Maintenance plan optimization method and equipment for new energy consumption power system

Abstract

The invention relates to a maintenance plan optimization method and equipment for a new energy consumption power system, and the method comprises the steps: carrying out the prediction of new energy power generation through different prediction models, and outputting an optimized maintenance plan according to a prediction result and a target function for maximizing the new energy consumption and minimizing the maintenance cost, and the optimized maintenance plan accords with a preset constraint condition. Meanwhile, an optimization algorithm is combined for solving, a maintenance plan can be adjusted in real time according to a prediction result of new energy power generation, the scientificity and feasibility of the plan are improved, potential risks during implementation of the maintenance plan are reduced, and the problems that power grid stability is out of control during maintenance and new energy power generation cannot effectively consume efficiency are solved. The defect that the maintenance plan is adjusted according to expert experience traditionally is overcome.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a method and device for optimizing a maintenance plan for a power system absorbing new energy. Background Art

[0002] With the vigorous advocacy and development of clean energy around the world, the proportion of new energy in the power grid continues to rise. New energy (such as wind energy, solar energy, etc.) has the significant advantages of being clean and sustainable, but its volatility, intermittency and uncertainty have brought many severe challenges to the safe and stable operation of the power grid. With the large-scale access of new energy to the power grid, the difficulty of power generation forecasting has become extremely difficult. The error in accurately estimating the power generation of new energy is large, which makes it difficult to effectively maintain the power balance of the power grid, which in turn seriously affects the stability of the power grid. At the same time, the continuous growth of electricity demand and the rapid development of social economy have promoted the vigorous development of various industrial production activities. The improvement of residents' living standards has also made the use of household appliances more extensive and frequent, which has led to a continuous upward trend in the load demand of the power grid. Under such complex circumstances, the daily dispatching work of the power grid has become extremely complicated. Dispatchers need to find a balance between the uncertainty of new energy generation and the growth of electricity demand. Every decision is full of risks and challenges. The reasonable arrangement of maintenance plans should be a key link to ensure the normal operation of power grid equipment and improve the reliability of the power grid, but it is difficult in this situation. On the one hand, due to the instability of renewable energy power generation, the uncertainty of the power grid operation status has increased, and the time period available for safety maintenance has become scarce and scattered. On the other hand, most of the current maintenance plan formulation methods for power systems in my country still rely on expert experience. This traditional model may have played a certain role in the past, but it is gradually unable to cope with the complex power grid environment where renewable energy is connected on a large scale. The operation of the power grid involves many complex factors, such as the coordination and complementarity between different energy sources, the diversified operating characteristics of power equipment, and the dynamic changes in the topology of the power grid. It is difficult to comprehensively and meticulously consider every key factor by relying solely on the subjective judgment of experts. This may lead to a series of problems during the implementation of the maintenance plan due to the inability to accurately estimate various potential risks, such as the loss of control of the stability of the power grid during the maintenance period, the inability to effectively absorb renewable energy power generation, etc., which in turn poses a serious threat to the safe and stable operation of the power grid, affects the quality and reliability of power supply, and brings many inconveniences and losses to social production and residents' lives. Summary of the invention

[0003] In order to at least overcome to some extent the problem in related technologies that power system maintenance plans formulated solely based on the subjective judgment of experts cannot accurately estimate various potential risks, the present application provides a maintenance plan optimization method and equipment for a new energy consumption power system.

[0004] The scheme of this application is as follows:

[0005] According to a first aspect of an embodiment of the present application, a method for optimizing a maintenance plan of a new energy consumption power system is provided, comprising:

[0006] Acquire real-time data of renewable energy power generation; the real-time data of renewable energy power generation includes: real-time data of wind power generation and real-time data of photovoltaic power generation;

[0007] Inputting the real-time wind power generation data into a pre-trained first prediction model to obtain a wind power generation prediction result;

[0008] Inputting the photovoltaic power generation real-time data into a pre-trained second prediction model to obtain a photovoltaic power generation prediction result;

[0009] Inputting the wind power generation prediction result and the photovoltaic power generation prediction result into a pre-trained maintenance plan optimization model, and outputting an optimized maintenance plan;

[0010] The objective function of the maintenance plan optimization model includes: maximizing new energy consumption and minimizing maintenance costs;

[0011] The constraints of the maintenance plan optimization model include: renewable energy generation constraints, load balance constraints, equipment maintenance time window constraints, equipment association constraints and energy storage equipment constraints;

[0012] The maintenance plan optimization model solves the objective function using an optimized genetic algorithm or an optimized particle swarm algorithm.

[0013] Preferably, the method further comprises:

[0014] The total power generation of the wind farm at time t is calculated as:

[0015]

[0016] Among them, P ω (t) represents the total power generation of the wind farm at time t, n is the total number of wind turbines in the wind farm, v i (t) represents the wind speed of the i-th wind turbine at time t, P ω,i (v) represents the power characteristic curve of the i-th wind turbine at time t;

[0017] Calculate the power consumption of the wind farm in the time interval [T1, T2]:

[0018]

[0019] Among them, E ω It represents the amount of electricity consumed by the wind farm in the time interval [T1, T2], and d is the set parameter;

[0020] The total power generation of the photovoltaic array at time t is calculated as:

[0021] P p (t) = ηI(t)A;

[0022] Among them, P P (t) represents the total power generation of the photovoltaic array at time t, I(t) represents the light intensity received by the photovoltaic array at time t, η represents the conversion efficiency of the photovoltaic array, and A represents the area of ​​the photovoltaic array;

[0023] Calculate the power consumption of the photovoltaic array in the time interval [T1, T2]:

[0024]

[0025] Among them, E P It represents the amount of electricity consumed by the photovoltaic array in the time interval [T1, T2], and d is the set parameter;

[0026] The total amount of new energy consumption is obtained based on the total amount of wind farm consumption and the total amount of photovoltaic array consumption:

[0027] E ren =E ω +E P ;

[0028] Then the objective function of maximizing the consumption of new energy is expressed as maximizeE ren .

