Real-time load intelligent regulation and control method for runoff type cascade hydropower station group

Abstract

The invention belongs to the hydropower station dispatching technology, and particularly provides a real-time load intelligent regulation and control method for a runoff cascade hydropower station group, which comprises the following steps: dividing different hydropower stations into a dispatching mode and a non-dispatching mode according to the total load demand of the current hydropower station group; a hydropower station constraint condition, a hydropower unit constraint condition and system-level constraint adjustment are established; for a scheduling mode power station, predicting a load plan curve deviation in advance, matching three types of scheduling sub-models according to an economic operation strategy, and realizing a multi-target real-time optimization target of the hydropower station in different scenes; for a power station in a non-scheduling mode, balance of water level safety and operation economy is achieved through dynamic switching of the two types of models; and finally, carrying out real-time intelligent regulation and control on the load of the hydropower station based on a complex constraint normalization strategy and a mixed integer programming peak clipping model based on MILP.

Description

Technical Field

[0001] The present invention belongs to the technical field of hydropower station dispatching, and in particular relates to a real-time load intelligent control method for a run-of-river cascade hydropower station group. Background Art

[0002] The coordinated operation of cascaded hydropower stations presents challenges such as sensitive water level fluctuations (daily regulation of hydropower stations with fluctuations exceeding 10cm can trigger accidents), frequent load adjustments (requiring significant manual intervention), and delayed dispatch responses. Traditional dispatch models (such as pure generation plan optimization) are unable to adapt to the requirements of power system frequency stability and suffer from the "curse of dimensionality" when solving nonlinear multi-constraint models. Existing optimized dispatch technologies, which are mostly based on generation plan optimization, are unable to respond to grid load deviations in real time, resulting in low hydropower utilization. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a real-time intelligent load control method for a group of run-of-river cascade hydropower stations, which realizes high-precision automatic load adjustment and avoids the risk of water level exceeding the limit.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: a real-time load intelligent control method for a run-of-river cascade hydropower station group, comprising the following steps: Step 1: Divide the hydropower station group into dispatching mode and non-dispatching mode according to the total load demand of the current hydropower station group; Step 2: Establish hydropower station-level constraints, hydropower unit-level constraints, and system-level constraints; Step 3: Build a real-time load control model: For dispatching power plants, predict the load plan curve deviation in advance and match the economic dispatch model according to the economic operation strategy. For non-dispatching power plants, match the non-dispatching model based on the water level zoning. The non-dispatching model includes a normal water level maintenance model and a PID control model. Step 4: Real-time control of the hydropower station load.

[0005] In a preferred solution, in step 1, the dispatching mode is applicable to power stations that are allowed to accept control instructions 15 minutes in advance, and the non-dispatching mode is applicable to power stations that are not allowed to accept control instructions 15 minutes in advance.

[0006] In the preferred scheme, the hydropower station-level constraints in step 2 include water balance constraints, reservoir water level constraints, initial and final water level constraints, outflow constraints, power station output constraints, water level-reservoir capacity constraints, and tailwater level-discharge flow relationship; the hydropower unit-level constraints include unit output constraints, unit power generation flow constraints, unit vibration zone constraints, unit start-up and shutdown duration constraints, unit output ramp constraints, unit power generation head constraints, head loss function and unit dynamic characteristics relationship; system-level constraints include power balance constraints, transmission channel constraints, and power generation plan constraints.

[0007] In a preferred solution, in step three, the economic dispatch model includes a stable water level model, a small regulation optimization model and a water-based electricity adjustment model.

[0008] In the preferred solution, the applicable conditions of the stable water level model are: when the reservoir water level is within the control range or the high water level operation zone, but the inflow is less than the outflow; or when the water level is in the dead water level operation zone, but the inflow is greater than the outflow, the applicable conditions are expressed as: ; in, Z j,t For power stations j In time t The water level in front of the dam; Z h,t For power stations j High water level operating zone threshold; Z d,t For power stations j Dead zone operating zone threshold; Q in , j , t For power stations j In time t Inbound flow; Q out , j , t For power stations j In time t outbound flow; ∧ represents the logic "AND", V represents the logic "OR"; The stable water level model aims to minimize the change of water level in front of the dam and distributes the load among plants according to the flow balance. The objective function is: ; in: For a specific time period T The water level change; For the time period t Inner Power Station i changes in water levels; N is the number of power stations; The application conditions of the small adjustment optimization model are: ; in, f i, t is the power station i In time t Load adjustment frequency; F th is the load adjustment frequency threshold; P i, t is the power station i In time t Load adjustment amount; P th is the load adjustment threshold; The optimization goal of the small adjustment optimization model is to convert the comprehensive penalty amount into F Minimize, the objective function: ; in: k i is the penalty coefficient for power plant load adjustment; l i is the penalty coefficient of the power station load adjustment frequency; F The comprehensive penalty amount for all hydropower stations in the small regulation optimization model; N is the total number of hydropower stations in the cascade hydropower station group; T is the total number of scheduling time periods; The water-based electricity adjustment model is suitable for scenarios where the water level is stable but the water energy utilization rate needs to be optimized. The goal is to generate electricity with the highest efficiency. The objective function is: ; in: E The power generation of the power station; is the power station output coefficient; is the power station water flow of the power station in a specified period; The average net head height of the power station during a specified period.

