A load prediction and photovoltaic fluctuation power coordination control system for power plants

Abstract

This invention discloses a power plant power coordination control system for load forecasting and photovoltaic fluctuations. This invention relates to the field of power control system technology and solves the problem that existing power plant power coordination control only aims at instantaneous power amplitude balance, failing to incorporate the directional characteristics of load and photovoltaic rate of change into the coordination framework. This results in the inability to achieve pre-emptive interception and difficulty in identifying rotating rate of change coupling modes, leading to poor power oscillation suppression and low energy conduction efficiency. This invention constructs a two-dimensional state grid and introduces a virtual potential field, incorporating the directional characteristics of load power and photovoltaic power rate of change into a unified coordination framework. When a rapid increase in load and a rapid decrease in photovoltaic power occur simultaneously, the spin recognition module can identify the upward spinning mode and pre-direct the energy into the discharge queue, completing the energy storage response before the impact is transmitted to the grid connection point, thus achieving pre-emptive interception and avoiding the deficiency of traditional control systems that can only perform delayed adjustments after the impact occurs.

Description

Technical Field

[0001] This invention relates to the field of power control system technology, specifically to a power plant power coordination control system for load forecasting and photovoltaic fluctuations. Background Technology

[0002] With the large-scale grid connection of renewable energy, the penetration rate of photovoltaic power generation in the power system is constantly increasing. Due to the intermittent and fluctuating nature of photovoltaic output, its dynamic matching with load demand has become a key issue in the coordinated control of power plant power. Currently, power plants are usually equipped with energy storage systems to smooth out power fluctuations. Their control strategies are mostly aimed at balancing instantaneous power amplitude, that is, adjusting the output of energy storage to make the power at the grid connection point track the given command value.

[0003] However, existing power coordination control methods have the following shortcomings: First, traditional control only takes the equality of instantaneous power amplitude as the coordination target, and does not incorporate the load power change rate and photovoltaic power fluctuation change rate into a unified coordination framework. When a rapid increase in load and a rapid decrease in photovoltaic power occur simultaneously, or when a rapid decrease in load and a rapid increase in photovoltaic power occur simultaneously, even if the instantaneous power amplitude reaches equilibrium, the superimposed impact caused by the opposite directions of their change rates will still cause a drastic jump in power at the grid connection point, triggering grid frequency and voltage fluctuations. Traditional control can only perform delayed adjustment after the impact occurs, which is a post-event remedy and cannot achieve pre-event interception before the impact is transmitted. Second, existing methods cannot identify the rotational rate of change coupling mode formed by the alternating fluctuations of load and photovoltaic power. For example, when the load rises rapidly and the photovoltaic power falls rapidly, the rate of change vector shows a rotational trend. The power oscillations caused by such complex disturbances are difficult to suppress accurately, resulting in a high rate of misoperation and low energy conduction efficiency. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a power plant power coordination control system for load forecasting and photovoltaic fluctuations. This system solves the problem that existing power plant power coordination control only aims at balancing instantaneous power amplitudes and does not incorporate the directional characteristics of load and photovoltaic rate of change into the coordination framework. This results in the inability to achieve pre-emptive interception and difficulty in identifying rotating rate of change coupling modes, leading to poor power oscillation suppression and low energy conduction efficiency.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a power plant power coordination control system for load forecasting and photovoltaic fluctuations, comprising: The data mapping module is used to collect instantaneous values ​​of load power and photovoltaic power at fixed intervals, divide the power range into M and N intervals to construct a two-dimensional grid map, determine the current grid coordinates, and calculate the rate of change of power based on the previous and next time intervals to form a vector. The potential field channel module treats one boundary of the entire mesh as an energy storage receiving area, assigns initial potential energy to the mesh and iterates to a steady state, and traces to the energy storage boundary along the potential energy gradient descent direction to generate a unidirectional edge channel; The spin recognition module is used to calculate the rotation direction and intensity based on the multi-time rate of change vector and generate upspin or downspin tags. The package unblocking module is used to calculate the energy that needs to be adjusted based on the load power, photovoltaic power and grid connection point target power, encapsulate it with spin tags into an energy package, and inject the package into the corresponding unidirectional edge channel according to the current grid coordinates; The energy storage scheduling module is used to set up the upper and lower spin queues in the energy storage receiving area, store the arriving packages into the corresponding queues according to their spin tags, count the arrival rate of packages in each queue, and generate discharge power commands and charging power commands based on the arrival rate.