[0029] Preferably, the method further comprises:

[0030] Calculate the cost of overhauling a single piece of equipment:

[0031] C j =c m,j +c l,j +c u,j *t out,j ;

[0032] Among them, C j represents the maintenance cost of equipment j, C m,j represents the maintenance material cost of equipment j, C l,j represents the labor cost of equipment j, C u,j represents the power outage loss cost per unit time of device j, t out,j Indicates the power outage time;

[0033] Calculate the maintenance cost of all equipment:

[0034]

[0035] Where m is the total number of devices;

[0036] Then the objective function of minimizing the maintenance cost is expressed as minimizeC;

[0037] The objective function of the maintenance plan optimization model is expressed as F = ω1E ren -ω2C;

[0038] Among them, F represents the objective function value, ω1 and ω2 represent the weight coefficients of maximizing new energy consumption and minimizing maintenance costs, respectively.

[0039] Preferably, the method further comprises:

[0040] According to the technical parameters of the wind turbine and the local wind speed historical statistical data, the upper and lower limits of wind power generation are set as the constraints of wind power generation:

[0041] P ω,min ≤P ω (t)≤P ω,max ;

[0042] Among them, P ω,min Represents the lower limit of wind power generation in the wind farm, P ω,max Indicates the upper limit of wind power generation capacity of the wind farm;

[0043] According to the characteristics of photovoltaic modules and the range of light conditions, the upper and lower limits of photovoltaic power generation are set as the constraints of photovoltaic power generation:

[0044] P p,min ≤P p (t)≤P p,max ;

[0045] Among them, P P,min Indicates the lower limit of photovoltaic power generation of photovoltaic modules, P P,max Indicates the upper limit of photovoltaic power generation of photovoltaic modules;

[0046] Set the total power generation of wind power and photovoltaic power generation to no less than the minimum expected value P ren,min As a constraint on renewable energy generation:

[0047] P ω +P P ≥P ren,min ;

[0048] Determine the load balancing constraint based on the total generated power:

[0049] P ω (t)+P p (t)+P trad (t) = L(t ) ;

[0050] Among them, P trad (t) represents the power generated by traditional energy, and L(t) represents the total power generated;

[0051] Determine the equipment maintenance time window constraint based on the earliest and latest allowable maintenance times of the equipment:

[0052] t start,j ≤t j ≤t end,j ;

[0053] Among them, t start,j represents the earliest allowable maintenance time of equipment j, t en d ,j represents the latest allowable maintenance time of equipment j, t j represents the maintenance time of equipment j;

[0054] The equipment association constraints include: maintenance order constraints, simultaneous maintenance constraints and non-simultaneous maintenance constraints;

[0055] The maintenance sequence constraint is expressed as

[0056] Among them, j1 and j2 are devices connected in series. Indicates the latest permissible maintenance time of equipment j1. Indicates the earliest allowed maintenance time of equipment j2;

[0057] The simultaneous maintenance constraint is expressed as t start,j3 =t start,j4 and t en d ,j3 =t en d ,j4 ;

[0058] Among them, j3 and j4 represent the equipment that must be repaired at the same time. Indicates the earliest allowed maintenance time of equipment j2. Indicates the earliest permissible maintenance time of equipment j4. Indicates the latest allowed maintenance time of equipment j3. Indicates the latest allowed maintenance time of equipment j4;

[0059] The constraint that no simultaneous maintenance can be performed is expressed as t end,j5 <t end,j6 or end,j6 <t end,j5 ;

[0060] Among them, j5 and j6 represent equipment that cannot be repaired at the same time. Indicates the latest allowed maintenance time of equipment j5. Indicates the latest permissible maintenance time of equipment j6;

[0061] Energy storage device constraints include:

[0062] 0≤P ch (t)≤P ch,max ;

[0063] 0≤P dis (t)≤P dis,max ;

[0064] S min ≤S(t)≤S max ;

[0065] Among them, P ch (t) represents the charging power of the energy storage device at time t, P ch,max Indicates the maximum charging power of the energy storage device, P dis (t) represents the discharge power of the energy storage device at time t, P dis,max represents the maximum discharge power of the energy storage device, S(t) represents the current energy storage capacity of the energy storage device, S min Represents the minimum energy storage capacity of the energy storage device, S max Indicates the maximum energy storage capacity of the energy storage device.

[0066] Preferably, the method further comprises:

[0067] Adaptive crossover probability P c and mutation probability P m , the genetic algorithm is optimized by optimizing the chromosome encoding:

[0068]

[0069] Where P c1 , P c2 , P m1 , P m2 represents the preset parameter, f' represents the larger fitness value of the two individuals involved in the crossover, f represents the fitness value of the variant individual, and f avg represents the average fitness value of the population, f max Indicates the maximum fitness value of the population;

[0070] The optimized genetic algorithm dynamically adjusts the probability of crossover and mutation according to the diversity of the population and the fitness of the individuals.

