Flexible resource scheduling method and device of power system, terminal equipment and storage medium

Abstract

The invention discloses a flexible resource scheduling method and device of an electric power system, terminal equipment and a storage medium, and belongs to the field of electric power scheduling, the flexible resource scheduling method of the electric power system corrects wind power output, photovoltaic output and electrical load through error probability density functions of wind power output, photovoltaic output and electrical load, so that the scheduling efficiency of the electric power system is improved. Random characteristic distribution of renewable energy output and load fluctuation is obtained, quantification of uncertainty distribution of the net load is further achieved, and corresponding first scheduling amount and second scheduling amount can be generated for flexibility demand scenes of insufficient power and excessive power. The flexible resources are accurately scheduled according to the first scheduling amount and the second scheduling amount, vacancy can be supplemented or surplus power can be consumed in time, uplink and downlink flexibility requirements can be effectively met, stable operation of a power system is guaranteed, and the problem that the uplink and downlink flexibility requirements cannot be effectively met in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system dispatching, and in particular to a flexible resource dispatching method, device, terminal device and storage medium for a power system. Background Art

[0002] In the operation of a power system, the upward flexibility demand is the demand for the ability to increase the power generation or reduce the power consumption load when the system power is insufficient to maintain the power supply-demand balance; the downward flexibility demand is the demand for the ability to reduce the power generation or increase the power consumption load when the system power is excessive. Due to the randomness of the renewable energy output (such as wind power being affected by meteorology and photovoltaic being restricted by sunlight) and the volatility of the power consumption load (influenced by user behavior, seasons, etc.), the renewable energy output and the power consumption load have an uncertain distribution, resulting in a significant increase in the uncertainty of the net load distribution during the operation of the power system. Furthermore, the possibility of power shortage or excess in the power system increases, leading to an increase in the upward and downward flexibility demands, and the need to dispatch flexible resources to meet the upward and downward flexibility demands of the power system. Therefore, accurately quantifying the uncertainty of the net load is crucial for reasonably dispatching flexible resources and meeting the upward and downward flexibility demands.

[0003] However, traditional technologies often respond to the upward and downward flexibility demands based on the average-based dispatching method. For example, the historical averages of the renewable energy output and the power consumption load are calculated and used as the basis for dispatching. When responding to the upward flexibility demand, the flexible resources are dispatched after predicting the insufficient power according to the average value; when responding to the downward flexibility demand, the flexible resources are dispatched after predicting the excess power according to the average value. Therefore, traditional technologies do not consider the uncertainty of the net load during the actual operation of the power system. Since the actual operation may deviate greatly from the average value, when the renewable energy output or the power consumption load deviates significantly from the average value, the dispatching result will be inaccurate, and there is a problem that the upward and downward flexibility demands cannot be effectively addressed, which may lead to power shortages or surpluses, and thus the stable operation of the power system cannot be ensured. Summary of the Invention

[0004] Embodiments of the present invention provide a flexible resource dispatching method, device, terminal device and storage medium for a power system, which realizes the quantification of the uncertainty distribution of the net load. For the uncertainty scenarios of power shortage and power surplus, the corresponding first dispatching quantity and the second dispatching quantity can be generated respectively. When the net load changes, the flexible resources are accurately dispatched through the first dispatching quantity and the second dispatching quantity to timely supplement the shortage or absorb the excess power, ensuring the stable operation of the power system, and effectively solving the problem in the prior art that the upward and downward flexibility demands cannot be effectively addressed, which may lead to power shortages or surpluses, and thus the stable operation of the power system cannot be ensured.

[0005] An embodiment of the present invention provides a method for scheduling flexibility resources of a power system, including:

[0006] Perform error correction on the current wind power output according to the probability density function of the wind power output error to generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; perform error correction on the current photovoltaic power output according to the probability density function of the photovoltaic power output error to generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic power output; perform error correction on the current electricity load according to the probability density function of the electricity load error to generate a load probability density function for quantifying the uncertainty distribution of the electricity load;

[0007] Generate a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function;

[0008] Generate a first scheduling quantity for responding to the upward flexibility demand of the power system and a second scheduling quantity for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic power output, the current electricity load, and the net load probability density function;

[0009] When responding to the upward flexibility demand of the power system, schedule the flexibility resources of the power system according to the first scheduling quantity; or, when responding to the downward flexibility demand of the power system, schedule the flexibility resources of the power system according to the second scheduling quantity.

[0010] Preferably, before generating the first output probability density function and the second output probability density function, it further includes:

[0011] Obtain the current temperature and the current wind speed;

[0012] When it is determined that the current temperature is not less than the preset temperature threshold or the current wind speed is not less than the preset wind speed threshold, input the current temperature, the current wind speed, the current wind power output, and the current photovoltaic power output into a preset output error prediction model, so that the output error prediction model extracts meteorological features and output features according to the current temperature, the current wind speed, the current wind power output, and the current photovoltaic power output; wherein, the meteorological features are used to characterize the dynamic change characteristics of the meteorology; the output features are used to characterize the operating characteristics of the power generation equipment;

[0013] Based on the cross-attention mechanism, perform weighted fusion on the meteorological features and the output features to generate fused features; wherein, the fused features are used to characterize the coupling relationship between the meteorology and the operation of the power generation equipment;

[0014] Generate the wind power output error corresponding to the current wind power output and the photovoltaic power output error corresponding to the current photovoltaic power output according to the fused features;

[0015] Correct the current wind power output according to the wind power output error to generate the corrected wind power output; correct the current photovoltaic power output according to the photovoltaic power output error to generate the corrected photovoltaic power output;

[0016] Respectively use the corrected wind power output and the corrected photovoltaic power output as the current wind power output corrected according to the wind power output error probability density function and the current photovoltaic power output corrected according to the photovoltaic power output error probability density function;

[0017] When it is determined that the current temperature is less than the preset temperature threshold or the current wind speed is less than the preset wind speed threshold, do not perform correction operations on the current wind power output and the current photovoltaic power output.

[0018] Preferably, the training process of the output error prediction model includes:

[0019] Generate a number of training samples according to the temperature historical samples, wind speed historical samples, wind power output historical samples and photovoltaic power output historical samples;

[0020] Iteratively train the output error prediction model to be trained according to each training sample until the model converges, and output the trained output error prediction model;

[0021] Wherein, in each training, input a training sample into the output error prediction model so that the output error prediction model outputs the wind power output error prediction result and the photovoltaic power output error prediction result corresponding to the training sample;

[0022] Compare the wind power output error prediction result and the photovoltaic power output error prediction result corresponding to the training sample with the actual results of the wind power output error and the photovoltaic power output error of the training sample respectively, and adjust the network parameters of the output error prediction model according to the comparison results.

[0023] Preferably, it further includes:

[0024] Obtain the installed capacity ratio of each wind turbine in the power system, the wind power output data of each wind turbine, the installed capacity ratio of each photovoltaic unit, the photovoltaic power output data of each photovoltaic unit, and the power consumption load data;

[0025] Generate a number of wind power output intervals according to the installed capacity ratio of each wind turbine; among them, different wind turbine intervals correspond to different average wind power outputs;

[0026] The kernel density estimation method is used to respectively fit the error distribution of the wind power output data in each wind power output interval, and generate a wind power output error probability density function; wherein, the wind power output error probability density function is used to characterize the wind power output error probability distribution of different wind power output intervals;

[0027] According to the installed capacity ratio of each photovoltaic unit, a number of photovoltaic output intervals are generated; wherein, different photovoltaic output intervals correspond to different average photovoltaic outputs;

[0028] The kernel density estimation method is used to respectively fit the error distribution of the photovoltaic output data in each photovoltaic output interval, and generate a photovoltaic output error probability density function; wherein, the photovoltaic output error probability density function is used to characterize the photovoltaic output error probability distribution of different photovoltaic output intervals;

[0029] The electricity load data is divided into a number of electricity load intervals; wherein, different electricity load intervals correspond to different average electricity loads;

[0030] The kernel density estimation method is used to respectively fit the error distribution of the electricity load data in each electricity load interval, and generate an electricity load error probability density function; wherein, the electricity load error probability density function is used to characterize the electricity load error probability distribution of different electricity load intervals.