[0006] As a further aspect of the present invention: the data mapping module is specifically used for: Based on historical operating data, the minimum and maximum load power values ​​Plmin and Plmax, as well as the minimum and maximum photovoltaic power values ​​Pvmin and Pvmax, are statistically analyzed. The load power range is divided into M continuous and non-overlapping intervals, with the length of each interval denoted as dL. The photovoltaic power range is divided into N continuous and non-overlapping intervals, with the length of each interval denoted as dV, forming an M×N two-dimensional grid diagram. At each sampling time t, the current grid coordinates (i, j) are determined by the following formula based on the instantaneous values ​​of load power Pl(t) and photovoltaic power Pv(t) collected; The value of i is obtained by calculating the difference between Pl(t) and Plmin, dividing it by dL, rounding it down, and then adding 1. The value of j is obtained by calculating the difference between Pv(t) and Pvmin, dividing it by dV, rounding it down, and then adding 1. It is ensured that i is in the range of 1 to M and j is in the range of 1 to N. When the power value falls exactly on the interval boundary, the interval with the smaller index is selected.

[0007] As a further aspect of the present invention: the data mapping module is also used for: Based on the power value at the current time t and the power value at the previous time t-dt, calculate the load change rate rl(t) and the photovoltaic change rate rv(t), where rl(t) is the load power difference divided by the sampling period dt, and rv(t) is the photovoltaic power difference divided by the sampling period dt. The change rate vector r(t) is formed by rl(t) and rv(t). Centered on the current grid (i, j), consider all its existing neighboring grids, calculate the direction vector d = (i'-i, j'-j) from the current grid to each neighboring grid (i', j') and the direction matching degree cosθ. When |r(t)| = 0, take cosθ = 0. The path weights from the current grid to the adjacent grid are updated based on the direction matching degree and the rate of change vector. The update formula is Wnew=Wold+a×(cosθ×|r(t)|-b), where a is the learning rate coefficient and b is the decay constant. The updated weights are restricted to the range [0, Wmax].

[0008] As a further aspect of the present invention: the potential field channel module is specifically used for: Assign potential energy Φ(i,j) to each grid cell, take one boundary of the entire grid as the energy storage receiving area, fix its potential energy to 0, fix the potential energy of the other boundaries to 1, and set the initial potential energy of all internal grids to 1. The potential energy of each grid is calculated iteratively as the average potential energy of its four adjacent grids (up, down, left, and right) until the maximum change in all grids is less than a preset convergence threshold, thus obtaining a steady-state potential energy distribution.

[0009] As a further aspect of the present invention: the potential field channel module is also used for: Calculate the potential gradient Q(i,j) for each grid cell. Use central difference for internal grid cells and one-sided difference for boundary grid cells. Path tracing is performed along the gradient descent direction. At each step, the device moves to the grid with the lowest potential energy in the adjacent grid and is lower than the potential energy of the current grid. This process is repeated until the energy storage boundary with zero potential energy is reached. Each tracing grid sequence is recorded as an edge channel. All paths that start from different points but eventually converge to the same energy storage boundary entrance are merged to form a complete set of unidirectional edge channels.

[0010] As a further aspect of the present invention: the spin recognition module is specifically used for: Based on the rate of change vectors r(t-2dt), r(t-td), and r(t) of each grid at the three most recent moments, each rate of change vector is represented in complex form, with its real part being the load rate of change and its imaginary part being the photovoltaic rate of change. Calculate the change in argument between two consecutive time points, namely the change in argument dθ1 from t-2dt to t-dt and the change in argument dθ2 from t-dt to t, and adjust dθ1 and dθ2 so that their values ​​are between negative π and π. Calculate the average angular velocity V(t), which is the sum of dθ1 and dθ2 divided by twice the sampling period dt; if V(t) is greater than zero, it is marked as upspin; if V(t) is less than zero, it is marked as downspin; if the absolute value of V(t) is less than a preset threshold, it is marked as no rotation. Simultaneously calculate the rotational intensity S(t), which is the absolute value of V(t) multiplied by the magnitude of the current rate of change vector r(t).