[0009] In the preferred solution, in step 3, for a power station in non-dispatch mode, the water level zones include a high water level zone, a dead water level zone, and a normal zone. Thresholds corresponding to the three types of water level zones are defined. When the water level continuously deviates from the normal zone and there is no regression trend, the PID control model is triggered to adjust the water level. The applicable conditions and expressions of the normal water level model and PID control model are: 1) Normal water level maintenance model Applicable to the reservoir water level in the normal range, the expression is: ; When the water level is within the normal range, there is no need to adjust the load and maintain the current operating state. ; Where: is the water level in front of the dam at the current moment; / It is the lower / upper limit of the water level in the normal area; is the load adjustment; 2) PID control model When the water level continuously deviates from the normal area and has no regression trend, the expression is: or ; At this time, the expression of load adjustment is: ; in: e ( t ) represents the input deviation; Kp is the proportionality coefficient; Ki is an integer coefficient; Kd is the difference coefficient.

[0010] In a preferred solution, a correction parameter equation is used to correct the proportional coefficient of the PID control model: ; Where, K 0 is the baseline coefficient, obtained by NHQ curve fitting; Correction constant for flow rate change; is the flow change rate of the upstream power station; Upstream benchmark flow value; is the proportional coefficient of the PID control model.

[0011] In the preferred solution, in step four, a complex constraint normalization processing strategy and a mixed integer programming peak shaving model based on MILP are used to perform real-time intelligent control of the hydropower station load, a complex constraint normalization processing strategy is used to perform constraint normalization and determine the proportion function of the output of the hydropower station that needs to be adjusted to the total peak-shaving amount, and then the optimal load distribution plan is obtained according to the mixed integer programming peak shaving model based on MILP.

[0012] In a preferred solution, the complex constraint normalization processing strategy adopts a three-stage processing, and the operation steps are: 1) Constraint classification: The physical constraints of cascade power stations are divided into rigid constraints and elastic constraints; 2) Constraint standardization: Use dimensionless processing to unify the constraint dimensions: ; in: is the normalized constraint value, c i is the original constraint value, is the lower limit of the constraint value, is the upper limit of the constraint value; 3) Constructing a peak-shaving electricity distribution function for hydropower stations ; in ; Where: For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, It means that the sum of the peak load ratios of m hydropower stations is 1; , , is the weight coefficient, ; Indicates power station i During monitoring time Intra-temporal variation; and Power stations i Inbound / outbound traffic; For hydropower stations i During monitoring time Output value within; For hydropower stations i Output ratio; is the number of cascade hydropower stations, and the minimum number of upper hydropower stations is ,default =1, the maximum number of the lowest hydropower station is .

[0013] In a preferred solution, the objective function and constraints of the MILP-based mixed integer programming peak shaving model are: 1) The objective function is: ; in, For the future t The total peak-shaving electricity of all hydropower stations during the period, For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, Adjust output penalty coefficients for hydropower stations; 2) The constraints that must be met are as follows: future t Maximum water level fluctuation of the hydropower station during this period Less than the maximum limit : ; Hydropower Station (Total N Units j In the future t Number of starts and stops during a time period Less than the maximum limit : ; future t The output change rate of the hydropower station during the period is less than the hydropower station climbing limit : ; ; In the above formula: N Indicates the number of units; express t The maximum output of the hydropower station during the period, express t The minimum output of the hydropower station during the period.

[0014] The present invention provides a real-time intelligent load control method for a group of run-of-river cascade hydropower stations, which has the following beneficial effects: 1. By dividing the water level into dispatch mode and non-dispatch mode, an economic dispatch model and a non-dispatch model were constructed. The study used multi-objective dynamic programming to optimize water level control and combined it with mixed integer programming to reduce the difficulty of solving real-time load distribution.