[0071] Preferably, the method further comprises:

[0072] The dynamic inertia weight is introduced to optimize the particle swarm algorithm:

[0073]

[0074] Among them, ω max and ω min Indicates the upper and lower limits of the inertia weight, t indicates the current number of iterations, t max Indicates the maximum number of iterations.

[0075] Preferably, the method further comprises:

[0076] In the process of solving the objective function, for solutions that violate the constraint conditions, their fitness is reduced through a penalty function.

[0077] Preferably, the method further comprises:

[0078] Build a neural network model;

[0079] The wind speed, wind direction, air pressure, temperature and wind turbine operating status parameters of the wind farm are obtained as nodes of the input layer of the neural network model;

[0080] At least one hidden layer is set, and the number of nodes in the hidden layer is determined; the activation function of the hidden layer is a ReLU function; the variable in the ReLU function is wind speed, wind direction, air pressure or temperature;

[0081] An output layer is set; the number of nodes in the output layer is 1, and the output is the predicted wind power; the activation function of the output layer is selected according to the range of the prediction target, and if the predicted power is a positive value, a linear activation function is selected;

[0082] Divide the processed historical wind power generation data into a training set, a validation set, and a test set;

[0083] Take the training set as the input vector and calculate the predicted output through the forward propagation of the neural network model;

[0084] The mean square error is used as the loss function, and the gradient of the loss function to the weights and biases of each layer is calculated through the back propagation algorithm;

[0085] The gradient from the output layer to the hidden layer weights is calculated according to the chain rule.

[0086] Update weights and biases using gradient descent algorithm;

[0087] Execute the loop until the loss function converges or the preset number of training times is reached;

[0088] The trained neural network model is used as the first prediction model.

[0089] Preferably, the method further comprises:

[0090] Obtain historical light intensity data, solar radiation angle data, cloud coverage data, ambient temperature data and characteristic parameters of photovoltaic modules as historical photovoltaic power generation data;

[0091] Build a support vector machine model;

[0092] The kernel function of the support vector machine model will be determined;

[0093] Construct optimization objectives and constraints for support vector machine models;

[0094] Training the support vector machine model based on the historical photovoltaic power generation data;

[0095] The trained support vector machine model is used as the second prediction model.

[0096] According to a second aspect of an embodiment of the present application, there is provided a maintenance plan optimization device for a new energy consumption power system, comprising:

[0097] Processor and memory;

[0098] The processor and the memory are connected via a communication bus:

[0099] Wherein, the processor is used to call and execute the program stored in the memory;

[0100] The memory is used to store a program, and the program is used at least to execute a maintenance plan optimization method for a new energy consumption power system as described in any of the above items.

[0101] The technical solution provided by this application may have the following beneficial effects:

[0102] The maintenance plan optimization method of the new energy consumption power system in the present application includes: obtaining real-time data of new energy power generation; the real-time data of new energy power generation includes: real-time data of wind power generation and real-time data of photovoltaic power generation; inputting the real-time data of wind power generation into a pre-trained first prediction model to obtain wind power generation prediction results; inputting the real-time data of photovoltaic power generation into a pre-trained second prediction model to obtain photovoltaic power generation prediction results; inputting the wind power generation prediction results and the photovoltaic power generation prediction results into a pre-trained maintenance plan optimization model, and outputting an optimized maintenance plan; wherein, the objective function of the maintenance plan optimization model includes: maximizing new energy consumption and minimizing maintenance costs; the constraints of the maintenance plan optimization model include: new energy power generation constraints, load balance constraints, equipment maintenance time window constraints, equipment association constraints and energy storage equipment constraints; the maintenance plan optimization model solves the objective function using an optimized genetic algorithm or an optimized particle swarm algorithm.

[0103] In this technical solution, new energy power generation is predicted through different prediction models, and the optimized maintenance plan is output based on the prediction results and the objective functions of maximizing new energy consumption and minimizing maintenance costs, and the optimized maintenance plan meets the preset constraints. At the same time, combined with the optimization algorithm solution, the maintenance plan can be adjusted in real time according to the prediction results of new energy power generation, improving the scientificity and feasibility of the plan, reducing the potential risks when implementing the maintenance plan, avoiding the problem of grid stability out of control during maintenance and the inability to effectively consume new energy power generation, and overcoming the shortcomings of traditional maintenance plan adjustment based on expert experience.

[0104] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0105] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0106] Figure 1 It is a flowchart of a maintenance plan optimization method for a new energy consumption power system provided by an embodiment of the present application;

[0107] Figure 2 It is a structural schematic diagram of a maintenance plan optimization device for a new energy consumption power system provided by an embodiment of the present application.

[0108] Reference numerals: processor-21; memory-22. DETAILED DESCRIPTION

[0109] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0110] Embodiment 1

[0111] Figure 1 This is a flow chart of a method for optimizing a maintenance plan for a new energy consumption power system provided by an embodiment of the present application, with reference to Figure 1 , a maintenance plan optimization method for a new energy consumption power system, comprising:

[0112] S11: Acquire real-time data of renewable energy power generation; the real-time data of renewable energy power generation includes: real-time data of wind power generation and real-time data of photovoltaic power generation;

[0113] S12: inputting the real-time data of wind power generation into a pre-trained first prediction model to obtain a wind power generation prediction result;

[0114] S13: inputting the real-time data of photovoltaic power generation into a pre-trained second prediction model to obtain a photovoltaic power generation prediction result;

[0115] S14: inputting the wind power generation prediction results and the photovoltaic power generation prediction results into a pre-trained maintenance plan optimization model, and outputting an optimized maintenance plan;

[0116] Among them, the objective functions of the maintenance plan optimization model include: maximizing new energy consumption and minimizing maintenance costs;

[0117] The constraints of the maintenance plan optimization model include: renewable energy generation constraints, load balance constraints, equipment maintenance time window constraints, equipment association constraints, and energy storage equipment constraints;

[0118] The maintenance plan optimization model solves the objective function using the optimized genetic algorithm or the optimized particle swarm algorithm.