[0011] As a further aspect of the present invention: the package unblocking module is specifically used for: Based on the current load power Pl(t), photovoltaic power Pv(t), and grid connection target power Pref, calculate the energy E(t) that needs to be adjusted at the current moment. Its value is Pl(t) minus Pv(t) minus Pref multiplied by the sampling period dt. Encapsulate E(t) with the spin tag output by the spin identification module and the current timestamp t into an energy package in the format {E, spin, tarrive}, where tarrive = t; Based on the current grid coordinates, obtain the corresponding channel in the unidirectional edge channel generated by the potential field channel module, inject the energy package into the channel, and record the channel number.

[0012] As a further aspect of the present invention: the energy storage scheduling module is specifically used for: In the energy storage receiving area, an upward spin queue and a downward spin queue are set up. When an energy package arrives at the channel exit, its spin tag is parsed. If it is an upward spin, it is inserted at the end of the upward spin queue; if it is a downward spin, it is inserted at the end of the downward spin queue; if it has no spin, it is discarded. Calculate the package arrival rates Rup(t) and Rdown(t) for each queue, as well as the accumulated energy.

[0013] As a further aspect of the present invention: the energy storage scheduling module is also used for: The discharge power command Pdist(t) = Pdist(t-dt) + Kup × (Rup(t) - Rup(t-dt)) is generated based on the arrival rate of the upward-spinning queue, and is limited to the range [0, Pdistmax]. The charging power command Pch(t) = Pch(t-dt) + Kdowan × (Rdown(t) - Rdown(t-dt)) is generated based on the arrival rate of the downspin queue and is limited to the range [0, Pchmax]. Kup and Kdown are proportional coefficients.

[0014] As a further aspect of the present invention: the energy storage scheduling module is also used for: Obtain the operating status of all available energy storage units within the energy storage receiving area, including the state of charge (SOCm), maximum allowable discharge power (Pdismaxm), maximum allowable charging power (Pchmaxm), and health status. For discharge commands, select healthy cells with SOCm higher than the lower threshold to form a discharge candidate set, sort them from high to low SOCm, and allocate discharge power according to the allocation ratio proportional to SOCm. At the same time, check whether each cell exceeds the maximum allowable discharge power. If it exceeds the limit, reallocate. For charging instructions, select healthy cells with SOCm below the upper limit threshold to form a charging candidate set, sort them from low to high SOCm, and allocate charging power according to the allocation ratio proportional to the remaining capacity (1-SOCm). At the same time, check whether the maximum allowable charging power is exceeded. If the limit is exceeded, reallocate the power. Finally, power command values ​​are sent to each energy storage unit to perform discharge or charging operations.

[0015] This invention provides a power plant power coordination control system for load forecasting and photovoltaic fluctuations. Compared with existing technologies, it has the following advantages: (1) By constructing a two-dimensional state grid and introducing a virtual potential field, this invention incorporates the directional characteristics of the rate of change of load power and photovoltaic power into a unified coordination framework. When the load rises rapidly and the photovoltaic power falls rapidly at the same time, the spinning mode can be identified by the spin identification module, and the energy package can be guided to the discharge queue in advance. The energy storage response is completed before the impact is transmitted to the grid connection point, thus achieving pre-interception and avoiding the defect of traditional control that can only perform delayed adjustment after the impact occurs. (2) By converting the load change rate and photovoltaic change rate at continuous moments into complex numbers and calculating the rotational angular velocity, this invention can accurately identify the rotational rate of change coupling mode formed by the alternating fluctuations of load and photovoltaic, effectively solving the problem that the existing technology is difficult to suppress the power oscillation caused by complex disturbances and reducing the error rate; (3) This invention constructs a unidirectional edge channel from each grid to the energy storage receiving area by solving the virtual potential field iteratively, so that energy can only flow unidirectionally along the predetermined path and cannot flow back. Combined with the processing method of diverting energy packages to different queues according to spin tags, it realizes refined energy diversion for different fluctuation characteristics, significantly improves energy diversion efficiency, and dynamically generates charging and discharging power commands according to the arrival rate of packages in each queue, so that the energy storage power smoothly follows the trend conflict changes of load and photovoltaic, avoids power mutation, ensures that the power at the grid connection point always remains stable, and improves the stability of grid frequency and voltage. Attached Figure Description