[0015] 2. This method achieved high-precision automatic load adjustment in a pilot cascade hydropower station, avoiding the risk of overshooting water levels. The number of adjustments was reduced by 6%, reducing the workload of dispatchers. Furthermore, by dividing the system into high-water, normal, and dead-water zones, combined with the NHQ curve and dynamic PID parameter correction, it effectively addressed the problem of abrupt water level changes in small-capacity hydropower stations, improving the centralized control capabilities and economic operation of the watershed. This research finding has important theoretical and practical significance for promoting the intelligent development of real-time load scheduling in cascade hydropower stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 Flowchart of the present invention; Figure 2 This is the flow chart for real-time load control. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0018] Example 1: A real-time intelligent load control method for a run-of-river cascade hydropower station group comprises the following steps: Step 1: Divide the hydropower station group into dispatching mode and non-dispatching mode according to the total load demand of the current hydropower station group.

[0019] The dispatching mode is applicable to power plants that are allowed to accept control instructions 15 minutes in advance, and the non-dispatching mode is applicable to power plants that are not allowed to accept control instructions 15 minutes in advance.

[0020] In dispatch mode, the next phase of load calculations must be made 15 minutes in advance, and the system automatically matches the three sub-models in the economic operation strategy. The control objectives are to achieve economic load distribution and flow balance between power stations.

[0021] The non-dispatch mode triggers PID control through water level partitioning to prevent water level from exceeding the limit, distinguish the differences in the response capabilities of hydropower stations, and improve dispatching flexibility and adaptability.

[0022] Step 2: Establish hydropower station-level constraints, hydropower unit-level constraints, and system-level constraints.

[0023] The constraints of a hydropower station include water balance constraints, reservoir water level constraints, initial and final water level constraints, outflow constraints, power station output constraints, water level-reservoir capacity constraints, and the relationship between tailwater level and discharge flow, which are expressed as follows: (1) Water balance constraints ; in, , Respectively for scenes s Next i Hydropower stations in t +1 and t Storage capacity at a given moment; For the scene s Next i Hydropower stations during the period t Inbound flow; For the scene s Next i Hydropower stations during the period t Outbound flow; is the scheduling time interval; For the i -1 hydropower station to i Water flow stagnation time at each hydropower station; For the scene s Next, after considering the water flow delay i -1 hydropower station during the period Traffic volume; For the scene s Lower Power Station i -1 and power station i Between the time period t Interval flow; For the scene s Next i Hydropower stations during the period t Power generation flow; Representation scene s Next i Hydropower stations during the period t of discarded water flow.

[0024] (2) Reservoir water level constraints ; in, For the scene s Next i Hydropower stations in t The water level in front of the dam at the moment; 、 Respectively i Hydropower stations in t The lower and upper water level limits at each moment. The lower water level limit is often the dead water level, the lower limit of the dispatching line, etc., and the upper water level limit is the flood limit water level, the normal high water level, or the upper limit of the dispatching line. This constraint limits the fluctuation range of the water level.

[0025] (3) Initial and final water level constraints ; in, 、 are the initial water level and the controlled water level at the end of the scheduling period; is the allowable water level deviation at the end of the scheduling period to avoid affecting the water volume in the next scheduling period. is the allowable water level deviation at the end of the scheduling period.

[0026] (4) Outbound flow constraints ; in, For the scene s Next i Hydropower stations during the period t Outbound flow; 、 Respectively i Hydropower stations during the period t The lower limit of the outflow is usually determined by the ecological water demand, while the upper limit of the outflow is determined by the reservoir's discharge capacity. This constraint ensures that the reservoir's discharge flow is within a normal range.

[0027] (5) Power station output constraints ; in, For the i Hydropower stations during the period t contribution; 、 Respectively i Hydropower stations during the period t The lower and upper limits of output.

[0028] (6) Water level-reservoir capacity constraints ; in, For power stations i The nonlinear relationship curve function between water level and storage capacity of the reservoir.

[0029] (7) Tailwater level-discharge rate relationship ; in, For power stations i The nonlinear relationship curve function between tailwater level and discharge flow; For the scene s Lower Power Station i In the period t tailwater level.

[0030] The constraints of the hydropower stage group include unit output constraint, unit power flow constraint, unit vibration zone constraint, unit start-up and shutdown duration constraint, unit output ramp constraint, unit power head constraint, head loss function and unit dynamic characteristic relationship, which are respectively expressed as follows: (1) Unit output constraints ; in, 、 、 For the scene s Lower Power Station i No. n Units in the period t Output and its upper and lower limits; u i,n,t Indicates whether the unit is in operation. If the unit is in operation, u i,n,t =1 ； The unit is in shutdown state. u i,n,t =0 。

[0031] power station i The output can be expressed as: ; in, Indicates power station i The number of units. (2) Unit power generation flow constraints ; in, 、 、 For the scene s Lower Power Station i No. n Units in the period t Power generation flow and its upper and lower limits; power station i The power generation flow can be expressed as: ; (3) Unit vibration zone constraints ; in, 、 For power stations i No. n The first unit k The upper and lower limits of the output of each vibration zone.