[0119] It should be noted that the method also includes:

[0120] The total power generation of the wind farm at time t is calculated as:

[0121]

[0122] Among them, P ω (t) represents the total power generation of the wind farm at time t, n is the total number of wind turbines in the wind farm, v i (t) represents the wind speed of the i-th wind turbine at time t, P ω,i (v) represents the power characteristic curve of the i-th wind turbine at time t;

[0123] Calculate the power consumption of the wind farm in the time interval [T1, T2]:

[0124]

[0125] Among them, E ω It represents the amount of electricity consumed by the wind farm in the time interval [T1, T2], and d is the set parameter;

[0126] The total power generation of the photovoltaic array at time t is calculated as:

[0127] P p (t) = ηI(t)A;

[0128] Among them, P P(t) represents the total power generation of the photovoltaic array at time t, I(t) represents the light intensity received by the photovoltaic array at time t, η represents the conversion efficiency of the photovoltaic array, and A represents the area of ​​the photovoltaic array;

[0129] Calculate the power consumption of the photovoltaic array in the time interval [T1, T2]:

[0130]

[0131] Among them, E P It represents the amount of electricity consumed by the photovoltaic array in the time interval [T1, T2], and d is the set parameter;

[0132] The total amount of new energy consumption is obtained based on the total amount of wind farm consumption and the total amount of photovoltaic array consumption:

[0133] E ren =E ω +E P ;

[0134] Then the objective function of maximizing the consumption of new energy is expressed as maximizeE ren .

[0135] It should be noted that the method also includes:

[0136] Calculate the cost of overhauling a single piece of equipment:

[0137] C j =c m,j +c l,j +c u,j *t out,j ;

[0138] Among them, C j represents the maintenance cost of equipment j, C m,j represents the maintenance material cost of equipment j, C l,j represents the labor cost of equipment j, C u,j represents the power outage loss cost per unit time of device j, t out,j Indicates the power outage time;

[0139] Calculate the maintenance cost of all equipment:

[0140]

[0141] Where m is the total number of devices;

[0142] Then the objective function of minimizing the maintenance cost is expressed as minimizeC;

[0143] The objective function of the maintenance plan optimization model is expressed as F = ω1E ren -ω2C;

[0144] Among them, F represents the objective function value, ω1 and ω2 represent the weight coefficients of maximizing new energy consumption and minimizing maintenance costs, respectively.

[0145] The maintenance plan optimization model in this embodiment has the main goals of minimizing maintenance costs and maximizing new energy consumption.

[0146] Objective function 1: Maximizing the consumption of new energy. In order to accurately quantify the amount of new energy consumption, calculation models are established for different types of new energy power generation (wind power generation and photovoltaic power generation).

[0147] Objective function 2: Minimize maintenance costs. Maintenance costs include many aspects, such as the material cost required for equipment maintenance, the labor cost of maintenance personnel, and the economic loss cost caused to users by power outages due to maintenance.

[0148] Taking the above two aspects into consideration, the comprehensive objective function can be expressed as: F = ω1E ren -ω2C min imizeC. The weight coefficient is used to balance the importance of new energy consumption and maintenance cost in the objective function.

[0149] It should be noted that the method also includes:

[0150] According to the technical parameters of the wind turbine and the local wind speed historical statistical data, the upper and lower limits of wind power generation are set as the constraints of wind power generation:

[0151] P ω,min ≤P ω (t)≤P ω,max ;

[0152] Among them, P ω,min Represents the lower limit of wind power generation in the wind farm, P ω,max Indicates the upper limit of wind power generation capacity of the wind farm;

[0153] According to the characteristics of photovoltaic modules and the range of light conditions, the upper and lower limits of photovoltaic power generation are set as the constraints of photovoltaic power generation:

[0154] P p,min ≤P p (t)≤P p,max ;

[0155] Among them, P P,min Indicates the lower limit of photovoltaic power generation of photovoltaic modules, P P,max Indicates the upper limit of photovoltaic power generation of photovoltaic modules;

[0156] Set the total power generation of wind power and photovoltaic power generation to no less than the minimum expected value P ren,minAs a constraint on renewable energy generation:

[0157] P ω +P P ≥P ren,min ;

[0158] Due to the uncertainty of renewable energy power generation, there are upper and lower limits to its power generation capacity, so constraints on renewable energy power generation are formulated.

[0159] During the maintenance period, the power grid must meet the load balance requirement, that is, the total power generation (including new energy power generation and traditional energy power generation) is equal to the load demand, so the load balance constraint is determined based on the total power generation:

[0160] P ω (t)+P p (t)+P trad (t) = L(t);

[0161] Among them, P trad (t) represents the power generated by traditional energy, and L(t) represents the total power generated;

[0162] Preferably, taking into account the dynamic changing characteristics of load demand, such as the difference between peak and valley in different time periods, it is necessary to establish a dynamic load balance constraint model to ensure that the power grid can operate stably at any time and avoid problems such as voltage fluctuations and frequency anomalies caused by power imbalance.

