a random access memory server, a magnetic disk or an optical disk, and various storage media that can store program instructions.

[0146] In addition, it should be noted that the combination of the technical features in the present case is not limited to the combination of the claims in the present case or the combination of the embodiments described in the present case. All technical features described in the present case can be freely combined or combined, unless they contradict each other.

[0147] It should be noted that the above only lists specific embodiments of the present application. Obviously, the present application is not limited to the above embodiments, and there are many similar changes. All modifications directly derived or inferred from the disclosure of the present application by those skilled in the art shall fall within the scope of the present application.

[0148] The above is only a preferred embodiment of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A power grid energy scheduling method based on multi-source data fusion, characterized in that, Comprise the following steps: S1, obtain the water and electricity environment data, photovoltaic environment data and load data of the target power grid area, the water and electricity environment data includes ice jam degree parameter for reflecting the ice jam degree of river channel, the photovoltaic environment data includes morning mist light transmittance and morning mist duration for reflecting the influence of morning mist on photovoltaic module, and the load data includes load pulse signal of electric heating equipment; S2, construct a water and electricity output attenuation model based on the ice jam degree parameter, and construct a photovoltaic efficiency inhibition model based on the morning mist light transmittance and morning mist duration, to generate water and electricity output constraint condition and photovoltaic efficiency correction condition respectively; S3, the water and electricity output data and photovoltaic output data are preprocessed by applying the water and electricity output constraint condition and photovoltaic efficiency correction condition, the time alignment of the preprocessed water and electricity output data and photovoltaic output data is carried out through dynamic time warping algorithm, and a fusion feature vector containing environmental constraints is generated; S4, the fusion feature vector is taken as input, a multi-objective optimization model is constructed in combination with the water and electricity output constraint condition and photovoltaic efficiency correction condition, and a rolling optimization algorithm is adopted to obtain a scheduling scheme containing water and electricity output strategy, photovoltaic output strategy and thermal power peak shaving strategy; The multi-objective optimization model in the step S4 takes thermal power peak shaving cost and power grid voltage deviation as optimization objectives, and the objective function is represented as: The minimum thermal power peak shaving cost and weighted voltage deviation square sum, that is: In the formula, is the cost of thermal power peak shaving, is the voltage quality weight, is the node voltage deviation; The constraint condition of the multi-objective optimization model comprises: wherein is the hydroelectric power output, is the photovoltaic power output, is the photovoltaic rated power output, is the thermal power output, is the real-time load, is the rated load, is the power balance tolerance. 2.The power grid energy scheduling method based on multi-source data fusion of claim 1, wherein, The acquisition of the water and electricity environment data in the step S1 comprises: The ice jam thickness is acquired through ultrasonic sensor network deployed in the river channel, and the ice jam degree parameter is determined based on the ice jam thickness; The acquisition of the photovoltaic environment data comprises: The morning mist light transmittance and duration are acquired through sensors on the surface of photovoltaic module; The acquisition of the load data comprises: The start-stop signal and power fluctuation data of electric heating equipment are acquired through intelligent electric meter in transformer area. 3.The power grid energy scheduling method based on multi-source data fusion of claim 2, characterized in that, The water and electricity output attenuation model in the step S2 is represented as: wherein is the ice jam extent parameter, is a function fitted based on the river hydrodynamics characteristics to characterize the constraint of ice jam extent on the upper limit of the hydropower available output; The photovoltaic efficiency inhibition model is represented as: wherein is the haze transmission of the morning mist, is the duration of the morning mist, is a function based on light scattering theory and the temperature rise characteristics of the component, is used to characterize the correction of the photovoltaic power generation efficiency by the morning mist.

4. The power grid energy scheduling method based on multi-source data fusion according to claim 3, characterized in that, The preprocessing of water and electricity output data in the step S3 specifically comprises: In the formula, is the original data of the water and electricity output, is the upper limit of the water and electricity available output output by the water and electricity output attenuation model; The preprocessing of photovoltaic output data specifically comprises: wherein is the photovoltaic power prediction data, is the photovoltaic efficiency correction factor output by the photovoltaic efficiency suppression model.

5. The power grid energy scheduling method based on multi-source data fusion according to claim 4, characterized in that, The dynamic time warping algorithm in the step S3 specifically comprises: A space-time cost function is constructed: In the formula, is a time deviation weight, is a power deviation weight, are respectively sampling times of the hydropower output data and the photovoltaic output data, are respectively the hydropower output data and the photovoltaic output data after preprocessing. Based on the space-time cost function, the optimal alignment path of water and electricity output data and photovoltaic output data is solved, so that the time error does not exceed the preset threshold, and the time-aligned joint output sequence is generated; The joint output sequence and the load data are spliced to obtain the fusion feature vector. 6.The power grid energy scheduling method based on multi-source data fusion of claim 1, wherein, Further comprising: The actual execution data and predicted data of the scheduling scheme are compared, if the deviation exceeds the preset threshold, the parameters of the water and electricity output attenuation model and photovoltaic efficiency inhibition model are adaptively corrected; The parameters comprise river channel characteristic coefficient and atmospheric attenuation coefficient.

7. A power grid energy scheduling system based on multi-source data fusion, applied to the method of any one of claims 1-6, characterized in that, Comprise: An edge computing layer is configured to acquire water and electricity environment data, photovoltaic environment data and load data of a target power grid area, the water and electricity environment data includes ice jam degree parameter, the photovoltaic environment data includes morning mist light transmittance and morning mist duration, and the load data includes load pulse signal of electric heating equipment. The edge computing layer is also used for generating a water and electricity output constraint condition and a photovoltaic efficiency correction condition based on the ice plug degree parameter and the morning fog parameter, preprocessing and time aligning water and electricity output data and photovoltaic output data, and generating a fusion feature vector; The cloud decision layer is used for receiving the fusion feature vector, constructing a multi-objective optimization model containing the water and electricity output constraint condition and the photovoltaic efficiency correction condition, and solving the scheduling scheme by using a rolling optimization algorithm; The communication control layer is used for realizing data interaction of the edge computing layer and the cloud decision layer, and sending the scheduling scheme to a power generation equipment for execution.

Images