[0031] Preferably, the generating a net load probability density function for quantifying the net load uncertainty distribution according to the first output probability density function, the second output probability density function and the load probability density function includes:

[0032] Monte Carlo sampling is performed on the first output probability density function, the second output probability density function and the load probability density function to generate a number of sampling data; wherein, each sampling data includes: the wind power output to be processed, the photovoltaic output to be processed and the electricity load to be processed;

[0033] For each sampling data, a net load sample corresponding to the sampling data is generated according to the wind power output to be processed, the photovoltaic output to be processed and the electricity load to be processed in the sampling data;

[0034] Based on the kernel density estimation method, the net load samples are fitted to generate a net load probability density function.

[0035] Preferably, the generating a first dispatch amount for responding to the upward flexibility demand of the power system and a second dispatch amount for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic output, the current electricity load and the net load probability density function includes:

[0036] Integrate the net load probability density function to generate a net load cumulative distribution function;

[0037] Generate a current net load prediction interval according to a preset confidence level and the net load cumulative distribution function; wherein, the current net load prediction interval includes: a net load prediction upper limit value and a net load prediction lower limit value;

[0038] Generate a first dispatch quantity and a second dispatch quantity according to the current wind power output, the current photovoltaic power output, the current power consumption load, the net load prediction upper limit value, and the net load prediction lower limit value.

[0039] Preferably, the current wind power output includes the wind power outputs corresponding to different moments within the day; the current photovoltaic power output includes the photovoltaic power outputs corresponding to different moments within the day; the current power consumption load includes the power consumption loads corresponding to different moments within the day; the current net load prediction interval includes the net load prediction upper limit values corresponding to different moments within the day and the net load prediction lower limit values corresponding to different moments;

[0040] The generating the first dispatch quantity and the second dispatch quantity according to the current wind power output, the current photovoltaic power output, the current power consumption load, and the current net load prediction interval includes:

[0041] Obtain the current dispatch time period within the day; wherein, the dispatch time period includes: a first time period or a second time period; the end moment of the first time period is less than the start moment of the second time period;

[0042] Generate net load prediction values corresponding to different moments within the day according to the current wind power output, the current photovoltaic power output, and the current power consumption load;

[0043] When it is determined that the current dispatch time period is the first time period, use the net load prediction value corresponding to the start moment of the first time period as the first net load prediction value; use the next moment of the current moment corresponding to the first net load prediction value as the first target moment; generate a first dispatch quantity according to the difference between the first net load prediction value and the net load prediction upper limit value corresponding to the first target moment; generate a second dispatch quantity according to the difference between the first net load prediction value and the net load prediction lower limit value corresponding to the first target moment;

[0044] When it is determined that the current dispatch time period is the second time period, use the net load prediction values corresponding to each moment within the second time period as the second net load prediction values; use the largest second net load prediction value as the first net load prediction value to be calculated, and use the smallest second net load prediction value as the second net load prediction value to be calculated;

[0045] Determine whether the moment corresponding to the second net load prediction value to be calculated is less than the moment corresponding to the first net load prediction value to be calculated;

[0046] If so, use the moment corresponding to the first net load prediction value to be calculated as the second target moment, and generate a first dispatch amount according to the difference between the upper limit value of the net load prediction corresponding to the second target moment and the second net load prediction value to be calculated; Generate a second dispatch amount according to the difference between the second net load prediction value to be calculated and the lower limit value of the net load prediction corresponding to the second target moment;

[0047] If not, use the moment corresponding to the second net load prediction value to be calculated as the third target moment, and generate a first dispatch amount according to the difference between the upper limit value of the net load prediction corresponding to the third target moment and the first net load prediction value to be calculated; Generate a second dispatch amount according to the difference between the first net load prediction value to be calculated and the lower limit value of the net load prediction corresponding to the third target moment.

[0048] Based on the above method embodiments, the present invention correspondingly provides apparatus embodiments.

[0049] An embodiment of the present invention provides a flexibility resource scheduling device for a power system, including: a first probability density function generation module, a second probability density function generation module, a dispatch amount generation module, and a flexibility resource scheduling module;

[0050] The first probability density function generation module is configured to perform error correction on the current wind power output according to the wind power output error probability density function to generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; Perform error correction on the current photovoltaic output according to the photovoltaic output error probability density function to generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic output; Perform error correction on the current electricity load according to the electricity load error probability density function to generate a load probability density function for quantifying the uncertainty distribution of the electricity load;

[0051] The second probability density function generation module is configured to generate a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function;

[0052] The dispatch amount generation module is configured to generate a first dispatch amount for responding to the upward flexibility demand of the power system and a second dispatch amount for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic output, the current electricity load, and the net load probability density function;

[0053] The flexibility resource scheduling module is used to schedule the flexibility resources of the power system according to the first scheduling amount when responding to the upstream flexibility demand of the power system; or, to schedule the flexibility resources of the power system according to the second scheduling amount when responding to the downstream flexibility demand of the power system.

[0054] Based on the above method embodiments, the present invention correspondingly provides embodiments of a terminal device.

[0055] Another embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the method for scheduling flexibility resources of a power system described in the above embodiments of the present invention.

[0056] Based on the above method embodiments, the present invention correspondingly provides embodiments of a storage medium.

[0057] Another embodiment of the present invention provides a storage medium. The computer-readable storage medium includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the method for scheduling flexibility resources of a power system described in the above embodiments of the present invention.

[0058] By implementing the present invention, the following beneficial effects are achieved:

[0059] The embodiments of the present invention provide a flexible resource scheduling method, device, terminal device and storage medium for a power system. In the present invention, for wind power output, photovoltaic power output and power consumption load, their error probability density functions are respectively used to correct the current values, generating a first output probability density function for quantifying the uncertainty distribution of the current wind power output, a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic power output, and a load probability density function for quantifying the uncertainty distribution of the power consumption load, so as to more accurately reflect the actual uncertainty characteristics of renewable energy output and power consumption load; further, after obtaining the above three probability density functions, the present invention generates a net load probability density function for quantifying the uncertainty distribution of the net load based on this. Since the comprehensive impact of the uncertainties of wind power, photovoltaic power output and power consumption load on the net load is considered, the various possible value situations and corresponding probabilities of the net load in actual operation can be more accurately characterized. Finally, according to the current wind power output, photovoltaic power output, power consumption load and net load probability density function, the scheduling amounts for responding to the upward and downward flexibility requirements of the power system can be accurately generated. Compared with the prior art, the present invention corrects the current wind power output, photovoltaic power output and power consumption load through the error probability density functions of wind power, photovoltaic power output and power consumption load, can accurately obtain the random characteristic distribution of renewable energy output and load fluctuations, and further realizes the quantification of the uncertainty distribution of the net load. For the uncertainty scenarios of power shortage and power surplus, the corresponding first scheduling amount and second scheduling amount can be respectively generated. When the net load changes, the flexible resources can be accurately scheduled through the first scheduling amount and the second scheduling amount, the shortage can be timely supplemented or the surplus power can be absorbed, effectively reducing the risk of system power imbalance and ensuring the stable operation of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 FIG. is a schematic flowchart of a flexible resource scheduling method for a power system provided by an embodiment of the present invention.

