Distributed photovoltaic scheduling method based on mirror image energy storage model

Abstract

The invention relates to a distributed photovoltaic scheduling method based on a mirror image energy storage model, and the method can achieve the analysis of the power generation condition and the transmission cost corresponding to a power user through the calculation of a flexible response value and a power generation load value, and accurately reflects the demands of a supply side. The optimal power supply scheduling task in the next time period can be screened out from the local perspective through the virtual energy storage model, so that the optimal power supply, energy storage and local power utilization time windows and power supply modes can be determined, and the generated electric energy can be used at a relatively high utilization rate; in addition, photovoltaic power generation electric energy scheduling strategies of different electricity users are judged through a game model, so that the optimal order of power supply scheduling tasks of the user is obtained based on selective evolution of other users, and the result output by the user each time is optimal.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and more specifically, to a distributed photovoltaic dispatching method based on a mirror energy storage model. Background Technology

[0002] Currently, energy storage technology is a crucial component of power management systems. With the popularization of distributed generation and energy storage technology, more and more power users are adopting local energy storage technology. On the one hand, it can make greater use of surplus electricity, and on the other hand, it can reduce the losses caused by power generation and transmission. Moreover, it can provide uninterrupted power supply output support when the power grid fails, thereby improving power efficiency, reducing the load on the power grid, and meeting the requirements of green economic development. Patent CN116760097A discloses a flexible grid-connected photovoltaic control system. This system generates output strategies for each photovoltaic module through power prediction, providing optimal scheduling methods for short-distance power supply and community power supply. However, in practical applications, it has been found that power generation users belonging to the same microgrid may experience power supply conflicts. This means that decisions may simultaneously be made to supply power to the community or to borrow power, as theoretically nearby photovoltaic modules may have similar power generation levels. This leads to the rapid fulfillment of community power supply needs in a short period, while excess power is re-planned, resulting in waste. Although the increased application of energy storage technology can solve some of the power redundancy problem through localized energy storage, new problems have arisen: 1. Increased power consumption or grid load. 2. Deterioration of power quality for users. 3. Insufficient energy storage capacity when needed, failing to meet user demand. The above reasons are all due to the inability to accurately configure the energy storage access time window. Since the electricity consumption of users is different, if a uniform energy storage strategy is configured, it will not be able to adapt to the actual needs of users, resulting in greater waste of resources. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a distributed photovoltaic scheduling method based on a mirror energy storage model.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is: a distributed photovoltaic scheduling method based on a mirror energy storage model, comprising:

[0005] Step S1: Obtain flexible response value and power generation load value;

[0006] Step S2: Input the flexible response value and the power generation load value into the pre-constructed virtual energy storage distribution model to generate several virtual output instruction sets. The virtual output instruction sets include several virtual output instructions. Each virtual output instruction corresponds to the power supply scheduling task of a photovoltaic power generation unit. The local cost value of the virtual output instruction set is calculated through a preset local cost algorithm.

[0007] Step S3: Capture the current user's energy storage redundancy value using a preset energy storage capture strategy;

[0008] Step S4: Input the virtual output instruction set and energy storage redundancy value into the preset comprehensive game model to update the instruction sequence in the virtual output instruction set, and calculate the scheduling cost of the virtual output instruction set through the preset scheduling cost algorithm.

[0009] Step S5: Generate a comprehensive evaluation value for the virtual output instruction set and the cost algorithm based on the local cost value and the scheduling cost value, and generate an output strategy based on the virtual output instruction set with the highest comprehensive evaluation value.

[0010] Further: Step S2 includes configuring a virtual model construction strategy, which comprises:

[0011] Step A1: Obtain the local user's power consumption architecture and power consumption information to configure the power consumption simulation architecture;

[0012] Step A2: Input the user's historical electricity consumption data into the electricity consumption simulation architecture to correct the electrical characteristic parameters of the electricity consumption simulation architecture;

[0013] Step A3: Virtually run the power supply scheduling task through the power consumption simulation architecture until all power supply scheduling tasks in the preset power supply scheduling task library are traversed, and power supply scheduling tasks are filtered through the preset horizontal coverage sub-strategy.

[0014] Step A4: Configure prediction parameters to calculate the comprehensive constraint vector of each power supply scheduling task under the power consumption simulation architecture, and construct the execution triggering network of the power supply scheduling task with the power supply scheduling task as the node based on the comprehensive constraint vector;

[0015] Step A5: Generate a demand boundary function for each power supply scheduling task based on the dynamic preference demand factor. The demand boundary function reflects the relationship between the execution time of the power supply scheduling task and the demand benefit. The dynamic preference demand factor reflects the user's current power demand for different types of power supply scheduling tasks. Configure the demand boundary function with the execution triggering network.