[0016] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1 Please see Figure 1 This application provides a power plant power coordination control system for load forecasting and photovoltaic fluctuations, comprising: The data mapping module is used to collect instantaneous values ​​of load power and photovoltaic power at fixed intervals, divide the power range into M and N intervals to construct a two-dimensional grid map, determine the current grid coordinates, and calculate the rate of change of power based on the previous and next time intervals to form a vector. The potential field channel module treats one boundary of the entire mesh as an energy storage receiving area, assigns initial potential energy to the mesh and iterates to a steady state, and traces along the potential energy gradient in the descent direction to the energy storage boundary to generate a unidirectional edge channel. The spin recognition module is used to calculate the rotation direction and intensity based on the multi-time rate of change vector and generate upspin or downspin tags. The package unblocking module is used to calculate the energy that needs to be adjusted based on the load power, photovoltaic power and grid connection point target power, encapsulate it with spin tags into an energy package, and inject the package into the corresponding unidirectional edge channel according to the current grid coordinates; The energy storage scheduling module is used to set up the upper and lower spin queues in the energy storage receiving area, store the arriving packages into the corresponding queues according to their spin tags, count the arrival rate of packages in each queue, and generate discharge power commands and charging power commands based on the arrival rate.

[0019] Example 2 Furthermore, based on Example 1, a power plant power coordination control system for load forecasting and photovoltaic fluctuations is described in more detail, specifically including: S1: Data from the power station is collected at a fixed sampling period dt, and the instantaneous values ​​of load power Pl(t) and photovoltaic power generation power Pv(t) are continuously obtained, where t is the current time; Based on the historical operating data of the power plant, the minimum and maximum load power Plmin and the minimum and maximum photovoltaic power Pvmin and Pvmax are calculated. The load power range is divided into M continuous and non-overlapping intervals, each with a length dL=(Plmax-Plmin) / M. The photovoltaic power range is divided into N continuous and non-overlapping intervals, each with a length dV=(Pvmax-Pvmin) / N. This forms an M×N two-dimensional network diagram. The grid cell (i, j) corresponds to the load power located in the i-th interval [Plmin+(i-1)dL, Plmin+i*dL] and the photovoltaic power located in the j-th interval [Pvmin+(j-1)dV, Pvmin+j*dV]. At each sampling time t, the collected Pl(t) is compared with the load interval boundary to determine its interval index i: i = floor[(Pl(t)) - Plmin / dL] + 1; Where floor is the floor function, which rounds down to the largest integer less than or equal to the value in parentheses. For example, floor[3.7]=3, and ensures that 1≤i≤M. Similarly, the interval index j to which the photovoltaic power belongs is calculated: j = floor[(Pv(t)) - Pvmin / dV] + 1; Finally, the grid coordinates (i, j) of the current running state point are obtained. If the power value happens to fall on the interval boundary, the interval with the smaller index is selected to ensure that each point is mapped to only one grid. Calculate the load change rate rl(t) and the photovoltaic change rate rv(t) using the power values ​​at the current time t and the previous time t-dt: rl(t) = [Pl(t) - Pl(t - dt)] / dt; rv(t) = [Pv(t) - Pv(t - dt)] / dt; Centered on the current grid (i, j), and based on all its adjacent grids, including up (i, j+1), down (i, j-1), left (i+1, j), right (i+1, j), top left (i-1, j+1), top right (i+1, j+1), bottom left (i-1, j-1), and bottom right (i+1, j-1), a total of 8 directions (if this point is a boundary, only adjacent grids are considered); For each adjacent grid (i', j'), calculate the direction vector d = (i'-i, j'-j) from the current grid to the adjacent grid and the direction matching degree cosθ = (r(t)×d) / (|r(t)|×|d|), where r(t) is the rate of change vector, r(t) = [rl(t), rv(t)]; If |r(t)|=0, then take cosθ=0; The value of the direction matching degree cosθ is between -1 and 1. The closer it is to 1, the more consistent the direction of the rate of change is with the direction of the path. Based on the orientation matching degree and the rate of change vector, update the path weight Wnew from the current grid to the adjacent grid. The path weight is initialized to a small positive number, and the update formula is: Wnew=Wold+a×(cosθ×|r(t)|-b), where a is the learning rate coefficient, b is the decay constant, and Wold represents the old path weight value from the current grid to a certain adjacent grid, that is, the weight set initially. The updated weights must be limited to the range [0, Wmax]. For example, Wmax = 1. If the updated weight is less than 0, it is set to 0. The larger the weight, the more the energy tends to flow along the path at the current rate of change.