[0163] Each device has its allowed maintenance time range. The device maintenance time window constraint is determined based on the earliest allowed maintenance time and the latest allowed maintenance time of the device:

[0164] t start,j ≤t j ≤t end,j ;

[0165] Among them, t start,j represents the earliest allowable maintenance time of equipment j, t end,j represents the latest allowable maintenance time of equipment j, t j represents the maintenance time of equipment j;

[0166] Equipment association constraints include: maintenance order constraints, simultaneous maintenance constraints, and non-simultaneous maintenance constraints;

[0167] In order to improve maintenance efficiency and reduce the impact on power grid operation, for some highly correlated devices, such as series devices or devices with great mutual influence, it is also necessary to consider the maintenance sequence constraints between them. The maintenance sequence constraint is expressed as

[0168] Among them, j1 and j2 are devices connected in series. Indicates the latest permissible maintenance time of equipment j1. Indicates the earliest allowed maintenance time of equipment j2;

[0169] In addition to the maintenance time sequence constraint, there are also some logical association constraints between devices. For example, some devices must be maintained at the same time, and the simultaneous maintenance constraint is expressed as t start,j3 =t start,j4 and t end,j3 =t end,j4 ;

[0170] Among them, j3 and j4 represent the equipment that must be repaired at the same time. Indicates the earliest allowed maintenance time of equipment j2. Indicates the earliest permissible maintenance time of equipment j4. Indicates the latest allowed maintenance time of equipment j3. Indicates the latest allowed maintenance time of equipment j4;

[0171] For some mutually exclusive maintenance equipment, the constraint that they cannot be repaired at the same time is expressed as t end,j5 <t end,j6 or end,j6 <t end,j5 ;

[0172] Among them, j5 and j6 represent equipment that cannot be repaired at the same time. Indicates the latest allowed maintenance time of equipment j5. Indicates the latest permissible maintenance time of equipment j6;

[0173] Consideration of these equipment-related constraints helps ensure the rationality and feasibility of the maintenance plan and avoid power grid operation failures caused by improper equipment maintenance arrangements.

[0174] If energy storage equipment is installed in the power grid, the charging and discharging characteristics and capacity limitations of the energy storage equipment need to be considered. Energy storage equipment constraints include:

[0175] 0≤P ch (t)≤P ch,max ;

[0176] 0≤P dis (t)≤P dis,max ;

[0177] S min ≤S(t)≤S max ;

[0178] Among them, P ch (t) represents the charging power of the energy storage device at time t, P ch,max Indicates the maximum charging power of the energy storage device, P dis(t) represents the discharge power of the energy storage device at time t, P dis,max represents the maximum discharge power of the energy storage device, S(t) represents the current energy storage capacity of the energy storage device, S min Represents the minimum energy storage capacity of the energy storage device, S max Indicates the maximum energy storage capacity of the energy storage device.

[0179] In the process of optimizing the maintenance plan, the energy storage and release functions of the energy storage equipment are rationally utilized. For example, excess electricity is stored when there is excess power generation from renewable energy, and electricity is released when there is insufficient power generation from renewable energy or maintenance causes a load gap, to help maintain the power balance and stable operation of the power grid.

[0180] In this technical solution, the maintenance plan optimization model uses an optimized genetic algorithm or an optimized particle swarm algorithm to solve the objective function. Intelligent optimization algorithms such as genetic algorithms and particle swarm optimization algorithms have strong global search capabilities, can find better solutions in complex solution spaces, and have relatively high computational efficiency, but may have problems such as slow convergence or falling into local optimal solutions, so the genetic algorithm or particle swarm algorithm needs to be optimized in this technical solution.

[0181] Specifically, the optimization of the genetic algorithm includes:

[0182] Adaptive crossover probability P c and mutation probability P m , the genetic algorithm is optimized by optimizing the chromosome encoding:

[0183]

[0184]

[0185] Where P c1 , P c2 , P m1 , P m2 represents the preset parameter, f' represents the larger fitness value of the two individuals involved in the crossover, f represents the fitness value of the variant individual, and f avg represents the average fitness value of the population, f max Indicates the maximum fitness value of the population;

[0186] The optimized genetic algorithm dynamically adjusts the probability of crossover and mutation according to the diversity of the population and the fitness of the individuals.

[0187] For genetic algorithms, the chromosome encoding method is optimized to make it more suitable for representing the equipment maintenance status and schedule in the maintenance plan; genetic operations (selection, crossover, mutation) are improved to improve the algorithm's search efficiency and convergence speed.

[0188] The optimization of particle swarm algorithm includes:

[0189] Introducing dynamic inertia weight to optimize the particle swarm algorithm:

[0190]

[0191] Among them, ω max and ω min Indicates the upper and lower limits of the inertia weight, t indicates the current number of iterations, t max Indicates the maximum number of iterations.

[0192] For the particle swarm optimization algorithm, a dynamic inertia weight is introduced to adjust the flight speed of the particles according to the iterative process of the algorithm, thereby enhancing the balance between the global search capability and the local search capability of the algorithm.

[0193] It should be noted that the method also includes:

[0194] In the process of solving the objective function, the fitness of the solution that violates the constraints is reduced through the penalty function.

[0195] In this technical solution, an effective penalty function is designed in combination with the actual constraints of the problem. For solutions that violate the constraints, the penalty function is used to reduce their fitness, so that they are gradually eliminated during the algorithm iteration process, guiding the algorithm to search for feasible solution space.