[0016] By setting it up in this way, a simulated power consumption architecture is constructed to simulate power consumption conditions. First, the power simulation architecture is corrected using historical power consumption data to make the physical model closer to the actual situation of the user. This way, when actual commands are input, feedback data close to reality can be obtained. The execution of power supply scheduling tasks is first filtered to remove power supply scheduling tasks that are not suitable for the current user environment, ensuring that the amount of data is small. Then, the power consumption simulation architecture is comprehensively judged by a comprehensive constraint vector to construct a corresponding execution trigger network so that the influence relationship between power supply scheduling tasks can be reflected in the model. The dynamic preference demand factor can reflect the benefits generated by each power supply scheduling task under the user demand item. In this way, all the downward influence relationship of power supply scheduling tasks can be digitized by this virtual model, so that the virtual model can accurately schedule power supply strategies.

[0017] Furthermore: the lateral coverage sub-strategy includes:

[0018] Step A3-1: Group different power supply scheduling tasks according to their objectives;

[0019] Step A3-2: Configure a virtual power consumption scenario in the power consumption simulation architecture, simulate power consumption conditions according to the virtual power consumption scenario, and execute power supply scheduling tasks to obtain feedback information from the virtual monitoring nodes;

[0020] Step A3-3: Configure several feedback anomaly evaluation items to generate corresponding feedback evaluation values ​​based on feedback information;

[0021] Step A3-4: In any group, compare the power supply scheduling tasks pairwise. If one power supply scheduling task meets the exclusion condition with another power supply scheduling task, remove the power supply scheduling task from the group until no two power supply scheduling tasks meet the exclusion condition.

[0022] By employing a horizontal coverage sub-strategy, power scheduling tasks that do not match the current user situation in each group are removed from the entire database. This means that other power scheduling tasks can fully cover the function of this power scheduling task in the model. By calculating the feedback evaluation value of each abnormal evaluation item, the power scheduling task is quantitatively evaluated, giving it an evaluation basis and greatly saving the data burden of computation.

[0023] Furthermore, the virtual energy storage distribution model generates comprehensive scheduling constraints based on the flexible response value and the power generation load value, selects different combinations of power supply scheduling tasks according to the comprehensive scheduling constraints to form a set of instructions to be output, and configures the duration of each power supply scheduling task according to the total scheduling duration in the comprehensive scheduling constraints to maximize demand benefits.

[0024] By generating comprehensive scheduling constraints to select tasks, and constructing specific duration selections for each instruction with the aim of maximizing benefits, the instruction set consisting of power supply scheduling tasks determined in this way can better meet actual usage needs.

[0025] Furthermore: the energy storage capture strategy includes

[0026] Step B1: Obtain battery state of charge, charge / discharge current, terminal voltage and temperature data and input them into the preset energy storage health assessment algorithm to form an energy storage health value.

[0027] Step B2: Obtain the current remaining energy storage value and calculate the corresponding energy storage redundancy value based on the energy storage health value. The energy storage capture strategy uses the energy storage health value and remaining energy storage value to determine whether the energy storage unit can support each power supply stage during the execution of the instruction set, ensuring that energy is not wasted.

[0028] Furthermore, the comprehensive game theory model includes configuring a decision function for each electricity user, and configuring a contention priority level for each corresponding scheduling relationship. The electricity user configuration decision function includes a benefit sub-function, a cost sub-function, and a contention loss sub-function. The benefit sub-function is used to calculate the scheduling benefit of the current virtual output instruction set, the cost sub-function is used to calculate the scheduling cost of the current virtual output instruction set, and the contention loss sub-function is used to calculate the loss of contention for a certain scheduling target. This process is iterated through each electricity user until the result of the electricity user configuration decision function for each electricity user is optimal. By configuring the electricity user configuration decision function in this way, and analyzing it through functions of benefit, cost, and contention, the power supply needs of each user are met under the optimal overall conditions.

[0029] Furthermore, the cost sub-function is constructed based on the energy storage redundancy value corresponding to the current electricity user.

[0030] Furthermore, step A3-3 also includes: the virtual power consumption scenario includes an event randomization module, which is used to generate random events in the power consumption simulation architecture to obtain feedback information under random events. By generating random events, the corresponding abnormal event capability is verified, and the safety and reliability of the module are ensured by testing extreme conditions.

[0031] Furthermore, the method includes step S6, configuring a feedback correction strategy. This strategy acquires the actual feedback data from each monitoring node, calculates the feedback deviation between the actual and predicted feedback data using a preset deviation correction algorithm, and corrects the prediction parameters of the virtual electricity consumption model based on the feedback deviation. The feedback correction strategy ensures that the actual feedback data remains within the preset feedback deviation, and the prediction parameters enable effective calculation of the feedback deviation, ensuring that the virtual model can be continuously optimized during actual operation to more closely approximate the actual situation.

[0032] The main technical effects of this invention are reflected in the following aspects: By setting it up in this way, the power generation situation and the transmission cost corresponding to the power user can be analyzed through the calculation of flexible response value and power generation load value, which can more accurately reflect the demand on the supply side. Furthermore, the virtual energy storage model can select the optimal power supply scheduling task in the next time period from the local perspective. This can determine the optimal time window and power supply mode for power supply, energy storage, and local power consumption, so that the generated electricity can be used with a high utilization rate. In addition, the game model is used to judge the scheduling strategy of photovoltaic power generation for different power users, and the optimal order of power supply scheduling tasks for the user is obtained based on the choices of other users, so that the output result of each user is optimal. Attached Figure Description

[0033] Figure 1 : Flowchart of a distributed photovoltaic scheduling method based on a mirror energy storage model according to the present invention;

[0034] Figure 2 : A flowchart of the virtual model construction strategy of this invention;

[0035] Figure 3 : A flowchart of the horizontal coverage sub-strategy of this invention;

[0036] Figure 4 : A flowchart of the energy storage and capture strategy of this invention;

[0037] Figure 5 : A block diagram of the strategy architecture of the comprehensive game theory model of this invention. Detailed Implementation

[0038] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, so that the technical solution of the present invention can be more easily understood and mastered.