[0061] Figure 2 FIG. is a schematic structural diagram of a flexible resource scheduling device for a power system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0063] Such as Figure 1As shown, to solve the problem in the prior art that it is impossible to effectively meet the flexibility requirements of the uplink and downlink, which may lead to power deficits or surpluses and thus cannot ensure the stable operation of the power system, an embodiment of the present invention provides a method for scheduling flexibility resources in a power system, including:

[0064] Step S1: Perform error correction on the current wind power output according to the wind power output error probability density function to generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; perform error correction on the current photovoltaic power output according to the photovoltaic power output error probability density function to generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic power output; perform error correction on the current electricity load according to the electricity load error probability density function to generate a load probability density function for quantifying the uncertainty distribution of the electricity load.

[0065] Illustratively, the present invention can obtain the current wind power output, current photovoltaic power output, and current electricity load corresponding to the current day, and thus respond to the uplink flexibility requirement and downlink flexibility requirement of the current day according to the obtained wind power output, photovoltaic power output, and electricity load, so as to realize the scheduling of flexibility resources.

[0066] Specifically, the current wind power output can be error-corrected according to the wind power output error probability density function to obtain the first output probability density function. After error-correcting the current wind power output, the first output probability density function can be fitted based on the current wind power output that has completed error correction, and then the uncertainty distribution of the current wind power output can be obtained based on the first output probability density function;

[0067] The current photovoltaic power output can be error-corrected according to the photovoltaic power output error probability density function to obtain the second output probability density function. After error-correcting the current photovoltaic power output, the second output probability density function can be fitted based on the photovoltaic power output that has completed error correction, and then the uncertainty distribution of the current photovoltaic power output can be obtained based on the second output probability density function;

[0068] The current electricity load can be error-corrected according to the electricity load error probability density function to obtain the load probability density function. After error-correcting the electricity load, the load probability density function can be fitted based on the electricity load that has completed error correction, and then the uncertainty distribution of the electricity load can be obtained based on the load probability density function.

[0069] It can be understood that, due to the influence of volatile factors such as meteorological conditions on the output of wind power and photovoltaic power, and the interference of factors such as user behavior on the electricity load, there is inherent uncertainty. In the embodiments of the present invention, error correction is performed through respective error probability density functions, and these sources of uncertainty can be accurately presented in the form of a probability distribution. For example, the output of wind power fluctuates at different wind speeds. By using the error probability density function of wind power output, the possibility and degree of such fluctuations can be quantified, and the obtained first output probability density function can more accurately reflect the uncertainty distribution of wind power output. The same applies to photovoltaic power and electricity load.

[0070] Compared with only using the original point predictions (such as the current wind power output, the current photovoltaic power output, and the current electricity load), by generating the probability density function after error correction, the present invention can provide a reliable uncertainty description for wind power, photovoltaic power, and electricity load, and reduce the prediction deviation and dispatching decision-making errors caused by insufficient uncertainty estimation.

[0071] Step S2: Generate a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function;

[0072] Schematically, the net load is the remaining load after subtracting the wind power and photovoltaic power outputs from the electricity load, and reflects the power demand that needs to be balanced by other conventional power sources (such as thermal power, hydropower, etc.) in the power system. Since the first output probability density function, the second output probability density function, and the load probability density function respectively describe the uncertainties of wind power output, photovoltaic power output, and electricity load, in the embodiments of the present invention, by combining them to generate the net load probability density function, the influence of the uncertainties of wind power, photovoltaic power, and load on the net load can be comprehensively considered, so as to more accurately reflect the changes in the net load during the actual operation of the power system.

[0073] The uncertainties of wind power output, photovoltaic power output, and electricity load interact with each other and propagate to the net load. For example, when the wind power and photovoltaic power outputs fluctuate greatly at the same time, the change in the net load may be more severe. Therefore, the net load probability density function of the present invention can capture the propagation and superposition effects of the uncertainties of wind power output, photovoltaic power output, and electricity load, and provide more accurate net load dynamic change information that conforms to the actual operation of the power system for obtaining the first dispatching amount and the second dispatching amount subsequently.

[0074] Step S3: Generate a first dispatching amount for responding to the upward flexibility demand of the power system and a second dispatching amount for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic power output, the current electricity load, and the net load probability density function;

[0075] Schematically, there is uncertainty in the current wind power, photovoltaic power output and electricity load, and the net load probability density function quantifies this uncertainty. Generating the first and second scheduling quantities based on the net load probability density function can accurately meet the actual flexibility requirements of the system. For example, when the wind power output fluctuates greatly, the scheduling quantities (such as the first scheduling quantity and the second scheduling quantity) required to cope with the increase (upward flexibility requirement) or decrease (downward flexibility requirement) of the net load can be accurately calculated according to the net load probability density function, avoiding under-scheduling or over-scheduling, and enabling the system to quickly respond to the current operating conditions and ensure the dynamic balance of power supply and demand.

[0076] Step S4: When responding to the upward flexibility requirement of the power system, schedule the flexibility resources of the power system according to the first scheduling quantity; or, when responding to the downward flexibility requirement of the power system, schedule the flexibility resources of the power system according to the second scheduling quantity.

[0077] Schematically, when the power system has a power shortage and needs to respond to the upward flexibility requirement, the present invention can schedule the flexibility resources of the power system according to the first scheduling quantity adapted to the actual situation generated previously; for example, quickly start the standby generator set to increase the power generation, or negotiate with users to cut some interruptible loads and other measures, so as to quickly make up for the power shortage of the power system and maintain the balance of power supply and demand.

[0078] When the power system has a power surplus and needs to respond to the downward flexibility requirement, the present invention schedules the flexibility resources of the power system according to the second scheduling quantity. For example, adjust the output of the generator set to reduce the power generation, or start the adjustable load to increase the electricity consumption to consume the excess power and avoid damage to the power system caused by power surplus.

[0079] Therefore, according to the accurately obtained first scheduling quantity and second scheduling quantity, the present invention can reasonably allocate flexibility resources such as thermal power, hydropower, and energy storage. After calculating the upward or downward scheduling quantity, the allocation task can be optimized according to factors such as the regulation performance and cost of each resource to improve the resource utilization efficiency. For example, when it is necessary to increase the power generation, give priority to calling the generator set with good regulation performance and low cost.