[0020] S2: Assign a potential energy Φ(i, j) to each grid cell (i, j). Initially, the potential energy of all internal grid cells is 1. Consider one boundary of the entire grid graph (e.g., the left boundary) as the "energy storage receiving area" with its potential energy fixed at 0, and fix the potential energies of the right boundary (i = M), the upper boundary (j = N), and the lower boundary (j = 1) at 1. Set the initial value of the potential energy of all internal grids (i.e., grids not on any boundary, satisfying 1 < i < M and 1 < j < N) to 1; For each grid, calculate the average potential energy of its four adjacent points (up, down, left, and right) as the new potential energy of this grid. Finally, obtain the new potential energy set A1. Use the potential energy of the new potential energy set A1 as new data and calculate again to obtain the average potential energy of the four adjacent points (up, down, left, and right) of the current grid point as the new potential energy of this grid; Repeat the calculation with the newly obtained data as new data. When the maximum change amount Φmax of all grids is less than the preset convergence threshold, stop the calculation. After convergence, the potential energy change in the internal region is stable (small gradient), equivalent to an insulator, while the potential energy in the region near the energy storage boundary drops rapidly, forming a conductive edge; Calculate the potential energy gradient of each grid Q(i, j) = [Φ(i + 1, j) - Φ(i - 1, j), Φ(i, j + 1) - Φ(i, j - 1)]. For boundary grids, use one-sided differences: If the grid is on the left boundary (the first column), the horizontal component takes the potential energy of the grid's right neighbor minus the potential energy of the grid itself; for the vertical component, if the grid is not on the upper and lower boundaries, still use the central difference (i.e., the upper minus the lower), and adjust accordingly if it is on the boundary at the same time; If the grid is on the right boundary (the last column), the horizontal component takes the potential energy of the grid itself minus the potential energy of its left neighbor; If the grid is on the lower boundary (the first row), the vertical component takes the potential energy of the grid's upper neighbor minus the potential energy of the grid itself; for the horizontal component, use the central difference if it is not on the left and right boundaries; If the grid is on the upper boundary (the last row), the vertical component takes the potential energy of the grid itself minus the potential energy of its lower neighbor; For the four corner grids (e.g., the lower left corner), use the corresponding one-sided differences simultaneously: the horizontal direction takes the right minus itself, and the vertical direction takes the upper minus itself to obtain the gradient of this point; The opposite direction of the gradient, -Q(i,j), is the direction of energy flow. Starting from each grid, path tracing is performed along the gradient descent direction. At each step, the grid with the lowest potential energy in the adjacent grid and lower than the current grid's potential energy is moved. This process is repeated until the energy storage boundary (the grid with zero potential energy) is reached. Each traversed grid sequence is recorded as an edge channel. All paths that start from different points but eventually converge at the same energy storage boundary entrance are merged to form a complete set of unidirectional edge channels. Each channel has a unique entrance and exit, and the coordinate sequence of all grids on the channel is recorded.