[0039] Reference Figure 1-5 As shown, a distributed photovoltaic scheduling method based on a mirror energy storage model includes...

[0040] Step S1: Obtain flexible response value and power generation load value; First, regarding how to obtain the flexible response value and power generation load value, the predicted power generation value is input into the load calculation algorithm to obtain the power generation load value of each user's photovoltaic module; the load calculation algorithm is as follows: Among them, E a Let β1 be the power generation load value, ΔP be the historical power generation efficiency from historical power generation information, and β2 be the preset topology loss weight. Then β1 + β2 = 1, h i Let g be the power loss transmitted along the i-th transmission path in the user's photovoltaic module. i Let be the transmission quality loss ratio corresponding to the i-th transmission path in the user's photovoltaic module, and n be the total number of power transmission paths for the user's photovoltaic module. Historical power generation efficiency is calculated using the following formula: Where χ p Here are the preset power generation efficiency parameters, t0 is the current time, and t j Let T be the start time of the j-th generation waveform with the same timing characteristics. a For the first preset time, f j (x) represents the j-th generation waveform with the same timing characteristics, and m represents the total number of generation waveforms with the same timing characteristics. By calculating the load of each node under different power supply levels, the burden of excess power consumption on the entire system is calculated, thus providing a basis for determining whether to use this power consumption for local energy storage or direct power supply to the community. On the one hand, historical power generation data is acquired to calculate historical power generation efficiency and obtain the changes in actual power generation due to different locations or equipment, thus obtaining more accurate results. On the other hand, based on the burden of power transmission, the actual loss situation is judged, which is more reliable and accurate. The flexible calculation module is equipped with a flexible response algorithm, which is B = K4*Ec + K5*Eq / A4 + K6*Qf; where B is the flexible response value, Ec is the user's energy storage, Eq is the power generation quality, Qf is the floating power consumption, K4 is the preset fourth weight, K5 is the preset fifth weight, K6 is the preset sixth weight, and A4 is the preset fourth conversion value. In practical applications, K4 is set to 0.3, K5 to 0.4, K6 to 0.1, and A4 to 0.5. The user's energy storage Ec is obtained as 800kW, the power generation quality Eq as 200, and the floating power consumption Qf as 2000kW. Therefore, the flexible response value B for user 1 is calculated to be 600. The user's photovoltaic modules are configured with indoor power supply branches and community power supply branches. Indoor power supply tasks are set for the indoor power supply branches, and community power supply tasks are set for the community power supply branches. The floating power consumption calculation formula is as follows: Where ε1 is the preset indoor power supply weight, ε2 is the preset community power supply weight, and q laThe floating power supply value q corresponds to the la-th indoor power supply task. lb Let ka be the floating power supply value corresponding to the lb-th community power supply task, kb be the total number of indoor power supply tasks, and kb be the total number of community power supply tasks. The user data acquisition unit obtains the power consumption required for dynamic power consumption tasks through user uploads, marking it as floating power consumption. Dynamic power consumption tasks are power activities outside the scope of daily power consumption; for example, dynamic power consumption tasks might be power consumption tasks that require the user to maintain the power generation impact value corresponding to the indoor temperature or to heat water. Through interaction with the user-side intelligent platform, the floating power consumption corresponding to the dynamic power consumption tasks is obtained. When the amount of power output from the photovoltaic side to the grid is less, the power can be used for these floating tasks. The specific method has already been disclosed in publication number CN116760097A and will not be elaborated here.