[0080] For step S1, in a preferred embodiment, when the present invention corrects the errors of the current wind power output and the current photovoltaic power output according to the wind power output error probability density function and the photovoltaic power output error probability density function, it can also discriminate the current meteorological data. If the current weather is in an unstable or extreme weather condition, such as the temperature is not less than the preset temperature threshold or the wind speed is not less than the preset wind speed threshold, then the current wind power output and the current photovoltaic power output can be pre-corrected first (can be corrected according to the meteorological data), and then the error correction is performed through the error probability density function, then there is:

[0081] Before generating the first output probability density function and the second output probability density function, it further includes:

[0082] Obtain the current temperature and the current wind speed;

[0083] When it is determined that the current temperature is not less than the preset temperature threshold or the current wind speed is not less than the preset wind speed threshold, input the current temperature, the current wind speed, the current wind power output, and the current photovoltaic output into a preset output error prediction model, so that the output error prediction model extracts meteorological features and output features according to the current temperature, the current wind speed, the current wind power output, and the current photovoltaic output; wherein, the meteorological features are used to characterize the dynamic change characteristics of the meteorology; the output features are used to characterize the operating characteristics of the power generation equipment;

[0084] Based on the cross-attention mechanism, weight and fuse the meteorological features and the output features to generate the fused features; wherein, the fused features are used to characterize the coupling relationship between the meteorology and the operation of the power generation equipment;

[0085] Generate the wind power output error corresponding to the current wind power output and the photovoltaic output error corresponding to the current photovoltaic output according to the fused features;

[0086] Correct the current wind power output according to the wind power output error to generate the corrected wind power output; correct the current photovoltaic output according to the photovoltaic output error to generate the corrected photovoltaic output;

[0087] Respectively use the corrected wind power output and the corrected photovoltaic output as the current wind power output corrected according to the wind power output error probability density function and the current photovoltaic output corrected according to the photovoltaic output error probability density function;

[0088] When it is determined that the current temperature is less than the preset temperature threshold or the current wind speed is less than the preset wind speed threshold, no correction operation is performed on the current wind power output and the current photovoltaic output.

[0089] Schematically, the wind and light output will fluctuate violently in extreme weather. Assuming that the current wind power output and the current photovoltaic output are predicted by a model or algorithm, then in extreme weather, the prediction errors of the current wind power output and the current photovoltaic output will increase significantly.

[0090] To solve the above problems, the present invention can obtain the current temperature and wind speed, compare them with the preset threshold, and identify extreme weather. When extreme weather occurs (the current temperature is not less than the preset temperature threshold or the current wind speed is not less than the preset wind speed threshold), the wind and light output can be corrected for the special situation in extreme weather to improve the prediction accuracy.

[0091] Specifically, in the embodiments of the present invention, meteorological features and output features can be extracted through an output error prediction model and fused using a cross-attention mechanism. The meteorological features characterize the meteorological dynamic change characteristics, and the output features characterize the operating characteristics of the power generation equipment. The fused features reflect the coupling relationship between meteorology and the operation of the power generation equipment. For example, the change in wind speed under strong wind weather not only affects the power generation efficiency of the wind turbine but also causes the wind turbine to adjust its output due to the protection mechanism. Through the fused features, these factors can be comprehensively considered, so as to more comprehensively grasp the internal mechanism of the change in the output of wind and light, and accurately predict the error in the output of wind and light.

[0092] Based on the fused features, the output errors of wind power and photovoltaic power are generated, and then the current wind power and photovoltaic power can be corrected. Taking the corrected output as the basis for subsequent error correction according to the error probability density function can further improve the prediction accuracy. Under high-temperature weather, the efficiency of photovoltaic modules will decrease. By correcting the current photovoltaic output with the photovoltaic output error calculated by the model, the output prediction value can be made closer to the actual output.

[0093] When it is determined that the current temperature is less than the preset temperature threshold or the current wind speed is less than the preset wind speed threshold, that is, in non-extreme weather conditions, no correction operation is performed on the current wind power output and the current photovoltaic power output; for example, the current wind power output and the current photovoltaic power output are continued to be used as the current wind power output for error correction according to the wind power output error probability density function and the current photovoltaic power output for error correction according to the photovoltaic power output error probability density function, respectively. It can be understood that the output of wind and light is relatively stable under non-extreme weather, and the prediction error is relatively small, so there is no need to perform complex correction calculations, thus reasonably optimizing the computing resources and improving the overall computing efficiency.

[0094] Further, in a preferred embodiment, the training process of the output error prediction model includes:

[0095] Generating a number of training samples according to the temperature historical samples, wind speed historical samples, wind power output historical samples, and photovoltaic power output historical samples;

[0096] Iteratively training the output error prediction model to be trained according to each training sample until the model converges, and outputting the trained output error prediction model;

[0097] Wherein, in each training, a training sample is input into the output error prediction model so that the output error prediction model outputs the wind power output error prediction result and the photovoltaic power output error prediction result corresponding to the training sample;

[0098] Compare the predicted results of wind power output error and photovoltaic power output error corresponding to the training samples with the actual results of wind power output error and photovoltaic power output error of the training samples respectively, and adjust the network parameters of the output error prediction model according to the comparison results.

[0099] Among them, in the process of generating training samples, since the sensitivities of wind and light to different meteorological elements are different, extreme weather on historical data can be marked according to meteorological forecast data. For example, when the temperature is greater than 30 °C, it is recorded as a high-temperature sample, and when the wind speed is greater than 10.8 m / s, it is recorded as a strong-wind sample. Thus, these high-temperature samples and strong-wind samples are used as temperature historical samples and wind speed historical samples in the training samples.

[0100] For the strong-wind samples in the historical data, calculate the predicted wind power output error ΔP' W , for the high-temperature samples in the historical data, calculate the predicted photovoltaic power output error ΔP' S , P W,obs , P S,obs are the true output data of wind power and photovoltaic power, and P W,pred , P S,pred are the predicted output data of wind power and photovoltaic power. Then there is:

[0101]

[0102] Use the above temperature historical samples, wind speed historical samples, wind power output historical samples and photovoltaic power output historical samples as inputs, and the actual error of the output as the output to train and construct an output error prediction model.

[0103] Furthermore, considering that the number of extreme weather samples is small, the present invention can also screen the input meteorological data, calculate the correlation coefficient between each meteorological feature and the prediction error, and retain the meteorological data with the absolute value of the correlation coefficient greater than 0.5 as the meteorological data of extreme weather, such as high-temperature meteorological data and strong-wind meteorological data.

[0104] Fit the above high-temperature meteorological data and strong-wind meteorological data according to the probability density function of the Gaussian mixture model GMM, and more training samples can be obtained. Schematically, GMM is a model based on probability distribution, which can capture the multimodal characteristics of data and generate new samples by sampling. Then there is:

[0105] First, fit the probability density function p(x, y) of GMM according to the historical data, and estimate the parameters of each Gaussian component. The formula is as follows:

[0106]

[0107] where \((x, y)\) represents the features and target values of the sample; \(K\) represents the number of Gaussian components; \(\pi\) k represents the weight of the \(k\)-th Gaussian component, satisfying \(\pi\) k \(\geq 0\) and while is the probability density function of the \(k\)-th Gaussian component, then we have:

[0108]

[0109] where \(D\) represents the data dimension, \(\mu\) k represents the mean vector of the \(k\)-th Gaussian component; \(\Sigma\) k represents the covariance matrix of the \(k\)-th Gaussian component.

[0110] According to the weight \(\pi\) k , randomly select a Gaussian component \(k\), and sample a new sample \((x', y')\) from the selected Gaussian component . Then, based on the augmented samples, a machine learning model can be used to capture the mapping relationship between meteorological features, wind and light power output and prediction errors, so as to obtain an output error prediction model under extreme weather conditions.

[0111] In a preferred embodiment, the output error prediction model of the embodiment of the present invention can be modeled using the integrated tree model XgBoost, and the model XgBoost is an efficient machine learning algorithm based on Gradient Boosting Decision Tree (GBDT). It minimizes the objective function by gradually adding weak learners (usually decision trees), and each step of the weak learner focuses on fitting the residuals (i.e., prediction errors) of the previous step model, thereby gradually improving the performance of the overall model. Then the objective function is:

[0112]

[0113] where \(L(y\) i , \(y\) i ) is the loss function, which is used to measure the difference between the predicted value \(y\) i and the true value \(y\) i ; \(\Omega(f\) k ) is the regularization term, which is used to control the model complexity and prevent overfitting; \(N\) is the number of samples, and \(K\) is the number of decision trees.