[0021] S3: Based on each grid, obtain the rate of change vectors r(t-2dt), r(t-td), and r(t) for the three most recent moments. Treat each vector as a complex number: z = load change rate + j × photovoltaic change rate, where j is the imaginary unit; Calculate the change in amplitude between two consecutive moments: dθ1=arg[z(t-td) / z(t-2dt)]; dθ2=arg[z(t-td) / z(t-2dt)], where arg(·) is the principal function of the complex number, and the return value is in the interval (-π, π]. If the change of adjacent amplitude angles exceeds π, then 2π is subtracted for adjustment; if it is less than -π, then 2π is added for adjustment. Calculate the average angular velocity V(t) = (dθ1 + dθ2) / (2dt). If V(t) > 0, it is a counterclockwise rotation, marked as an upward rotation. If V(t) < 0, it is a clockwise rotation, marked as a downward rotation. If V(t) is less than a preset small threshold e, it is marked as no rotation. Simultaneously calculate the rotational intensity S(t) = |V(t)| × |r(t)|, where |r(t)| = √(rl) 2 +rv 2 ); Calculate the energy that needs to be adjusted at the current moment: E(t) = (Pl(t) - Pv(t) - Pref) × dt, where Pref is the target power at the grid connection point; If E(t)>0, it means that energy storage and discharge are needed to make up for the shortfall; if E(t)<0, it means that energy storage and charging are needed to absorb excess energy; if E(t)=0, no adjustment is needed. Encapsulate E(t) with the current spin tag and timestamp t into a data structure called the energy package, with the format {E, spin, tarrive}, where tarrive = t; Based on the current grid coordinates (i, j), obtain the corresponding unidirectional edge channel in the grid graph, use this channel as the injection channel for the energy envelope at the current moment, and record the channel number C.

[0022] S4: Set up an upper queue Qup and a lower spin queue Qdown in the energy storage receiving area. If an energy package arrives at the exit of a certain channel, parse its spin tag: If it is an upward spiral, insert the package at the end of the upward spiral queue; If it is a downward spiral, insert the package at the end of the downward spiral queue; If there is no twist, discard the package and proceed with further processing. Calculate the arrival rates Rup(t), Rdown(t), and cumulative energy for each queue; For the upward-spinning queue, the discharge command Pdist(t) is generated based on its arrival rate Rup(t), and its calculation formula is as follows: Pdist(t) = Pdist(t-dt) + Kup × (Rup(t) - Rup(t-dt)), and Pdist(t) is limited to the allowable discharge power range of the energy storage system [0, Pdistmax]. If the calculated Pdist(t) is less than 0, the value is 0. Similarly, for the downward-spinning queue, the charging power command Pch(t) is generated, and its calculation formula is as follows: Pch(t) = Pch(t-dt) + Kdowan × (Rdown(t) - Rdown(t-dt)), and is limited to the allowable charging power range [0, Pchmax]; Where Kup and Kdowan are proportionality coefficients (e.g., 0.5); Obtain the operating status of all available energy storage units within the energy storage receiving area, including the state of charge (SOCm) and maximum operating discharge power of each unit. Maximum allowable charging power And health status, where m=1,2,...,Mess is the cell number; For the discharge command Pdist(t): First, select cells whose SOCm is higher than the set lower threshold (e.g., 20%) and whose health status is normal to form a discharge candidate set D; Sort the cells in set D from high to low according to SOCm, and prioritize the cells with high state of charge to undertake the discharge task. The total discharge power is distributed proportionally to these units, with the proportionality coefficient being directly proportional to SOCm. The calculation formula is as follows: ; For each cell, check whether the allocated value exceeds its maximum allowable discharge power. If it does, set the maximum value and redistribute the excess power difference to other cells in set D that have not exceeded the limit. Repeat the adjustment until all cell allocation values ​​are within the limit or the remaining power is zero. Similarly, for the charging instruction Pch(t), cells with SOCm higher than the set lower threshold (e.g., 20%) and that are healthy are selected to form a charging candidate set C; Sort the cells by SOCm from low to high, and prioritize the cells with low state of charge to undertake the charging task. The allocation ratio is directly proportional to the remaining capacity (1-SOCm), and the calculation formula is as follows: ; Check if the maximum allowable charging power is exceeded; if so, reallocate the power. Finally, power command values ​​are sent to each energy storage unit. and Then, perform the corresponding discharge and charge operations.