[0041] Step S2: Input the flexible response value and the power generation load value into the pre-constructed virtual energy storage distribution model to generate several virtual output instruction sets. The virtual output instruction set includes several virtual output instructions, each corresponding to the power supply scheduling task of a photovoltaic power generation unit. The local cost value of the virtual output instruction set is calculated through a preset local cost algorithm. First, the virtual energy storage distribution model is a mirror simulation model built based on the user's actual power consumption architecture and the characteristics of energy storage equipment. It can simulate the power distribution, energy storage charging and discharging process, and load matching under different combinations of power supply scheduling tasks. It can accurately reflect the interaction logic between power consumption and energy storage in real scenarios. Technical personnel can build this model using conventional simulation tools such as MATLAB / Simulink. The virtual energy storage distribution model generates comprehensive scheduling constraints based on the flexible response value and the power generation load value. Based on the comprehensive scheduling constraints, it selects different combinations of power supply scheduling tasks to form the output instruction set. The duration of each power supply scheduling task is configured according to the total scheduling duration in the comprehensive scheduling constraints to maximize demand benefits. Next, the flexible response value and generation load value are input into the model to generate a virtual output instruction set, and the local cost value is calculated. The specific process is as follows: First, a comprehensive scheduling constraint calculation is performed, which is the core basis for generating the instruction set of the virtual energy storage distribution model. The comprehensive scheduling constraint needs to cover three dimensions: power supply capacity, energy storage adaptation, and load matching. The calculation method for the constraint value of each dimension is as follows: First, the power supply capacity constraint value P const1 The calculation formula is P const1 =α1×P load +α2×B, where α1 is the weight of the power generation load, α2 is the weight of the flexible response, and α1+α2=1, with values ​​ranging from 0<α1, α2<1. In the stable illumination region, α1=0.6, α2=0.4; in the fluctuating illumination region, α1=0.4, α2=0.6. load P is the power generation load value calculated in step S1, and B is the flexible response value calculated in step S1.const1 The physical meaning is the maximum power supply capacity threshold that the current photovoltaic power generation unit can stably provide. If the total power demand of subsequent power dispatch task combinations exceeds this threshold, then the combination does not meet the power supply capacity constraint. Secondly, there is the energy storage adaptation constraint value E. const2 The calculation formula is E const2 =β1×E ava +β2×(B×γ), where β1 is the weight of available energy storage capacity, β2 is the weight of energy storage correlation in flexible response, and β1+β2=1, usually β1=0.7, β2=0.3; E ava E represents the current available energy storage capacity; γ is the energy extraction coefficient in the flexible response value B, calculated as γ = K4 / (K4 + K5 + K6), where K4, K5, and K6 are the preset weights of the flexible response algorithm in step S1. const2 The physical meaning is the maximum energy consumption threshold of the power supply scheduling tasks that the current energy storage system can support. If the total energy consumption of the task combination exceeds this threshold, the energy storage adaptation constraint is not met. Thirdly, the load matching constraint value L... const3 The calculation formula is L const3 =|P load -L real |×δ, where L real The current actual electricity load of the user; δ is the load deviation penalty coefficient, ranging from 1.2 to 1.8. A value of 1.8 is used for scenarios with large load fluctuations, and 1.2 is used for scenarios with small fluctuations. L const3 The physical meaning is the degree of deviation between the power supply and the actual load, and a load deviation threshold L needs to be preset. th If L const3 More than L th If so, the task combination does not satisfy the load matching constraint.

[0042] Integrated scheduling constraint value C total The calculation formula is C total =P const1 ×E const2 / (L const3 +ε), where ε is the local minimum, and we take ε = 10. -6 Technicians need to preset the comprehensive constraint threshold C. th Community users set C th =800, Industrial Park User Setting C th =1500, when C total ≥C th If the corresponding power supply scheduling task combination meets the comprehensive scheduling constraints, it can be included in the instruction set to be output; otherwise, it will be removed.

[0043] The subsequent power supply dispatch task combination selection and duration configuration are then performed. The pre-set power supply dispatch task library must include all possible power supply tasks on the user side, categorized into indoor task groups and community task groups based on their objectives. Indoor task groups include examples such as "Indoor Air Conditioner Power Supply" and "Indoor Refrigerator Power Supply," while community task groups include examples such as "Community Streetlight Power Supply" and "Community Charging Pile Power Supply." Each task must be labeled with its rated power P. i Minimum execution time Maximum execution time For example, "Indoor air conditioning power supply" is marked with P. i =1.5kW "Community charging station power supply" is marked P i =6kW Tasks that meet the comprehensive scheduling constraints are selected from the task library to form multiple task groups. The execution duration of each task group is then configured based on the total scheduling duration to maximize the desired benefit. Total scheduling duration T total The timeframe is determined by technicians by combining the photovoltaic power generation period with the peak electricity consumption period of users, for example, setting T... total =6h. The demand revenue needs to be calculated using the demand boundary function, which is R. i (t i )=k i ×t i -m i ×t i 2 , where R i (t i ) is the execution time t of the i-th task. i The demand and benefit at that time, k i It is the profit coefficient of the i-th task, m i It is the decay coefficient of the reward for the i-th task, and m i =0.5×l i ×10 -2 ;l i Determined by the dynamic preference demand factor, F i It represents the user's preference for the task, calculated using the following formula: ω p The priority weight is preferably 0.6, ω n The necessary weight is preferably 0.4. It is the user-defined task priority, N i It is the task necessity coefficient, k i =F i ×k base k_base is the base revenue coefficient, set to 10. Total demand revenue R total =Σ(R) i (t i The constraint condition is Σ(t)i ) = T total And t i >0, solve for max(R) using the Lagrange multiplier method. total ), to obtain the optimal duration Xt for each task. i The duration configuration of the task combination is completed to form a virtual output instruction set. Each instruction in the instruction set corresponds to the power supply scheduling task of a photovoltaic power generation unit.