[0032] (4) Unit start-up and shutdown duration constraints

[0033] in, For power stations i No. n Units in the period t The startup operation variable is 1, which means the startup operation; For power stations i No. n Units in the period t The shutdown operation variable is 1, which means the shutdown operation; 、 Power stations i No. n Minimum startup and shutdown duration of each unit; For power stations i No. n The maximum number of startup times of a unit within the scheduling cycle.

[0034] (5) Unit output ramp constraints ; in, For power stations i No. n The climbing ability of the unit.

[0035] (6) Unit generating head constraints ; in, 、 For the scene sLower Power Station i No. n Units in the period t The power generation head and head loss.

[0036] (7) Head loss function ; Where, 、 For power stations i Head loss coefficient and loss constant.

[0037] (8) Unit dynamic characteristics relationship ; Where, For power stations i No. n The nonlinear relationship function of output-head-flow of each unit.

[0038] System-level constraints: (1) Establish power balance constraints: ; Where, The time period for scene s t Channel output power; I Indicates the number of intelligent turbine units in a hydropower station. Indicates the turbine unit in the period t The average power.

[0039] (2) Establishing transport channel constraints: ; in, is the upper limit of the transmission channel power.

[0040] (3) Establish power generation plan constraints: ; Where, Indicates the day-ahead power generation plan issued by the power grid; A smaller deviation threshold is used to ensure the hydropower supply support capability to the power grid from the perspective of power balance and power supply stability.

[0041] Step 3: Build a real-time load control model: Figure 1 As shown in the figure, for dispatching mode power stations, the load planning curve deviation is predicted in advance, and the economic dispatch model is matched according to the economic operation strategy; for non-dispatching mode power stations, the non-dispatching model is matched based on the water level partitioning. The non-dispatching model includes a normal water level maintenance model and an abnormal water level regulation model, namely, a PID control model.

[0042] First, based on the ARIMA-LSTM hybrid model, the trend term of the historical load curve is decomposed ( R 2 ≥0.93) and period term (24 h ); then, the Pearson correlation coefficient is used to filter the associated power station data and construct the regional load characteristic matrix; finally, the extended Kalman filter (EKF) is used to implement online update of the prediction model parameters, and the prediction value is iteratively corrected every 15 minutes.

[0043] By predicting the deviation between the actual load and the planned curve (5%~15%), the output combination of hydropower units can be adjusted 1-4 hours in advance to avoid efficiency losses caused by temporary start-up and shutdown (frequent start-up and shutdown of units can reduce efficiency by up to 30%).

[0044] The economic dispatch model includes a stable water level model, a small regulation optimization model and a water-based electricity adjustment model.

[0045] (1) Stable water level model When the reservoir water level is within the control range or the high water level operation zone, but the inflow is less than the outflow, or when the water level is in the dead water level operation zone, but the inflow is greater than the outflow, under the premise of meeting various safety constraints, with the goal of minimizing the change in the water level in front of the dam, inter-plant load distribution is carried out according to flow balance, so as to achieve load and flow matching and stabilize the reservoir water level in front of the dam as much as possible.

[0046] The applicable conditions of the stable water level model are: when the reservoir water level is within the control range or the high water level operation zone, but the inflow is less than the outflow; or when the water level is in the dead water level operation zone, but the inflow is greater than the outflow.

[0047] The expressions for the applicable conditions are: ; in, Z j,t For power stations j In time t The water level in front of the dam; Z h,t is the high water level operation zone threshold of power station j; Z d,t is the dead zone operation zone threshold of power station j; Q in , j , t For power stations j In time t Inbound flow; Q out , j , t For power stations j In time toutbound flow; ∧ represents the logic "AND", V represents the logic "OR"; The stable water level model aims to minimize the change of water level in front of the dam and distributes the load among plants according to the flow balance. The objective function is: ; in: For a specific time period T The water level change; For the time period t Inner Power Station i changes in water levels; N is the number of power stations; Reasons for setting this sub-model in the scheduling model: From a safety and stability perspective: When the water level is in the high-water zone and the inflow is less than the outflow, the water level may continue to rise and exceed the limit. Conversely, when the water level is in the dead zone and the inflow is greater than the outflow, the water level may continue to fall and exceed the limit (as discussed in the background section of this specification regarding the sensitivity of water level fluctuations). This model prioritizes stabilizing the water level to avoid accidents.

[0048] Economic perspective: Reducing water level fluctuations can reduce the frequency of load adjustments, thereby reducing manual intervention.