[0114] ​Schematically, in the embodiments of the present invention, Gaussian mixture model (GMM) is used to fit the screened meteorological data to generate new samples, expanding the training sample size. Since GMM can capture the multimodal characteristics of data, the generated new samples cover more meteorological and output combination situations that may occur under extreme weather, enabling the model to learn richer patterns during the training process, enhancing the robustness of the model, and reducing the risk of model overfitting. The XgBoost algorithm based on gradient boosting decision tree can also be used to construct an output error prediction model. XgBoost effectively improves the prediction performance of the model by gradually adding weak learners to fit the residuals of the previous model and continuously optimizing the objective function.

[0115] Further, after determining whether to correct the current wind power output and current photovoltaic output according to the current temperature and current wind speed, error correction can also be performed on the wind power output, photovoltaic output, and electricity load based on the preset probability density functions of wind power output error, photovoltaic output error, and electricity load error. The generation processes of the different probability density functions of errors are as follows:

[0116] Obtain the installed capacity ratio of each wind turbine in the power system, the wind power output data of each wind turbine, the installed capacity ratio of each photovoltaic unit, the photovoltaic output data of each photovoltaic unit, and the electricity load data.

[0117] According to the installed capacity ratio of each wind turbine, generate several wind power output intervals; among them, different wind turbine intervals correspond to different average wind power outputs.

[0118] Use the kernel density estimation method to fit the error distribution of the wind power output data in each wind power output interval respectively to generate a wind power output error probability density function; wherein, the wind power output error probability density function is used to characterize the wind power output error probability distribution of different wind power output intervals.

[0119] According to the installed capacity ratio of each photovoltaic unit, generate several photovoltaic output intervals; among them, different photovoltaic output intervals correspond to different average photovoltaic outputs.

[0120] Use the kernel density estimation method to fit the error distribution of the photovoltaic output data in each photovoltaic output interval respectively to generate a photovoltaic output error probability density function; wherein, the photovoltaic output error probability density function is used to characterize the photovoltaic output error probability distribution of different photovoltaic output intervals.

[0121] Divide the electricity load data into several electricity load intervals; among them, different electricity load intervals correspond to different average electricity loads.

[0122] The kernel density estimation method is used to fit the error distribution of the electricity load data in each electricity load interval respectively, and an electricity load error probability density function is generated; wherein, the electricity load error probability density function is used to characterize the electricity load error probability distribution in different electricity load intervals.

[0123] Schematically, the output of wind power / photovoltaic is directly related to the installed capacity. Wind power output = installed capacity × utilization rate. Then, the installed capacity ratio is a standardized measure of the output level. For example, a 20% installed capacity ratio means that the current output is 20% of the rated capacity (low output), and an 80% installed capacity ratio means close to the rated output (high output). Then, after generating several wind power output intervals according to the installed capacity ratio of each wind turbine, the output mean corresponding to each wind power output interval is different. Therefore, when performing error correction, the interval corresponding to the current wind power output can be found, and the current wind power output can be error-corrected according to the wind power output error probability distribution of this interval. The same applies to photovoltaic output and electricity load.

[0124] Different installed capacity ratios correspond to different output levels. In the embodiments of the present invention, by dividing the output intervals according to the installed capacity ratio, the operation characteristic differences of different capacity units can be considered. For example, when the installed capacity ratio is low, the unit may be operating at part load, and its error characteristics are different from those when the installed capacity ratio is high (close to the rated output), so that the finally generated error probability density function can more accurately reflect the error distribution at different output levels, thereby more accurately quantifying the uncertainty of wind power and photovoltaic output.

[0125] Dividing the electricity load data into several intervals takes into account the error characteristics at different electricity load levels. Different electricity load intervals correspond to different electricity usage scenarios and user behaviors, and their error distributions are also different. By fitting the error distribution of each interval respectively, the uncertainty of the electricity load can be more finely characterized, providing more accurate information for load forecasting and scheduling.

[0126] Therefore, the embodiments of the present invention can reduce the risk of power system scheduling by more accurately quantifying uncertainty and correcting errors. When facing the fluctuations of wind power, photovoltaic output and electricity load, it can predict and respond more accurately, reduce scheduling mistakes caused by insufficient error estimation, and improve the stability and reliability of power system operation.

[0127] In a preferred embodiment, the present invention is based on the historical prediction data of wind and light output ΔP W,corr 、ΔP S,corr and the actual output data P W,obs 、ΔP S,obs , and calculate the historical prediction errors of wind power and photovoltaic output ΔP W 、ΔP S; Based on the historical load prediction data ΔP L,pred and the measured data ΔP L,obs , calculate the historical load prediction error ΔP L , then we have:

[0128] ΔP W = P W,obs - P W,corr

[0129] ΔP S = P S,obs - P S,corr

[0130] ΔP L = P L,obs - P L,pred ;

[0131] For wind power and photovoltaic data, according to the installed capacity ratio, divide the historical output prediction data into n intervals, and the number of data in each interval is m, with n = 6 by default; for load data, divide the intervals according to the historical load prediction, with n = 6 by default, and the number of data in each interval is m.

[0132] Within each interval, use the kernel density estimation method to fit the distribution of the prediction error, and the formula is as follows:

[0133]

[0134] In the formula, f type,n (ΔP) can respectively represent the probability density function of wind power output error, the probability density function of photovoltaic output error, and the probability density function of electricity load error. ΔP represents the error, and ΔP i represents the prediction error of the i-th sample in the set, where K(u) is the kernel function, h is the bandwidth, and K(u) is the Gaussian kernel by default, is the standard deviation of the sample, and m is the sample size.

[0135] For wind power and photovoltaic output, according to the probability density function of wind power and photovoltaic output error in each interval, correct the wind power and photovoltaic output on the D-th day to obtain the first output probability density function and the second output probability density function corresponding to the wind power output and the photovoltaic output respectively;

[0136] For the electricity load, based on the obtained probability density function of the electricity load error in each interval, correct the electricity load on the D-th day to obtain the load probability density function, then we have:

[0137]

[0138] Among them, They can be respectively expressed as the first output probability density function, the second output probability density function, and the load probability density function.

[0139] For step S2, in a preferred embodiment, generating a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function includes:

[0140] Performing Monte Carlo sampling on the first output probability density function, the second output probability density function, and the load probability density function to generate a number of sampling data; wherein, each sampling data includes: the wind power output to be processed, the photovoltaic output to be processed, and the electricity load to be processed;

[0141] For each sampling data, generating a net load sample corresponding to the sampling data according to the wind power output to be processed, the photovoltaic output to be processed, and the electricity load to be processed in the sampling data;

[0142] Based on the kernel density estimation method, fitting each net load sample to generate a net load probability density function.

[0143] In a preferred embodiment, Monte Carlo sampling can be performed respectively from the corresponding to the first output probability density function, the second output probability density function, and the load probability density function. If q is the number of samplings, then:

[0144]

[0145] Next, calculate the net load p` of each sampled sample data net (i):

[0146] p` net (j) = p` L (j) - p` W (j) - p` S (j), j = 1, 2,..., q;

[0147] Finally, using the kernel density estimation method to fit each net load sample, and defining the net load probability density function as φ net (p`), then:

[0148]

[0149] where K(u) is the kernel function and h is the bandwidth.