[0044] Next, the local cost of the virtual output instruction set is calculated using a preset local cost algorithm. The local cost includes the operating loss cost of the photovoltaic power generation unit, the charging and discharging loss cost of the local energy storage, and the task switching cost. The calculation formula is as follows: Where, ΔP i It is the power loss per unit time during the execution of the i-th task, determined by the photovoltaic module parameters. For monocrystalline silicon modules, it is typically taken as ΔP. i =0.2kW; η is the energy storage charging and discharging loss coefficient, taken as η=0.05, that is, 5% energy loss during charging and discharging. It is the amount of energy stored and charged when the i-th task is executed. For charging power, For charging time, λ is the energy storage and discharge amount when the i-th task is executed; λ is the task switching cost coefficient, which is taken as λ = 2, N switch This refers to the number of task switches within the instruction set. For easier subsequent comprehensive evaluation, the switching cost needs to be normalized to an equivalent energy consumption, making C... local The units are unified.

[0045] Reference Figure 2 As shown, step S2 configures a virtual model construction strategy, which includes:

[0046] Step A1, "Obtain the local user's power consumption architecture and power consumption information to configure the power consumption simulation architecture," refers to the following steps: The power consumption architecture includes the user's power supply branches, electrical equipment parameters, and energy storage device parameters; the power consumption information includes the user's daily power consumption periods, electrical equipment start-up and shutdown patterns, and historical power consumption fault records. When configuring the power consumption simulation architecture, the physical connection relationships of the power consumption architecture need to be converted into the topology of the simulation model, and the power consumption information parameters need to be entered into the corresponding node and branch attributes.

[0047] Step A2: Input historical user electricity consumption data into the electricity consumption simulation architecture to correct the electrical characteristic parameters of the simulation architecture. Historical electricity consumption data includes hourly power consumption data, hourly voltage data, and hourly current data. Electrical characteristic parameters include the resistance R, inductance L, capacitance C, and equivalent impedance of the equipment in the simulation architecture. The correction uses the least squares method, assuming a theoretical voltage U for a certain branch. sim =I sim ×R+L×dI sim / dt+(1 / C)×∫I sim dt, where I sim For simulated current, actual voltage U real Actual current I real Using historical data, construct the objective function F = Σ(U sim -U real ) 2 Taking partial derivatives with respect to R, L, and C and setting them to zero, we obtain the optimal parameter values ​​to replace the original parameters, making the deviation between the simulation output and the historical actual data ΔU=|U sim -U real | / U real The value is reduced by 100% to ensure the accuracy of the simulation architecture.

[0048] Reference Figure 3 As shown, step A3 involves virtually running the power supply scheduling task using a power consumption simulation architecture until all power supply scheduling tasks in the preset power supply scheduling task library are traversed. Power supply scheduling tasks are then filtered using a preset horizontal coverage sub-strategy. The horizontal coverage sub-strategy includes: first clarifying the feedback anomaly evaluation items and the calculation method for the feedback evaluation values; the feedback anomaly evaluation items include the power deviation rate ΔP. rate Voltage stability ΔU range Energy storage charging and discharging efficiency Task completion rate η complete The weights for each evaluation item are ω1 = 0.3, ω2 = 0.2, ω3 = 0.3, and ω4 = 0.2, respectively. The feedback evaluation value... S ranges from 0 to 100 points, with higher scores indicating better performance. The exclusion condition for the horizontal coverage sub-strategy is defined as follows: within the same group, if task A and task B satisfy any one of the following conditions, they are excluded: one is S. A ≤0.8×S B If the performance of task A is less than 80% of that of task B, and the performance of A in every other task is less than that of task B, then task B can be removed from the schedule.

[0049] Step A3-1: Group different power supply scheduling tasks according to their objectives;

[0050] Step A3-2: Configure a virtual power consumption scenario in the power consumption simulation architecture, simulate power consumption conditions according to the virtual power consumption scenario, and execute power supply scheduling tasks to obtain feedback information from the virtual monitoring nodes;

[0051] Step A3-3: Configure several feedback anomaly evaluation items to generate corresponding feedback evaluation values ​​based on feedback information;

[0052] Step A3-3 further includes: the virtual power consumption scenario includes an event randomization module, which is used to generate random events in the power consumption simulation architecture to obtain feedback information under random events.

[0053] Step A3-4: In any group, compare the power supply scheduling tasks pairwise. If one power supply scheduling task meets the exclusion condition with another power supply scheduling task, remove the power supply scheduling task from the group until no two power supply scheduling tasks meet the exclusion condition.

[0054] Step A4: Configure prediction parameters to calculate the comprehensive constraint vector of each power supply scheduling task under the power consumption simulation architecture. Based on the comprehensive constraint vector, construct the execution triggering network of the power supply scheduling task with the power supply scheduling task as the node. The prediction parameters include the solar irradiance G, hourly ambient temperature T, and hourly user electricity demand prediction value L for the next 24 hours. pred Comprehensive constraint vector in These are the predicted light intensity, ambient temperature, and power demand corresponding to the task execution period, respectively. The execution triggering network uses tasks as nodes. If the power supply margin after task A is executed can meet the power demand of task B, and the end of the execution time of A coincides with the start of the execution time of B, then a directed edge is established between nodes A and B. The weight of the edge is the task switching cost, forming a task network that can be executed sequentially.