[0049] (2) Small adjustment optimization model In practice, load fluctuations at certain power plants must be minimized to ensure minimal load variation assigned to them. Current output is achieved through inter-plant load distribution, with some power plants receiving more load and others receiving less. Based on inter-station economic operation scheduling, the amount and magnitude of load variations must be calculated for specific regulation targets, and penalty factors must be introduced accordingly. The ultimate optimization goal is to minimize the overall penalty amount to achieve the best regulation effect.

[0050] The small adjustment optimization model is used to reduce the frequency and amplitude of load adjustments and optimize economic efficiency. Its boundary conditions are based on the load fluctuation threshold and the current operating status (for example, it is applicable when the load adjustment frequency is too high).

[0051] The application conditions of the small adjustment optimization model are: ; in, f i, t is the power station i In time t Load adjustment frequency; F th is the load adjustment frequency threshold; For power stations i In time t The load adjustment amount; △Pth is the load adjustment threshold; The optimization goal of the small adjustment optimization model is to convert the comprehensive penalty amount into F Minimize, the objective function: ; in: k i is the penalty coefficient for power plant load adjustment; l i is the penalty coefficient of the power station load adjustment frequency; F The comprehensive penalty amount for all hydropower stations in the small regulation optimization model; N is the total number of hydropower stations in the cascade hydropower station group; T is the total number of scheduling time periods.

[0052] Analysis of the reasons for setting a small adjustment model in the scheduling model sub-model Economic perspective: When the load adjustment frequency or amplitude exceeds the threshold, frequent changes will increase penalty costs (such as equipment wear and scheduling intensity). The model optimizes economic efficiency by minimizing the penalty amount.

[0053] From the perspective of safety and stability: Reducing the number of adjustments can avoid frequent triggering of the unit vibration zone (such as the unit vibration zone constraint in step 2) and improve system stability.

[0054] (3) Water-based electricity adjustment model On the premise of meeting the water level control requirements of water conservancy comprehensive utilization departments at all levels and cascade power stations, the discharge flow of each cascade power station is calculated, striving to generate electricity with the highest efficiency, increase power generation, and improve the safe, stable and economic operation of the hydropower system.

[0055] The water-power adjustment model is used to maximize power generation efficiency within the safe water level range. It is suitable for scenarios where the water level is stable but the water energy utilization rate needs to be optimized. Its constraints are shown in Table 1.

[0056]

[0057] In order to calculate the discharge flow of each cascade power station while meeting the water level control requirements of water conservancy comprehensive utilization departments at all levels and cascade power stations, with the goal of generating electricity with the highest efficiency, the objective function is: ; in: E The power generation of the power station; is the power station output coefficient; is the power station water flow of the power station in a specified period; The average net head height of the power station during a specified period.

[0058] The boundary conditions of the three sub-models are mutually exclusive and cover all scheduling scenarios: the stable water level model prioritizes safety (when there is a risk of water level exceeding the limit), the small regulation optimization model prioritizes economy (when there are frequent load fluctuations), and the water-based power adjustment model prioritizes efficiency (when the water level is stable). This embodies the multi-objective optimization of the invention, ensuring timely, safe, and economical control responses in different scheduling scenarios.

[0059] For hydropower stations not in dispatch mode, regulation is based on the normal water level maintenance model and PID control model. For hydropower stations with small energy storage capacity and poor regulation capabilities, the water level is significantly affected by upstream water inflow. When the reservoir water level enters the high water level operating zone or the dead water level operating zone and shows no sign of returning to the operating zone, after a certain period of time, the hydropower station load will redistribute the water level based on the allocation results, bringing it as close as possible to the middle value of its operating zone, thereby achieving the goal of returning to the operating zone.

[0060] The applicable conditions and expressions of the normal water level model and PID control model are: 1) Normal water level maintenance model Applicable to the reservoir water level in the normal range, the expression is: ; When the water level is within the normal range, there is no need to adjust the load and maintain the current operating state. ; Where: is the water level in front of the dam at the current moment; / It is the lower / upper limit of the water level in the normal area; is the load adjustment amount.

[0061] The model maintains normal water level and keeps the load stable.

[0062] 2) PID control model When the water level continuously deviates from the normal area and has no regression trend, the expression is: or ; At this time, the expression of load adjustment is: ; in: e ( t ) represents the input deviation; Kp is the proportionality coefficient; Ki is an integer coefficient; Kd is the difference coefficient.

[0063] The PID controller, which incorporates proportional, integral, and differential components, has become a staple of industrial control due to its simple structure, excellent stability, reliable operation, and easy adjustment. By using the outflow from the dam, the water level above the dam, and the difference between the calculated and actual load values ​​as input, it can flexibly increase or decrease the operating load in real time, enhancing the robustness of the water level control effect at the dam.