[0196] For example, for solutions that violate load balancing constraints or equipment maintenance time window constraints, a larger penalty value is given:

[0197]

[0198] Among them, k1 and k2 are penalty coefficients, which ensure that the final maintenance plan meets all constraints.

[0199] It should be noted that in this technical solution, a high-precision renewable energy power generation prediction model is established, which comprehensively utilizes multi-source data such as historical data, meteorological data (such as wind speed, light intensity, temperature, etc.), geographic information, etc., and adopts machine learning algorithms (such as neural networks, support vector machines, etc.) or physical model methods to improve the accuracy of renewable energy power generation prediction.

[0200] Specifically, training the first prediction model includes:

[0201] Build a neural network model;

[0202] The wind speed, wind direction, air pressure, temperature and wind turbine operating status parameters of the wind farm are obtained as nodes of the input layer of the neural network model;

[0203] At least one hidden layer is set, and the number of nodes in the hidden layer is determined; the activation function of the hidden layer is the ReLU function; the variable in the ReLU function is wind speed, wind direction, air pressure or temperature;

[0204] Set the output layer; the number of nodes in the output layer is 1, and the output is the predicted wind power; the activation function of the output layer is selected according to the range of the prediction target. If the predicted power is a positive value, the linear activation function is selected;

[0205] Divide the processed historical wind power generation data into a training set, a validation set, and a test set;

[0206] Take the training set as the input vector and calculate the predicted output through the forward propagation of the neural network model;

[0207] The mean square error is used as the loss function, and the gradient of the loss function to the weights and biases of each layer is calculated through the back propagation algorithm;

[0208] The gradient from the output layer to the hidden layer weights is calculated according to the chain rule.

[0209] Update weights and biases using gradient descent algorithm;

[0210] Execute the loop until the loss function converges or the preset number of training times is reached;

[0211] The trained neural network model is used as the first prediction model.

[0212] In specific practice, a neural network model is constructed to predict wind power generation.

[0213] 1) Determine the number of input layer nodes based on the collected data. The input nodes can include the normalized wind speed υ norm (t), wind direction θ(t), air pressure ρ(t), air temperature T(t) and operating state parameters of the wind turbine (such as pitch angle β(t), etc.). Assuming there are n input variables in total, the number of input layer nodes is n.

[0214] 2) Set at least one hidden layer.

[0215] The number of hidden layer nodes h can be determined based on an empirical formula or through experiments. The empirical formula is:

[0216]

[0217] Among them, a is a constant between 1 and 10, and m can be temporarily set to 1 and can be adjusted according to actual conditions).

[0218] The activation function of the hidden layer can select the commonly used ReLU function to introduce nonlinear characteristics.

[0219] f(x)=max(0,x);

[0220] x can be wind speed, temperature, etc.

[0221] 3) Set up the output layer.

[0222] The number of nodes in the output layer is 1, and the output is the predicted wind power generation P ω (t). The activation function of the output layer is selected according to the range of the prediction target. If the predicted power is positive, a linear activation function can be selected.

[0223] f(x) = x;

[0224] 4) Data partitioning.

[0225] The preprocessed historical data is divided into training set, validation set and test set. Usually, the ratio can be 70%, 15%, and 15%. Suppose the training data set is {(x1, y1), (x2, y2), …, (x N ,y N )}, where x i is the input vector (including the above-mentioned various input data), y i is the corresponding actual wind power value.

[0226] 5) Forward propagation.

[0227] For the input vector x i , the predicted output is calculated through the forward propagation of the neural network.

[0228] First, the output of the input layer is equal to the input vector x i ;

[0229] Then, through the calculation of the hidden layer, the output of the jth node in the hidden layer is

[0230]

[0231] in, is the weight from the input layer to the hidden layer, is the bias of the hidden layer), and finally the output of the output layer

[0232]

[0233] in, is the weight from the hidden layer to the output layer, b 2 is the bias of the output layer.

[0234] 6) Loss function calculation.

[0235] The mean square error (MSE) is used as the loss function

[0236]

[0237] 7) Back propagation and parameter update.

[0238] The gradient of the loss function to the weights and biases of each layer is calculated by the back propagation algorithm. Gradient According to the chain rule calculation, the loss function is used to calculate the output layer output. The derivative of the output layer activation function, the derivative of the hidden layer output hj, etc. Then the gradient descent algorithm is used to update the weights and biases. The weight update formula is

[0239]

[0240] Where a is the learning rate. The bias update formula is similar.

[0241] Repeat the above process until the loss function converges or the preset number of training times is reached.

[0242] Training the second prediction model includes:

[0243] Obtain historical light intensity data, solar radiation angle data, cloud coverage data, ambient temperature data and characteristic parameters of photovoltaic modules as historical photovoltaic power generation data;

[0244] Build a support vector machine model;

[0245] The kernel function of the support vector machine model will be determined;

[0246] Construct the optimization objectives and constraints of the support vector machine model;

[0247] Training a support vector machine model based on historical photovoltaic power generation data;

[0248] The trained support vector machine model is used as the second prediction model.