[0055] Step A5: Generate a demand boundary function for each power supply scheduling task based on the dynamic preference demand factor. The demand boundary function reflects the relationship between the execution time of the power supply scheduling task and the demand benefit. The dynamic preference demand factor reflects the user's current electricity demand for different types of power supply scheduling tasks. Configure the demand boundary function with the execution triggering network. Calculate the demand boundary function parameters based on the dynamic preference demand factor F. i Reflecting the user's preference for the task, the benefit coefficient k is calculated using the aforementioned formula. i =F i ×k base The revenue decay coefficient m i =0.5×l i ×10 -2 This leads to the demand boundary function R. i (ti )=k i ×t i -m i ×t i 2 Configure this function in the corresponding node of the execution trigger network to supplement the node's demand and revenue calculation function.

[0056] Reference Figure 4 As shown, step S3 involves capturing the current user's energy redundancy value using a preset energy storage capture strategy; the energy storage capture strategy includes...

[0057] Step B1: Obtain battery state of charge, charge / discharge current, terminal voltage and temperature data and input them into the preset energy storage health assessment algorithm to form an energy storage health value.

[0058] Step B2: Obtain the current remaining energy storage value and calculate the corresponding energy storage redundancy value based on the energy storage health value.

[0059] This strategy is used to accurately assess the safe availability of redundant energy in an energy storage system. The core of this strategy is to first calculate the energy storage health value using battery operating parameters, and then derive the energy storage redundancy value by combining this with the remaining energy storage value. In step B1, the definitions of key parameters must be clearly defined: the battery state of charge (SOC) refers to the percentage of the battery's current remaining capacity relative to its rated capacity, ranging from 0 to 100%; the charge / discharge current (I) refers to the current value during the current charging and discharging process, positive during charging and negative during discharging; the terminal voltage (U) refers to the voltage between the positive and negative terminals of the battery; and the temperature (T) refers to the temperature of the battery's current operating environment. The energy storage health assessment algorithm calculates the energy storage health value (H) by weighted integration of the above parameters. The calculation formula is H = w1 × SOC + w2 × (U / U rated )×100+w3×(1-|I| / I rated )×100+w4×(1-|TT opt | / T range )×100, where w1, w2, w3, and w4 are weighting coefficients and their sum is 1. In practical applications, w1 = 0.4, w2 = 0.2, w3 = 0.2, and w4 = 0.2; U rated I is the rated terminal voltage of the battery. rated The rated charge / discharge current of the battery, commonly taken as 20A; T opt The optimal operating temperature for the battery is typically 25°C; T range It is half of the battery's allowable operating temperature range. If the battery's allowable operating temperature is 10℃ to 40℃, then T range =15℃. The energy storage health value H ranges from 0 to 100. A higher H value indicates a better health status of the energy storage system and a greater potential for safely available electrical energy. In step B2, the current remaining energy storage value E remThe current actual remaining electrical energy of the battery is directly collected by the energy storage system's power monitoring module; energy storage redundancy value E red The calculation formula is E red =E rem ×(H / 100) This formula reflects the impact of the energy storage health status on redundant electrical energy. The lower the health value, the less redundant electrical energy can be safely used, thus avoiding excessive use that could damage the energy storage equipment.

[0060] Step S4: Input the virtual output instruction set and energy storage redundancy value into the preset comprehensive game model to update the instruction sequence in the virtual output instruction set, and calculate the scheduling cost of the virtual output instruction set through the preset scheduling cost algorithm.

[0061] Reference Figure 5 As shown, the comprehensive game theory model includes configuring a decision function for each electricity user and configuring a contention priority level for each corresponding scheduling relationship. The electricity user's decision function includes a benefit sub-function, a cost sub-function, and a contention loss sub-function. The benefit sub-function calculates the scheduling benefit of the current virtual output instruction set, the cost sub-function calculates the scheduling cost of the current virtual output instruction set, and the contention loss sub-function calculates the loss of contention for a certain scheduling target. The model iterates through each electricity user until the result of the electricity user's configuration decision function is optimal. This model is used to coordinate scheduling resource conflicts among multiple users, making the updated virtual output instruction set more adaptable to the overall electricity consumption scenario. The comprehensive game theory model configures a decision function for each electricity user and configures a contention priority level for each corresponding scheduling relationship. The decision function includes a benefit sub-function, a cost sub-function, and a contention loss sub-function. The benefit sub-function calculates the scheduling benefit of the current virtual output instruction set, and the calculation formula is R. s =Σ(P i ×t i ×p), where P i The rated power of the i-th power supply scheduling task in the virtual output instruction set; t i is the execution duration of the task; p is the electricity price for the current time period. The cost sub-function is used to calculate the scheduling cost of the current virtual output instruction set, and is based on the energy storage redundancy value E corresponding to the current electricity user. red The construction and calculation formula is C. s =Σ[∫F(E) red -E qi )dΔE]×k,E qi Let F(.) be the energy storage value in the i-th power supply dispatching task, F(.) be the energy storage influence function, and ΔE = E red -E qi This reflects the impact of the energy storage redundancy difference on the energy storage unit, where k is the unit energy storage redundancy call-up cost. Contention loss sub-function L sThe formula for calculating the loss of contending for a certain scheduling target is L. s =Σ(Δr) i ×T i ), where the contention priority level is r, Δr i The difference T between the i-th power supply dispatching task and other dispatching tasks when they compete for the same time period. i To compete for time. Decision function F s The final expression is F s =R s -C s -L s F s A higher value indicates a better current scheduling scheme for that user. The iterative process of the comprehensive game model is as follows: first, initialize the virtual output instruction set for all electricity users, and then calculate their respective decision functions F. s Value; then, each user adjusts the execution period and power parameters of its own power supply scheduling task based on the current instruction set of other users, and recalculates F. s Value; repeat the above adjustment and calculation process until all users' F values ​​are reached. s The values ​​no longer increase, at which point the virtual output instruction set reaches its optimal state, completing the instruction sequence update. The cost sub-function is constructed based on the energy storage redundancy value corresponding to the current electricity user. The scheduling cost of the updated virtual output instruction set is calculated. The scheduling cost reflects the coordination cost of the instruction set during actual execution, and the calculation formula is C. sched =c1×N conflict +c2×Σ(P i ×t i ×d), where c1 and c2 are weighting coefficients and their sum is 1. In practical applications, c1 = 0.3 and c2 = 0.7 can be set; N conflict The number of scheduling conflicts between this instruction set and other user instruction sets is counted as one conflict if the power supply scheduling tasks of two users occupy the same transmission line in the same time period; d is the unit transmission loss rate of the transmission line, which is commonly taken as 0.05, i.e. 5%. This formula integrates the scheduling conflict cost and the transmission loss cost to comprehensively evaluate the actual execution cost of the scheduling scheme.