[0150] Schematically, there are uncertainties in the wind power output, photovoltaic output, and electricity load respectively, and the first and second output probability density functions and the load probability density function quantify these uncertainties. In the embodiments of the present invention, through Monte Carlo sampling, a large number of sampling data are generated from these probability density functions, comprehensively covering various possible output combination situations of wind, light, and load. Each sampling data includes the wind power output to be processed, the photovoltaic output to be processed, and the electricity load to be processed, fully considering the interaction of the uncertainties of the three, avoiding the errors caused by only considering a single factor or simple averaging, and thus more truly reflecting the uncertainty distribution of the net load in actual operation.

[0151] Based on the sampling data, a net load sample is generated, and the kernel density estimation method is used for fitting to obtain the net load probability density function. Since the non-parametric estimation method does not require a prior assumption about the distribution form of the net load, it can flexibly adapt to various complex distribution situations. Whether the net load distribution is normal, skewed, or multimodal, it can accurately describe its probability density. Under extreme weather conditions, changes in wind power output, photovoltaic output, and electricity load may lead to abnormal distribution of the net load, and the kernel density estimation method can effectively capture these changes, providing more practical net load distribution information for the flexible scheduling of the power system.

[0152] For step S3, in a preferred embodiment, the generating a first scheduling amount for responding to the upward flexibility demand of the power system and a second scheduling amount for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic output, the current electricity load, and the net load probability density function includes:

[0153] Integrate the net load probability density function to generate a net load cumulative distribution function;

[0154] Generate a current net load prediction interval according to the preset confidence level and the net load cumulative distribution function; wherein, the current net load prediction interval includes: a net load prediction upper limit value and a net load prediction lower limit value;

[0155] Generate a first scheduling amount and a second scheduling amount according to the current wind power output, the current photovoltaic output, the current electricity load, the net load prediction upper limit value, and the net load prediction lower limit value.

[0156] Specifically, based on the net load probability density function, the net load cumulative distribution function is obtained:

[0157]

[0158] By setting different confidence levels 1-α, the upper and lower limit values P net,up 、P net,down of the net load prediction interval under different confidence levels can be obtained, realizing the interval prediction of the net load:

[0159] P{P net,down <P` net <P net,up} = 1 - α;

[0160] In the embodiments of the present invention, the current net load prediction interval can be obtained according to the preset confidence level, so as to obtain different net load prediction upper limit values and net load prediction lower limit values in the interval.

[0161] In a preferred embodiment, the present invention can calculate the scheduling amounts in different time periods and calculate the flexibility requirements of different time scales, and there are: the current wind power output includes the wind power outputs corresponding to different moments within the day; the current photovoltaic power output includes the photovoltaic power outputs corresponding to different moments within the day; the current power consumption load includes the power consumption loads corresponding to different moments within the day; the current net load prediction interval includes the net load prediction upper limit values corresponding to different moments within the day and the net load prediction lower limit values corresponding to different moments within the day;

[0162] Generating a first scheduling amount and a second scheduling amount according to the current wind power output, the current photovoltaic power output, the current power consumption load, and the current net load prediction interval includes:

[0163] Obtain the current scheduling time period within the day; wherein, the scheduling time period includes: a first time period or a second time period; the end time of the first time period is less than the start time of the second time period;

[0164] Generate the net load prediction values corresponding to different moments within the day according to the current wind power output, the current photovoltaic power output, and the current power consumption load;

[0165] When it is determined that the current scheduling time period is the first time period, the net load prediction value corresponding to the start time of the first time period is used as the first net load prediction value; the next moment of the current moment corresponding to the first net load prediction value is used as the first target moment; a first scheduling amount is generated according to the difference between the first net load prediction value and the net load prediction upper limit value corresponding to the first target moment; a second scheduling amount is generated according to the difference between the first net load prediction value and the net load prediction lower limit value corresponding to the first target moment;

[0166] When it is determined that the current scheduling time period is the second time period, the net load prediction values corresponding to each moment within the second time period are used as the second net load prediction values; the largest second net load prediction value is used as the first net load prediction value to be calculated, and the smallest second net load prediction value is used as the second net load prediction value to be calculated;

[0167] Determine whether the moment corresponding to the second net load prediction value to be calculated is less than the moment corresponding to the first net load prediction value to be calculated;

[0168] If so, use the moment corresponding to the first net load prediction value to be calculated as the second target moment, and generate a first dispatch amount according to the difference between the net load prediction upper limit value corresponding to the second target moment and the second net load prediction value to be calculated; Generate a second dispatch amount according to the difference between the second net load prediction value to be calculated and the net load prediction lower limit value corresponding to the second target moment;

[0169] If not, use the moment corresponding to the second net load prediction value to be calculated as the third target moment, and generate a first dispatch amount according to the difference between the net load prediction upper limit value corresponding to the third target moment and the first net load prediction value to be calculated; Generate a second dispatch amount according to the difference between the first net load prediction value to be calculated and the net load prediction lower limit value corresponding to the third target moment.

[0170] It can be understood that in the embodiment of the present invention, the net load cumulative distribution function is obtained by integrating the net load probability density function, which can comprehensively describe the probability distribution characteristics of the net load. The current net load prediction interval is generated according to the preset confidence level and the net load cumulative distribution function, including the net load prediction upper limit values at different moments and the lower limit values at different moments. Since different confidence levels can reflect different risk preferences, an appropriate confidence level can be selected according to the actual situation to cope with different degrees of uncertainty.

[0171] Furthermore, the current wind power output, photovoltaic power output and power consumption load can be refined into data at different moments within the day, fully considering the dynamic change characteristics of wind, light and load within a day, and the scheduling time period is divided into a first time period and a second time period, and different scheduling amount calculation methods are adopted for different time periods.

[0172] In the first time period, the dispatch amount is calculated based on the net load prediction value at the starting moment; in the second time period, the dispatch amount is calculated according to the maximum and minimum net load prediction values and their moment relationship. And the above-mentioned scheduling method for distinguishing different time periods can realize short-term scheduling (such as the scheduling in the first time period) and long-term scheduling (such as the scheduling in the second time period), and can better adapt to the operating characteristics of the power system at different time periods, realize the refined scheduling of the power system, and improve the accuracy and effectiveness of scheduling.

[0173] In the first time period, by comparing the difference between the predicted net load value at a certain moment and the upper and lower limit values of the predicted net load at the next moment, the required dispatching amount of the system under different conditions can be accurately determined, ensuring that when there is a power deficit in the system, power generation can be increased or load can be reduced in a timely manner, and when there is a power surplus, power generation can be reduced or load can be increased in a timely manner, thereby ensuring the balance between power supply and demand and the stable operation of the power system.

[0174] In the second time period, by comparing the time relationship between the maximum and minimum predicted net load values to determine the calculation method of the dispatching amount, the complex changes of the net load in this time period can be better handled.

[0175] Specifically, since the time resolution of the predicted wind and solar power output data is generally 15 minutes, the characteristics of the power system demand vary at different time scales. The present invention can be divided according to the time scale, and the flexibility requirements can be divided into 15 minutes, 1 hour, 4 hours, and 24 hours. The first time period can be 15 minutes, and the second time period can be 1 hour, 4 hours, or 24 hours.