[0249] In specific practice, the use of support vector machines for photovoltaic power generation prediction includes:

[0250] 1) Data collection. Collect historical light intensity data I(t) (t represents time), and record the light intensity values ​​at different time points. The time resolution can be selected according to the actual situation, such as recording once every 15 minutes. Collect solar radiation angle data α(t), which affects the amount of solar radiation received by the photovoltaic module, and is also recorded at the corresponding time resolution. Cloud coverage data C(t), clouds will block sunlight and affect photovoltaic power generation. The corresponding data can be obtained through meteorological observations or satellite cloud images. Ambient temperature data T(t), temperature affects the conversion efficiency of photovoltaic modules, and the temperature values ​​are recorded in chronological order. Characteristic parameters of photovoltaic modules, such as the conversion efficiency η of the module, the open circuit voltage V oc, short circuit current I sc As well as the total area A of the component, these parameters are usually obtained through experimental tests.

[0251] 2) Support vector machine model construction

[0252] Kernel function selection. The support vector machine maps the input data to a high-dimensional feature space through a kernel function in order to find an optimal classification hyperplane in the high-dimensional space. Commonly used kernel functions include radial basis function (RBF)

[0253] K(x i ,x j ) = exp(-γ||x i -x j || 2 );

[0254] γ is the kernel parameter.

[0255] Polynomial kernel function:

[0256] K(x i ,x j )=(x i T x j +c) d ;

[0257] c is a constant and d is the degree of the polynomial.

[0258] For photovoltaic power generation prediction, the RBF kernel function usually performs better due to the nonlinear characteristics of its data, but other kernel functions can also be tried according to the actual data situation.

[0259] 3) Construct optimization goals

[0260] For photovoltaic power generation prediction, the optimization objective of the support vector machine is

[0261]

[0262] The constraints are:

[0263]

[0264] Where w is the weight vector, b is the bias, ξ i is a slack variable, and ε is an insensitive parameter in the loss function, which indicates the error tolerance range between the predicted value and the true value. The purpose of this optimization problem is to minimize the complexity of the model (measured by) while ensuring that the prediction error is within a certain range (controlled by), and to balance the training error and model complexity through the penalty parameter C.

[0265] In this technical solution, different prediction models are used to predict renewable energy power generation. Based on the prediction results and the objective functions of maximizing renewable energy consumption and minimizing maintenance costs, an optimized maintenance plan is output, and the optimized maintenance plan meets the preset constraints. At the same time, combined with the optimization algorithm solution, the maintenance plan can be adjusted in real time according to the prediction results of renewable energy power generation.

[0266] For example, if it is predicted that the power of renewable energy generation will increase significantly during a certain period of time, exceeding the power grid's absorption capacity, some highly flexible equipment can be arranged for maintenance in advance to increase the grid's absorption space; if there is a sudden failure or a sudden drop in the output of renewable energy generation, the maintenance plan can be adjusted in time to give priority to the safe and stable operation of the power grid, and the maintenance of some equipment can be suspended or postponed. Through real-time data feedback and dynamic adjustment mechanisms, the maintenance plan can better adapt to the uncertainty of renewable energy generation and improve the operation efficiency of the power grid and the level of renewable energy absorption.

[0267] Embodiment 2

[0268] Figure 2 This is a schematic diagram of a maintenance plan optimization device for a new energy consumption power system provided by an embodiment of the present application, referring to Figure 2 , a maintenance plan optimization device for a new energy consumption power system, comprising:

[0269] Processor 21 and memory 22;

[0270] The processor 21 and the memory 22 are connected via a communication bus:

[0271] The processor 21 is used to call and execute the program stored in the memory 22;

[0272] The memory 22 is used to store a program, and the program is used to execute at least one of the maintenance plan optimization methods for a new energy consumption power system in the above embodiment.

[0273] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0274] It should be noted that, in the description of this application, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of this application, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0275] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0276] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0277] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0278] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0279] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0280] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0281] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A maintenance plan optimization method for a new energy consumption power system, characterized in that: include: Obtain real-time data on renewable energy power generation; The real-time data of renewable energy power generation includes: real-time data of wind power generation and real-time data of photovoltaic power generation; Inputting the real-time wind power generation data into a pre-trained first prediction model to obtain a wind power generation prediction result; Inputting the photovoltaic power generation real-time data into a pre-trained second prediction model to obtain a photovoltaic power generation prediction result; Inputting the wind power generation prediction result and the photovoltaic power generation prediction result into a pre-trained maintenance plan optimization model, and outputting an optimized maintenance plan; The objective function of the maintenance plan optimization model includes: maximizing new energy consumption and minimizing maintenance costs; The constraints of the maintenance plan optimization model include: renewable energy generation constraints, load balance constraints, equipment maintenance time window constraints, equipment association constraints and energy storage equipment constraints; The maintenance plan optimization model solves the objective function using an optimized genetic algorithm or an optimized particle swarm algorithm.

2. The method according to claim 1, characterized in that The method further comprises: The total power generation of the wind farm at time t is calculated as: Among them, P ω (t) represents the total power generation of the wind farm at time t, n is the total number of wind turbines in the wind farm, v i (t) represents the wind speed of the i-th wind turbine at time t, P ω,i (v) represents the power characteristic curve of the i-th wind turbine at time t; Calculate the power consumption of the wind farm in the time interval [T1, T2]: Among them, E ω It represents the amount of electricity consumed by the wind farm in the time interval [T1, T2], and d is the set parameter; The total power generation of the photovoltaic array at time t is calculated as: P p (t)=ηI(t)A; Among them, P P (t) represents the total power generation of the photovoltaic array at time t, I(t) represents the light intensity received by the photovoltaic array at time t, η represents the conversion efficiency of the photovoltaic array, and A represents the area of ​​the photovoltaic array; Calculate the power consumption of the photovoltaic array in the time interval [T1, T2]: Among them, E P It represents the amount of electricity consumed by the photovoltaic array in the time interval [T1, T2], and d is the set parameter; The total amount of new energy consumption is obtained based on the total amount of wind farm consumption and the total amount of photovoltaic array consumption: AND ren =And ω +E P ; Then the objective function of maximizing the consumption of new energy is expressed as maximizeE ren .