[0062] Step S5: Generate a comprehensive evaluation value from the virtual output instruction set and the cost algorithm based on the local cost value and the scheduling cost value, and generate an output strategy based on the virtual output instruction set with the highest comprehensive evaluation value. It should be noted that the output strategy in this invention differs from the output strategy in the publicly available document CN116760097A. The output strategy in this invention fully considers the role of the energy storage unit in the entire process. By configuring a mirror model of the entire entity built around the energy storage unit, the analysis of energy storage characteristics is completed. This ensures power quality while the output strategy determines not only the output path and target but also the output time period, thus enabling the full scheduling of power generation resources for application in subsequent time periods.

[0063] Step S6: A feedback correction strategy is configured. This strategy acquires the actual feedback data of each monitoring node, calculates the feedback deviation between the actual feedback data and the predicted feedback data using a preset deviation correction algorithm, and corrects the prediction parameters of the virtual power consumption model based on the feedback deviation. Monitoring nodes refer to key data acquisition points preset in the user's power consumption architecture and the virtual power consumption model, including current monitoring points of user power supply branches, charging and discharging power monitoring points of energy storage devices, and output voltage monitoring points of photovoltaic power generation units. Actual feedback data refers to physical quantity data collected in real time by the sensing devices of the monitoring nodes during the execution period of the virtual output instruction set, such as real-time branch current, real-time charging and discharging power of energy storage, and real-time output voltage of photovoltaics. Predictive feedback data refers to the predicted values ​​of the corresponding physical quantities output by the virtual power consumption model for each monitoring node during the same execution period, such as predicted branch current, predicted charging and discharging power of energy storage, and predicted output voltage of photovoltaics. This data must completely correspond to the acquisition period and physical quantity type of the actual feedback data to ensure the effectiveness of the comparison. The specific steps of the feedback correction strategy are as follows: First, acquire actual feedback data. Using sensors at monitoring nodes, continuously collect data from all monitoring nodes within the execution period of the virtual output instruction set at 1-minute intervals. Integrate the collected discrete data into an actual feedback dataset according to the time series. Second, acquire predictive feedback data. Extract predictive data from the historical output records of the virtual electricity consumption model that completely corresponds to the actual feedback data collection period and monitoring nodes, and integrate it into a predictive feedback dataset. Third, calculate the feedback deviation using a deviation correction algorithm. The formula for calculating the feedback deviation D is D = k 1×|AP| / A+k2×|AP|, where D is the feedback deviation of a single monitoring node at a certain acquisition time, A is the actual feedback data value at that time, P is the predicted feedback data value at that time, and k1 and k2 are deviation weighting coefficients with k1+k2=1. This formula considers both relative and absolute deviations simultaneously, avoiding errors caused by calculating a single deviation. The fourth step is to correct the prediction parameters. The prediction parameters of the virtual electricity consumption model include predicted values ​​of light intensity, predicted values ​​of user electricity demand, and predicted values ​​of energy storage charging and discharging efficiency. These parameters are corrected based on the feedback deviation D, and the correction formula is P. new =P old -k×D, where P new P is the corrected prediction parameter value. old The original predicted parameter values ​​are given by P, where k is the correction coefficient. After correction, P will be... new The virtual power consumption model is re-entered for the generation of subsequent virtual output instruction sets, enabling dynamic optimization of the model.

[0064] Of course, the above are just typical examples of the present invention. In addition, the present invention may have many other specific embodiments. All technical solutions formed by equivalent substitution or equivalent transformation fall within the scope of protection claimed by the present invention.