[0176] First, according to the current wind power output P W,corr 、the current photovoltaic power output P S,corr and the current electricity load P L , the predicted net load values P net corresponding to different moments within the day are generated:

[0177] P net =P L -P W,corr -P S,corr ;

[0178] For the net load data at the 15-minute time (i.e., the first time period) scale, considering its volatility and uncertainty, the upward and downward flexibility requirements at time t, that is, the first and second dispatching amounts are respectively

[0179] is the first predicted net load value at the starting time t of the first time period, are respectively the upper and lower limit values of the predicted net load at the next moment corresponding to the first predicted net load value, then there are:

[0180]

[0181] For the 1-hour, 4-hour, or 24-hour time scale, that is, for the second time period, the upward and downward flexibility demand amounts are The maximum and minimum predicted net load values within the time window of the second time period are respectively The predicted net load upper limit value and the predicted net load lower limit value are respectively

[0182] If the occurrence time of the minimum net load within the time window of the second time period is less than the occurrence time of the maximum net load, that is, the moment corresponding to the second net load prediction value to be calculated is less than the moment corresponding to the first net load prediction value to be calculated, then there is:

[0183]

[0184] Wherein, is the upper limit value of the net load prediction corresponding to the second target moment, is the lower limit of the net load prediction corresponding to the second target moment.

[0185] If the occurrence time of the minimum net load within the time window of the second time period is greater than the occurrence time of the maximum net load, that is, the moment corresponding to the second net load prediction value to be calculated is greater than the moment corresponding to the first net load prediction value to be calculated, then there is:

[0186]

[0187] Wherein, is the upper limit value of the net load prediction corresponding to the third target moment, is the lower limit value of the net load prediction corresponding to the third target moment.

[0188] Therefore, the embodiment of the present invention can determine the calculation method of the scheduling amount by comparing the moment relationship between the maximum and minimum net load prediction values, and can better cope with the complex change situation of the net load in this time period. For example, when the net load first decreases and then increases, calculating the scheduling amount according to the minimum net load prediction value can better cope with the situation of power surplus; when the net load first increases and then decreases, calculating the scheduling amount according to the maximum net load prediction value can better cope with the situation of power shortage. Therefore, the scheduling amount calculation method of the present invention can improve the adaptability of the power system to complex operating conditions and enhance the stability and reliability of the system.

[0189] For step S4, in a preferred embodiment, the up and down flexibility requirements can be quantified based on the first scheduling amount and the second scheduling amount, and then the scheduling schemes of resources such as thermal power, energy storage, and demand response can be obtained according to the first scheduling amount and the second scheduling amount, and dynamic scheduling can be performed through a hierarchical control strategy (model predictive control of thermal power units, dual-loop power tracking of energy storage, virtual synchronous machine regulation on the demand side) to achieve the response to the up and down flexibility requirements of the power system.

[0190] As Figure 2 shown, based on the embodiments of the above various flexibility resource scheduling methods of the power system, the present invention correspondingly provides an apparatus embodiment;

[0191] An embodiment of the present invention provides a flexibility resource scheduling device for a power system, including: a first probability density function generation module, a second probability density function generation module, a scheduling quantity generation module, and a flexibility resource scheduling module;

[0192] The first probability density function generation module is configured to perform error correction on the current wind power output according to the wind power output error probability density function to generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; perform error correction on the current photovoltaic output according to the photovoltaic output error probability density function to generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic output; perform error correction on the current power consumption load according to the power consumption load error probability density function to generate a load probability density function for quantifying the uncertainty distribution of the power consumption load;

[0193] The second probability density function generation module is configured to generate a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function;

[0194] The scheduling quantity generation module is configured to generate a first scheduling quantity for responding to the upward flexibility demand of the power system and a second scheduling quantity for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic output, the current power consumption load, and the net load probability density function;

[0195] The flexibility resource scheduling module is configured to schedule the flexibility resources of the power system according to the first scheduling quantity when responding to the upward flexibility demand of the power system; or schedule the flexibility resources of the power system according to the second scheduling quantity when responding to the downward flexibility demand of the power system.

[0196] It should be noted that the device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical modules. They may be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that they have a communication connection, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0197] Those skilled in the art can clearly understand that for the sake of convenience and brevity, the specific working process of the device described above can refer to the corresponding process in the foregoing method embodiments, which will not be elaborated here.

[0198] Based on the embodiments of the above various flexibility resource scheduling methods for power systems, the present invention correspondingly provides embodiments of terminal devices.

[0199] An embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a flexibility resource scheduling method for a power system described in any method embodiment of the present invention.

[0200] The terminal device may be a computing terminal device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory.

[0201] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device through various interfaces and lines.

[0202] The memory may be used to store the computer program. The processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. Among them, the program storage area may store an operating system, application programs required for at least one function, etc.; the data storage area may store data created according to the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0203] Based on the embodiments of the above various power system flexibility resource scheduling methods, the present invention correspondingly provides embodiments of a storage medium item.

[0204] An embodiment of the present invention provides a storage medium, the storage medium includes a stored computer program, wherein, when the computer program runs, it controls the device where the computer-readable storage medium is located to execute a power system flexibility resource scheduling method described in any method item embodiment of the present invention.

[0205] The storage medium is a computer-readable storage medium, the computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0206] The above is the preferred implementation manner of the present invention. It should be noted that for those of ordinary skill in the art in the technical field, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. A flexibility resource scheduling method for a power system, characterized in that, Including: Performing error correction on the current wind power output according to the probability density function of wind power output error to generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; performing error correction on the current photovoltaic power output according to the probability density function of photovoltaic power output error to generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic power output; performing error correction on the current electricity load according to the probability density function of electricity load error to generate a load probability density function for quantifying the uncertainty distribution of the electricity load. Generating a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function. Generating a first dispatch quantity for responding to the upward flexibility demand of the power system and a second dispatch quantity for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic power output, the current electricity load, and the net load probability density function. When responding to the upward flexibility demand of the power system, dispatching the flexibility resources of the power system according to the first dispatch quantity; or, when responding to the downward flexibility demand of the power system, dispatching the flexibility resources of the power system according to the second dispatch quantity.

2. The flexibility resource scheduling method for a power system according to claim 1, wherein Before generating the first output probability density function and the second output probability density function, it further includes: Obtaining the current temperature and the current wind speed. When it is determined that the current temperature is not less than the preset temperature threshold or the current wind speed is not less than the preset wind speed threshold, inputting the current temperature, the current wind speed, the current wind power output, and the current photovoltaic power output into a preset output error prediction model, so that the output error prediction model extracts meteorological features and output features according to the current temperature, the current wind speed, the current wind power output, and the current photovoltaic power output; wherein, the meteorological features are used to characterize the dynamic change characteristics of the weather; the output features are used to characterize the operating characteristics of the power generation equipment. Based on the cross-attention mechanism, weighting and fusing the meteorological features and the output features to generate fused features; wherein, the fused features are used to characterize the coupling relationship between the weather and the operation of the power generation equipment. Generating the wind power output error corresponding to the current wind power output and the photovoltaic power output error corresponding to the current photovoltaic power output according to the fused features. Performing correction on the current wind power output according to the wind power output error to generate a corrected wind power output; performing correction on the current photovoltaic power output according to the photovoltaic power output error to generate a corrected photovoltaic power output. Taking the corrected wind power output and the corrected photovoltaic power output as the current wind power output for error correction according to the probability density function of wind power output error and the current photovoltaic power output for error correction according to the probability density function of photovoltaic power output error, respectively. When it is determined that the current temperature is less than the preset temperature threshold or the current wind speed is less than the preset wind speed threshold, no correction operation is performed on the current wind power output and the current photovoltaic power output.