[0064] Due to the varying water level differences from the operating reference level during abnormal water levels, unit load adjustments will also undergo different adjustments. Using the same PID adjustment parameters within the abnormal water level range is difficult to meet the adjustment requirements for different dam water levels, and may result in slow adjustment, large overshoot, and multiple oscillations.

[0065] The proportional coefficient of the PID control model is corrected using a correction parameter equation: ; Where, K 0 is the baseline coefficient, obtained by NHQ curve fitting; It is the flow change correction constant, which is obtained by dynamic adjustment based on the flow change rate of the upper power station, and is usually in the range of 0.1~0.3; is the flow change rate of the upstream power station; Upstream benchmark flow value; is the proportional coefficient of the PID control model.

[0066] This model is used for abnormal water level regulation to achieve rapid water level recovery and is applied to water level exceeding the limit during flood season and dead water level during dry season.

[0067] In non-dispatch mode, the normal water level model uses a "zero adjustment" strategy to reduce ineffective operations and minimize equipment wear. The abnormal water level model's PID control addresses time lag and nonlinearity in water level regulation. By modifying parameters to adapt to the hydraulic coupling characteristics of cascade power stations, control accuracy is improved. Dynamic switching between these two models achieves a balance between water level safety and operational economy.

[0068] Step 4: Real-time control of the hydropower station load.

[0069] Water level fluctuations at hydropower stations are influenced by numerous factors, such as upstream outflow, weather variations, and sensor measurement noise. Consequently, water level fluctuations are nonlinear, multi-constrained, and subject to time delays. By adjusting the load to cause changes in the dam water level, the dam water level self-adjusts to the operating range based on feedback and differential error states. By gradually approaching the optimal operating range, the effects of time delays and nonlinear links on the control process can be overcome. The regulation system possesses adaptive capabilities and can cope with the characteristic variations between the dam water level and the load. For run-of-the-river hydropower stations, water level control consists of two steps. First, the real-time change of the dam water level must be controlled toward the reference water level. Then, when the actual water level reaches the target water level, the water level must be kept stable. This is essentially a multi-objective scheduling problem. To address the "curse of dimensionality" in multi-objective dynamic programming, a cascade hydropower peak shaving method based on a complex constraint normalization strategy is proposed, and a mixed integer programming peak shaving model based on MILP is constructed.

[0070] A complex constraint normalization processing strategy and a mixed integer programming peak shaving model based on MILP are used to carry out real-time intelligent control of the hydropower station load. A complex constraint normalization processing strategy is used to perform constraint normalization and determine the proportion function of the hydropower station's output that needs to be adjusted to the total peak load. Then, the optimal load distribution plan is obtained according to the mixed integer programming peak shaving model based on MILP.

[0071] The complex constraint normalization processing strategy transfers the peak shaving effect of the upstream power station to the downstream power station by coordinating the storage and discharge timing relationship between cascade power stations. It adopts a three-stage processing method with the following operating steps: 1) Constraint classification: The physical constraints of cascade power stations are divided into rigid constraints and elastic constraints; 2) Constraint standardization: Use dimensionless processing to unify the constraint dimensions: ; in: is the normalized constraint value, c i is the original constraint value, is the lower limit of the constraint value, is the upper limit of the constraint value; 3) Constructing a peak-shaving electricity distribution function for hydropower stations ; in ; Where: For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, It means that the sum of the peak load ratios of m hydropower stations is 1; , , is the weight coefficient, ; Indicates power station i During monitoring time Intra-temporal variation; and are the inflow / outflow flows of power station i respectively; For hydropower stations i During monitoring time Output value within; For hydropower stations i Output ratio; is the number of cascade hydropower stations, and the minimum number of upper hydropower stations is ,default =1, the maximum number of the lowest hydropower station is .

[0072] The objective function and constraints of the MILP-based mixed integer programming peak shaving model are: 1) The objective function is: ; in, For the future t The total peak-shaving electricity of all hydropower stations during the period, For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, Adjust output penalty coefficients for hydropower stations; 2) The constraints that must be met are as follows: future t Maximum water level fluctuation of the hydropower station during this period Less than the maximum limit : ; Hydropower Station (Total N Units j In the future t Number of starts and stops during a time period Less than the maximum limit : ; future t The output change rate of the hydropower station during the period is less than the hydropower station climbing limit : ; ; In the above formula: N Indicates the number of units; express tThe maximum output of the hydropower station during the period, express t The minimum output of the hydropower station during the period.

[0073] The algorithm flow is as follows Figure 2 shown.