[0023] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0024] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A power plant power coordination control system for load forecasting and photovoltaic fluctuations, characterized in that, include: The data mapping module is used to collect instantaneous values ​​of load power and photovoltaic power at fixed intervals, divide the power range into M and N intervals to construct a two-dimensional grid map, determine the current grid coordinates, and calculate the rate of change of power based on the previous and next time intervals to form a vector. The potential field channel module treats one boundary of the entire mesh as an energy storage receiving area, assigns initial potential energy to the mesh and iterates to a steady state, and traces along the potential energy gradient in the descent direction to the energy storage boundary to generate a unidirectional edge channel. The spin recognition module is used to calculate the rotation direction and intensity based on the multi-time rate of change vector and generate upspin or downspin tags. The package unblocking module is used to calculate the energy that needs to be adjusted based on the load power, photovoltaic power and grid connection point target power, encapsulate it with spin tags into an energy package, and inject the package into the corresponding unidirectional edge channel according to the current grid coordinates; The energy storage scheduling module is used to set up the upper and lower spin queues in the energy storage receiving area, store the arriving packages into the corresponding queues according to their spin tags, count the arrival rate of packages in each queue, and generate discharge power commands and charging power commands based on the arrival rate.

2. The power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The data mapping module is specifically used for: Based on historical operating data, the minimum and maximum load power values ​​Plmin and Plmax, as well as the minimum and maximum photovoltaic power values ​​Pvmin and Pvmax, are statistically analyzed. The load power range is divided into M continuous and non-overlapping intervals, with the length of each interval denoted as dL. The photovoltaic power range is divided into N continuous and non-overlapping intervals, with the length of each interval denoted as dV, forming an M×N two-dimensional grid diagram. At each sampling time t, the current grid coordinates (i, j) are determined by the following formula based on the instantaneous values ​​of load power Pl(t) and photovoltaic power Pv(t) collected; The value of i is obtained by calculating the difference between Pl(t) and Plmin, dividing it by dL, rounding it down, and then adding 1. The value of j is obtained by calculating the difference between Pv(t) and Pvmin, dividing it by dV, rounding it down, and then adding 1. It is ensured that i is in the range of 1 to M and j is in the range of 1 to N. When the power value falls exactly on the interval boundary, the interval with the smaller index is selected.

3. The power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 2, characterized in that, The data mapping module is also used for: Based on the power value at the current time t and the power value at the previous time t-dt, calculate the load change rate rl(t) and the photovoltaic change rate rv(t), where rl(t) is the load power difference divided by the sampling period dt, and rv(t) is the photovoltaic power difference divided by the sampling period dt. The change rate vector r(t) is formed by rl(t) and rv(t). Centered on the current grid (i, j), consider all its existing neighboring grids, calculate the direction vector d = (i'-i, j'-j) from the current grid to each neighboring grid (i', j') and the direction matching degree cosθ. When |r(t)| = 0, take cosθ = 0. The path weights from the current grid to the adjacent grid are updated based on the direction matching degree and the rate of change vector. The update formula is Wnew=Wold+a×(cosθ×|r(t)|-b), where a is the learning rate coefficient and b is the decay constant. The updated weights are restricted to the range [0, Wmax].

4. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The potential field channel module is specifically used for: Assign potential energy Φ(i,j) to each grid cell, take one boundary of the entire grid as the energy storage receiving area, fix its potential energy to 0, fix the potential energy of the other boundaries to 1, and set the initial potential energy of all internal grids to 1. The potential energy of each grid is calculated iteratively as the average potential energy of its four adjacent grids (up, down, left, and right) until the maximum change in all grids is less than a preset convergence threshold, thus obtaining a steady-state potential energy distribution.

5. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The potential field channel module is also used for: Calculate the potential gradient Q(i,j) for each grid cell. Use central difference for internal grid cells and one-sided difference for boundary grid cells. Path tracing is performed along the gradient descent direction. At each step, the device moves to the grid with the lowest potential energy in the adjacent grid and is lower than the potential energy of the current grid. This process is repeated until the energy storage boundary with zero potential energy is reached. Each tracing grid sequence is recorded as an edge channel. All paths that start from different points but eventually converge to the same energy storage boundary entrance are merged to form a complete set of unidirectional edge channels.

6. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The spin recognition module is specifically used for: Based on the rate of change vectors r(t-2dt), r(t-td), and r(t) of each grid at the three most recent moments, each rate of change vector is represented in complex form, with its real part being the load rate of change and its imaginary part being the photovoltaic rate of change. Calculate the change in argument between two consecutive time points, namely the change in argument dθ1 from t-2dt to t-dt and the change in argument dθ2 from t-dt to t, and adjust dθ1 and dθ2 so that their values ​​are between negative π and π. Calculate the average angular velocity V(t), which is the sum of dθ1 and dθ2 divided by twice the sampling period dt; if V(t) is greater than zero, it is marked as upspin; if V(t) is less than zero, it is marked as downspin; if the absolute value of V(t) is less than a preset threshold, it is marked as no rotation. Simultaneously calculate the rotational intensity S(t), which is the absolute value of V(t) multiplied by the magnitude of the current rate of change vector r(t).

7. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The package unblocking module is specifically used for: Based on the current load power Pl(t), photovoltaic power Pv(t), and grid connection target power Pref, calculate the energy E(t) that needs to be adjusted at the current moment. Its value is Pl(t) minus Pv(t) minus Pref multiplied by the sampling period dt. Encapsulate E(t) with the spin tag output by the spin identification module and the current timestamp t into an energy package in the format {E, spin, tarrive}, where tarrive = t; Based on the current grid coordinates, obtain the corresponding channel in the unidirectional edge channel generated by the potential field channel module, inject the energy package into the channel, and record the channel number.

8. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 1, characterized in that, The energy storage scheduling module is specifically used for: In the energy storage receiving area, an upward spin queue and a downward spin queue are set up. When an energy package arrives at the channel exit, its spin tag is parsed. If it is an upward spin, it is inserted at the end of the upward spin queue; if it is a downward spin, it is inserted at the end of the downward spin queue; if it has no spin, it is discarded. Calculate the package arrival rates Rup(t) and Rdown(t) for each queue, as well as the accumulated energy.

9. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 8, characterized in that, The energy storage scheduling module is also used for: The discharge power command Pdist(t) = Pdist(t-dt) + Kup × (Rup(t) - Rup(t-dt)) is generated based on the arrival rate of the upward-spinning queue, and is limited to the range [0, Pdistmax]. The charging power command Pch(t) = Pch(t-dt) + Kdowan × (Rdown(t) - Rdown(t-dt)) is generated based on the arrival rate of the downspin queue and is limited to the range [0, Pchmax]. Kup and Kdown are proportional coefficients.

10. A power plant power coordination control system for load forecasting and photovoltaic fluctuations according to claim 9, characterized in that, The energy storage scheduling module is also used for: Obtain the operating status of all available energy storage units within the energy storage receiving area, including the state of charge (SOCm), maximum allowable discharge power (Pdismaxm), maximum allowable charging power (Pchmaxm), and health status. For discharge commands, select healthy cells with SOCm higher than the lower threshold to form a discharge candidate set, sort them from high to low SOCm, and allocate discharge power according to the allocation ratio proportional to SOCm. At the same time, check whether each cell exceeds the maximum allowable discharge power. If it exceeds the limit, reallocate. For charging instructions, select healthy cells with SOCm below the upper limit threshold to form a charging candidate set, sort them from low to high SOCm, and allocate charging power according to the allocation ratio proportional to the remaining capacity (1-SOCm). At the same time, check whether the maximum allowable charging power is exceeded. If the limit is exceeded, reallocate the power. Finally, power command values ​​are sent to each energy storage unit to perform discharge or charging operations.

Images