3. A flexibility resource scheduling method for a power system according to claim 2, characterized in that, The training process of the output error prediction model includes: Generate a number of training samples based on temperature history samples, wind speed history samples, wind power output history samples, and photovoltaic output history samples; Iteratively train the model to be trained according to each training sample until the model converges, and output the trained output error prediction model; Among them, during each training, input a training sample into the output error prediction model, so that the output error prediction model outputs the wind power output error prediction result and the photovoltaic output error prediction result corresponding to the training sample; Compare the wind power output error prediction result and the photovoltaic output error prediction result corresponding to the training sample with the actual wind power output error result and the actual photovoltaic output error result of the training sample respectively, and adjust the network parameters of the output error prediction model according to the comparison result.

4. A flexibility resource scheduling method for a power system according to claim 3, characterized in that, It also includes: Obtain the installed capacity ratio of each wind turbine in the power system, the wind power output data of each wind turbine, the installed capacity ratio of each photovoltaic unit, the photovoltaic output data of each photovoltaic unit, and the electricity load data; Generate a number of wind power output intervals according to the installed capacity ratio of each wind turbine; among them, different wind turbine intervals correspond to different average wind power outputs; Use the kernel density estimation method to fit the error distribution of the wind power output data in each wind power output interval respectively, and generate a wind power output error probability density function; among them, the wind power output error probability density function is used to characterize the wind power output error probability distribution of different wind power output intervals; Generate a number of photovoltaic output intervals according to the installed capacity ratio of each photovoltaic unit; among them, different photovoltaic output intervals correspond to different average photovoltaic outputs; Use the kernel density estimation method to fit the error distribution of the photovoltaic output data in each photovoltaic output interval respectively, and generate a photovoltaic output error probability density function; among them, the photovoltaic output error probability density function is used to characterize the photovoltaic output error probability distribution of different photovoltaic output intervals; Divide the electricity load data into a number of electricity load intervals; among them, different electricity load intervals correspond to different average electricity loads; Use the kernel density estimation method to fit the error distribution of the electricity load data in each electricity load interval respectively, and generate an electricity load error probability density function; among them, the electricity load error probability density function is used to characterize the electricity load error probability distribution of different electricity load intervals.

5. The flexibility resource scheduling method for a power system according to claim 4, wherein The generating a net load probability density function for quantifying the net load uncertainty distribution according to the first output probability density function, the second output probability density function, and the load probability density function includes: Perform Monte Carlo sampling on the first output probability density function, the second output probability density function, and the load probability density function to generate a number of sampling data; among them, each sampling data includes: the wind power output to be processed, the photovoltaic output to be processed, and the electricity load to be processed; For each sampling data, generate a net load sample corresponding to the sampling data according to the wind power output to be processed, the photovoltaic output to be processed, and the electricity load to be processed in the sampling data; Based on the kernel density estimation method, each net load sample is fitted to generate a net load probability density function.

6. The flexibility resource scheduling method for a power system according to claim 5, characterized in that, The generating of a first scheduling quantity for responding to the upward flexibility demand of the power system and a second scheduling quantity for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic power output, the current power consumption load, and the net load probability density function includes: Integrating the net load probability density function to generate a net load cumulative distribution function; Generating a current net load prediction interval according to a preset confidence level and the net load cumulative distribution function; generating a first scheduling quantity and a second scheduling quantity according to the current wind power output, the current photovoltaic power output, the current power consumption load, and the current net load prediction interval.

7. A flexibility resource scheduling method for a power system according to claim 6, characterized in that, The current wind power output includes the wind power output corresponding to different moments within the day; the current photovoltaic power output includes the photovoltaic power output corresponding to different moments within the day; the current power consumption load includes the power consumption load corresponding to different moments within the day; the current net load prediction interval includes the upper limit value of the net load prediction corresponding to different moments within the day and the lower limit value of the net load prediction corresponding to different moments. The generating of a first scheduling quantity and a second scheduling quantity according to the current wind power output, the current photovoltaic power output, the current power consumption load, and the current net load prediction interval includes: Obtaining the current scheduling time period within the day; wherein, the scheduling time period includes a first time period or a second time period; the end moment of the first time period is less than the start moment of the second time period. Generating the net load prediction value corresponding to different moments within the day according to the current wind power output, the current photovoltaic power output, and the current power consumption load; When it is determined that the current scheduling time period is the first time period, taking the net load prediction value corresponding to the start moment of the first time period as the first net load prediction value; taking the next moment of the current moment corresponding to the first net load prediction value as the first target moment; generating a first scheduling quantity according to the difference between the first net load prediction value and the upper limit value of the net load prediction corresponding to the first target moment; generating a second scheduling quantity according to the difference between the first net load prediction value and the lower limit value of the net load prediction corresponding to the first target moment; When it is determined that the current scheduling time period is the second time period, taking the net load prediction values corresponding to each moment within the second time period as the second net load prediction values; taking the largest second net load prediction value as the first net load prediction value to be calculated, and taking the smallest second net load prediction value as the second net load prediction value to be calculated; Judging whether the moment corresponding to the second net load prediction value to be calculated is less than the moment corresponding to the first net load prediction value to be calculated; If so, taking the moment corresponding to the first net load prediction value to be calculated as the second target moment, generating a first scheduling quantity according to the difference between the upper limit value of the net load prediction corresponding to the second target moment and the second net load prediction value to be calculated; generating a second scheduling quantity according to the difference between the second net load prediction value to be calculated and the lower limit value of the net load prediction corresponding to the second target moment. Otherwise, use the moment corresponding to the second net load prediction value to be calculated as the third target moment, and generate a first dispatch volume according to the difference between the upper limit value of the net load prediction corresponding to the third target moment and the first net load prediction value to be calculated; generate a second dispatch volume according to the difference between the first net load prediction value to be calculated and the lower limit value of the net load prediction corresponding to the third target moment.

8. A flexibility resource scheduling device for a power system, characterized in that Including: A first probability density function generation module, a second probability density function generation module, a dispatch volume generation module, and a flexibility resource dispatch module; The first probability density function generation module is configured to correct the error of the current wind power output according to the wind power output error probability density function, and generate a first output probability density function for quantifying the uncertainty distribution of the current wind power output; correct the error of the current photovoltaic output according to the photovoltaic output error probability density function, and generate a second output probability density function for quantifying the uncertainty distribution of the current photovoltaic output; correct the error of the current electricity load according to the electricity load error probability density function, and generate a load probability density function for quantifying the uncertainty distribution of the electricity load; The second probability density function generation module is configured to generate a net load probability density function for quantifying the uncertainty distribution of the net load according to the first output probability density function, the second output probability density function, and the load probability density function; The dispatch volume generation module is configured to generate a first dispatch volume for responding to the upward flexibility demand of the power system and a second dispatch volume for responding to the downward flexibility demand of the power system according to the current wind power output, the current photovoltaic output, the current electricity load, and the net load probability density function; The flexibility resource dispatch module is configured to dispatch the flexibility resources of the power system according to the first dispatch volume when responding to the upward flexibility demand of the power system; or dispatch the flexibility resources of the power system according to the second dispatch volume when responding to the downward flexibility demand of the power system.

9. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a flexibility resource dispatch method for a power system as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the storage medium is located to execute a flexibility resource dispatch method for a power system as described in any one of claims 1 to 7.

Images