[0074] Step 1: Obtain parameters such as the power plant NHQ curve, water level and storage capacity curve, tailwater level and flow curve, and obtain real-time data of the power plant, including upper and lower water levels of the dam, inflow and outflow, and real-time load parameters. Calculate the future peak power distribution function of the hydropower station. t Peak-shaving power of each hydropower station during the time period.

[0075] Step 2: Preprocessing: Discretize the time window, for example, 15 minutes per period, to generate the feasible region of each hydropower station unit combination.

[0076] Step 3: Model conversion Convert nonlinear constraints into linear inequalities Step 4: Branch and bound solution Use the CPLEX / Gurobi solver to backcalculate the current power station outflow and the reservoir water level at the end of the time period based on the load distribution. Then, determine whether the load distribution meets the requirements based on the water level, ramp rate, and start-stop constraints. If not, return to step 1. If the above constraints are met, output the solution.

[0077] Step 5: Output the optimal load distribution plan and unit start and stop plan.

[0078] Example 2: A load regulation test was conducted using a cascade of hydropower stations as an example. Basic parameters for each station are shown in Table 2. Stations S1, S2, and S3 all operated in dispatch mode, while stations S4 and S5 operated in non-dispatch mode. A random scenario was selected: an emergency peak-shaving scenario in which a flood peak during the flood season causes a sudden 30% increase in inflow to the hydropower station, resulting in a 15% decrease in total load demand (from 550 MW to 467.5 MW). The real-time parameters of the example hydropower stations are shown in Table 3. Stations S1, S2, S3, and S4 initially adopted a water-to-power model, then transitioned to a stable water level and low-load regulation model to ensure that each station achieved load regulation and stable water level goals. Stations S4 and S5 operated in an abnormal water level regulation mode, before transitioning to a stable water level regulation mode. The load adjustments for the final five stations were, in order, reductions of 21.2 MW, 19.8 MW, 17.6 MW, 12.9 MW, and 11.0 MW.

[0079] For the annual operating indicators of cascade hydropower stations, as shown in Table 4, the real-time load control method given in this patent has significantly improved the real-time load regulation capability compared with the current method of the example hydropower station, with a relative increase of 4.7% in power generation capacity and a relative reduction of 26.2% in peak-shaving power shortage.

[0080]

[0081]

[0082]

[0083] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A real-time intelligent load control method for a run-of-river cascade hydropower station group, characterized in that: The following steps are involved: Step 1: Divide the hydropower station group into dispatching mode and non-dispatching mode according to the total load demand of the current hydropower station group; Step 2: Establish hydropower station-level constraints, hydropower unit-level constraints, and system-level constraints; Step 3: Build a real-time load control model: For dispatching power plants, predict the load plan curve deviation in advance and match the economic dispatch model according to the economic operation strategy. For non-dispatching power plants, match the non-dispatching model based on the water level zoning. The non-dispatching model includes a normal water level maintenance model and a PID control model. Step 4: Real-time control of the hydropower station load.

2. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 1, characterized in that: In step 1, the dispatching mode is applicable to power plants that are allowed to accept control instructions 15 minutes in advance, and the non-dispatching mode is applicable to power plants that are not allowed to accept control instructions 15 minutes in advance.

3. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 1, characterized in that: The hydropower station-level constraints in step 2 include water balance constraints, reservoir water level constraints, initial and final water level constraints, outflow constraints, power station output constraints, water level-reservoir capacity constraints, and tailwater level-discharge flow relationship; the hydropower unit-level constraints include unit output constraints, unit power generation flow constraints, unit vibration zone constraints, unit start-up and shutdown duration constraints, unit output ramp constraints, unit power generation head constraints, head loss function, and unit dynamic characteristic relationship; the system-level constraints include power balance constraints, transmission channel constraints, and power generation plan constraints.

4. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 1, characterized in that: In the step three, the economic dispatch model includes a stable water level model, a small regulation optimization model and a water-based electricity adjustment model.

5. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 4, characterized in that: The applicable conditions of the stable water level model are: when the reservoir water level is within the control range or the high water level operation zone, but the inflow is less than the outflow; or when the water level is in the dead water level operation zone, but the inflow is greater than the outflow. The expression of the applicable conditions is: ； in, Z j,t For power stations j In time t The water level in front of the dam; Z h,t For power stations j High water level operating zone threshold; Z d,t For power stations j Dead zone operating zone threshold; Q in , j , t For power stations j In time t Inbound flow; Q out , j , t For power stations j In time t Outbound flow; ∧ represents logical "and", V represents logical "or"; The stable water level model aims to minimize the change of water level in front of the dam and distributes the load among plants according to the flow balance. The objective function is: ； in: For a specific time period T The water level change; For the time period t Inner Power Station i changes in water levels; N is the total number of hydropower stations in the cascade hydropower station group; The application conditions of the small adjustment optimization model are: ； in, f i, t is the power station i In time t Load adjustment frequency; F th is the load adjustment frequency threshold; Pi,t For power stations i In time t Load adjustment amount; P th is the load adjustment threshold; The optimization goal of the small adjustment optimization model is to convert the comprehensive penalty amount into F Minimize, the objective function: ； in: k i is the penalty coefficient for power plant load adjustment; l i is the penalty coefficient of the power station load adjustment frequency; F The comprehensive penalty amount for all hydropower stations in the small regulation optimization model; N is the total number of hydropower stations in the cascade hydropower station group; T is the total number of scheduling time periods; The water-based electricity adjustment model is suitable for scenarios where the water level is stable but the water energy utilization rate needs to be optimized. The goal is to generate electricity with the highest efficiency. The objective function is: ； in: E The power generation of the power station; is the power station output coefficient; is the power station water flow of the power station in a specified period; The average net head height of the power station during a specified period.

6. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 1, characterized in that: In step 3, for a non-dispatch mode power station, the water level zones include a high water level zone, a dead water level zone, and a normal zone. Thresholds corresponding to the three types of water level zones are defined. When the water level continuously deviates from the normal zone and there is no regression trend, the PID control model is triggered to adjust the water level. The applicable conditions and expressions of the normal water level model and PID control model are: 1) Normal water level maintenance model Applicable to the reservoir water level in the normal range, the expression is: ； When the water level is within the normal range, there is no need to adjust the load and maintain the current operating state. ; Where: is the water level in front of the dam at the current moment; / It is the lower / upper limit of the water level in the normal area; is the load adjustment; 2) PID control model When the water level continuously deviates from the normal area and has no regression trend, the expression is: or ; At this time, the expression of load adjustment is: ； in: e ( t ) represents the input deviation; Kp is the proportionality coefficient; Ki is an integer coefficient; Kd is the difference coefficient.

7. A real-time intelligent load control method for a run-of-river cascade hydropower station group according to claim 6, characterized in that: The proportional coefficient of the PID control model is corrected using a correction parameter equation: ； Where, K 0 is the baseline coefficient, obtained by NHQ curve fitting; Correction constant for flow rate change; is the flow change rate of the upstream power station; Upstream benchmark flow value; is the proportional coefficient of the PID control model.

8. The method for real-time intelligent load control of a run-of-river cascade hydropower station group according to claim 1, characterized in that: In the fourth step, a complex constraint normalization processing strategy and a mixed integer programming peak shaving model based on MILP are used to perform real-time intelligent control of the hydropower station load. The complex constraint normalization processing strategy is used to perform constraint normalization and determine the proportion function of the output of the hydropower station that needs to be adjusted to the total peak-shaving amount. Then, the optimal load distribution plan is obtained according to the mixed integer programming peak shaving model based on MILP.

9. A real-time intelligent load control method for a run-of-river cascade hydropower station group according to claim 8, characterized in that: The complex constraint normalization processing strategy adopts a three-stage process, and the operation steps are: 1) Constraint classification: The physical constraints of cascade power stations are divided into rigid constraints and elastic constraints; 2) Constraint standardization: Use dimensionless processing to unify the constraint dimensions: ； in: is the normalized constraint value, c i is the original constraint value, is the lower limit of the constraint value, is the upper limit of the constraint value; 3) Constructing a peak-shaving electricity distribution function for hydropower stations ； in ; Where: For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, express m The sum of the peak load ratios of the hydropower stations is 1; , , is the weight coefficient, ; Indicates power station i During monitoring time Intra-temporal variation; and Power stations i Inbound / outbound traffic; For hydropower stations i During monitoring time Output value within; For hydropower stations i Output ratio; is the number of cascade hydropower stations, and the minimum number of upper hydropower stations is ,default =1, the maximum number of the lowest hydropower station is .

10. A real-time intelligent load control method for a run-of-river cascade hydropower station group according to claim 9, characterized in that: The objective function and constraints of the MILP-based mixed integer programming peak shaving model are: 1) The objective function is: ； in, For the future t The total peak-shaving electricity of all hydropower stations during the period, For hydropower stations i The output that needs to be adjusted accounts for the proportion of the total peak load, Adjust output penalty coefficients for hydropower stations; 2) The constraints that must be met are as follows: future t Maximum water level fluctuation of the hydropower station during this period Less than the maximum limit : ； Hydropower Station (Total N Units j In the future t Number of starts and stops during a time period Less than the maximum limit : ； future t The output change rate of the hydropower station during the period is less than the hydropower station climbing limit : ； ； In the above formula: N Indicates the number of units; express t The maximum output of the hydropower station during the period, express t The minimum output of the hydropower station during the period.

Images