Claims

1. A distributed photovoltaic scheduling method based on a mirror energy storage model, characterized in that: include Step S1: Obtain flexible response value and power generation load value; Step S2: Input the flexible response value and the power generation load value into the pre-constructed virtual energy storage distribution model to generate several virtual output instruction sets. The virtual output instruction sets include several virtual output instructions. Each virtual output instruction corresponds to the power supply scheduling task of a photovoltaic power generation unit. The local cost value of the virtual output instruction set is calculated through a preset local cost algorithm. Step S3: Capture the current user's energy storage redundancy value using a preset energy storage capture strategy; Step S4: Input the virtual output instruction set and energy storage redundancy value into the preset comprehensive game model to update the instruction sequence in the virtual output instruction set, and calculate the scheduling cost of the virtual output instruction set through the preset scheduling cost algorithm. Step S5: Generate a comprehensive evaluation value for the virtual output instruction set and the cost algorithm based on the local cost value and the scheduling cost value, and generate an output strategy based on the virtual output instruction set with the highest comprehensive evaluation value.

2. The distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 1, characterized in that: Step S2 configures a virtual model construction strategy, which includes: Step A1: Obtain the local user's power consumption architecture and power consumption information to configure the power consumption simulation architecture; Step A2: Input the user's historical electricity consumption data into the electricity consumption simulation architecture to correct the electrical characteristic parameters of the electricity consumption simulation architecture; Step A3: Virtually run the power supply scheduling task through the power consumption simulation architecture until all power supply scheduling tasks in the preset power supply scheduling task library are traversed, and power supply scheduling tasks are filtered through the preset horizontal coverage sub-strategy. Step A4: Configure prediction parameters to calculate the comprehensive constraint vector of each power supply scheduling task under the power consumption simulation architecture, and construct the execution triggering network of the power supply scheduling task with the power supply scheduling task as the node based on the comprehensive constraint vector; Step A5: Generate a demand boundary function for each power supply scheduling task based on the dynamic preference demand factor. The demand boundary function reflects the relationship between the execution time of the power supply scheduling task and the demand benefit. The dynamic preference demand factor reflects the user's current power demand for different types of power supply scheduling tasks. Configure the demand boundary function with the execution triggering network.

3. The distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 2, characterized in that: The lateral coverage sub-strategy includes: Step A3-1: Group different power supply scheduling tasks according to their objectives; Step A3-2: Configure a virtual power consumption scenario in the power consumption simulation architecture, simulate power consumption conditions according to the virtual power consumption scenario, and execute power supply scheduling tasks to obtain feedback information from the virtual monitoring nodes; Step A3-3: Configure several feedback anomaly evaluation items to generate corresponding feedback evaluation values ​​based on feedback information; Step A3-4: In any group, compare the power supply scheduling tasks pairwise. If one power supply scheduling task meets the exclusion condition with another power supply scheduling task, remove the power supply scheduling task from the group until no two power supply scheduling tasks meet the exclusion condition.

4. The distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 3, characterized in that: The virtual energy storage distribution model generates comprehensive scheduling constraints based on flexible response values ​​and power generation load values. It selects combinations of different power supply scheduling tasks based on the comprehensive scheduling constraints to form a set of instructions to be output. It also configures the duration of each power supply scheduling task according to the total scheduling duration in the comprehensive scheduling constraints to maximize demand benefits.

5. The distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 1, characterized in that: The energy storage capture strategy includes Step B1: Obtain battery state of charge, charge / discharge current, terminal voltage and temperature data and input them into the preset energy storage health assessment algorithm to form an energy storage health value. Step B2: Obtain the current remaining energy storage value and calculate the corresponding energy storage redundancy value based on the energy storage health value.

6. The distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 1, characterized in that: The comprehensive game theory model includes configuring a decision function for each electricity user, and configuring a contention priority level for each corresponding scheduling relationship. The electricity user configuration decision function includes a benefit sub-function, a cost sub-function, and a contention loss sub-function. The benefit sub-function is used to calculate the scheduling benefit of the current virtual output instruction set, the cost sub-function is used to calculate the scheduling cost of the current virtual output instruction set, and the contention loss sub-function is used to calculate the loss of contention for a certain scheduling target. The model iterates through each electricity user until the result of the electricity user configuration decision function for each electricity user is optimal.

7. A distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 6, characterized in that: The cost sub-function is constructed based on the energy storage redundancy value corresponding to the current electricity user.

8. A distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 3, characterized in that: Step A3-3 further includes: the virtual power consumption scenario includes an event randomization module, which is used to generate random events in the power consumption simulation architecture to obtain feedback information under random events.

9. A distributed photovoltaic scheduling method based on a mirror energy storage model as described in claim 2, characterized in that: It also includes step S6, configuring a feedback correction strategy, wherein the feedback correction strategy acquires the actual feedback data of each monitoring node, calculates the feedback deviation between the actual feedback data and the predicted feedback data through a preset deviation correction algorithm, and corrects the prediction parameters of the virtual power consumption model according to the feedback deviation.

Images