3. The method according to claim 2, characterized in that The method further comprises: Calculate the cost of overhauling a single piece of equipment: C j =c m,j +c l,j +c u,j *t out,j ; Among them, C j represents the maintenance cost of equipment j, C m,j represents the maintenance material cost of equipment j, C l,j represents the labor cost of equipment j, C u,j represents the unit time power outage loss cost of device j, t out,j Indicates the power outage time; Calculate the maintenance cost of all equipment: Where m is the total number of devices; Then the objective function of minimizing the maintenance cost is expressed as minimizeC; The objective function of the maintenance plan optimization model is expressed as F = ω1E ren -ω2C; Among them, F represents the objective function value, ω1 and ω2 represent the weight coefficients of maximizing new energy consumption and minimizing maintenance costs, respectively.

4. The method according to claim 1, characterized in that: The method further comprises: According to the technical parameters of the wind turbine and the local wind speed historical statistical data, the upper and lower limits of wind power generation are set as the constraints of wind power generation: P ω,min ≤P ω (t)≤P ω,max ; Among them, P ω,min Represents the lower limit of wind power generation in the wind farm, P ω,max Indicates the upper limit of wind power generation capacity of the wind farm; According to the characteristics of photovoltaic modules and the range of light conditions, the upper and lower limits of photovoltaic power generation are set as the constraints of photovoltaic power generation: P p,min ≤P p (t)≤P p,max ; Among them, P P,min Indicates the lower limit of photovoltaic power generation of photovoltaic modules, P P,max Indicates the upper limit of photovoltaic power generation of photovoltaic modules; Set the total power generation of wind power and photovoltaic power generation to no less than the minimum expected value P ren,min As a constraint on renewable energy generation: P ω +P P ≥P ren,min ; Determine the load balancing constraint based on the total generated power: P ω (t)+P p (t)+P trad (t)=L(t); Among them, P trad (t) represents the power generated by traditional energy, and L(t) represents the total power generated; Determine the equipment maintenance time window constraint based on the earliest and latest allowable maintenance times of the equipment: t start,j ≤t j ≤t end,j ; Among them, t start,j represents the earliest allowable maintenance time of equipment j, t en d ,j represents the latest allowable maintenance time of equipment j, t j represents the maintenance time of equipment j; The equipment association constraints include: maintenance order constraints, simultaneous maintenance constraints and non-simultaneous maintenance constraints; The maintenance sequence constraint is expressed as Among them, j1 and j2 are devices connected in series. Indicates the latest permissible maintenance time of equipment j1. Indicates the earliest allowed maintenance time of equipment j2; The simultaneous maintenance constraint is expressed as t start,j3 =t start,j4 and t end,j3 =t end,j4 ; Among them, j3 and j4 represent the equipment that must be repaired at the same time. Indicates the earliest allowed maintenance time of equipment j2. Indicates the earliest permissible maintenance time of equipment j4. Indicates the latest allowed maintenance time of equipment j3. Indicates the latest allowed maintenance time of equipment j4; The constraint that no simultaneous maintenance can be performed is expressed as t end,j5 <t end,j6 or end,j6 <t end,j5 ; Among them, j5 and j6 represent equipment that cannot be repaired at the same time. Indicates the latest allowed maintenance time of equipment j5. Indicates the latest allowed maintenance time of equipment j6; Energy storage device constraints include: 0≤P ch (t)≤P ch,max ; 0≤P dis (t)≤P dis,max ; S min ≤S(t)≤S max ; Among them, P ch (t) represents the charging power of the energy storage device at time t, P ch,max Indicates the maximum charging power of the energy storage device, P dis (t) represents the discharge power of the energy storage device at time t, P dis,max represents the maximum discharge power of the energy storage device, S(t) represents the current energy storage capacity of the energy storage device, S min Represents the minimum energy storage capacity of the energy storage device, S max Indicates the maximum energy storage capacity of the energy storage device.

5. The method according to claim 1, characterized in that The method further comprises: Adaptive crossover probability P c and mutation probability P m , the genetic algorithm is optimized by optimizing the chromosome encoding: Where P c1 , P c2 , P m1 , P m2 represents the preset parameter, f' represents the larger fitness value of the two individuals involved in the crossover, f represents the fitness value of the variant individual, and f avg represents the average fitness value of the population, f max Indicates the maximum fitness value of the population; The optimized genetic algorithm dynamically adjusts the probability of crossover and mutation according to the diversity of the population and the fitness of the individuals.

6. The method according to claim 1, characterized in that The method further comprises: The dynamic inertia weight is introduced to optimize the particle swarm algorithm: Among them, ω max and ω min Indicates the upper and lower limits of the inertia weight, t indicates the current number of iterations, t max Indicates the maximum number of iterations.

7. The method according to claim 1, characterized in that The method further comprises: In the process of solving the objective function, for solutions that violate the constraint conditions, their fitness is reduced through a penalty function.

8. The method according to claim 1, characterized in that The method further comprises: Build a neural network model; The wind speed, wind direction, air pressure, temperature and wind turbine operating status parameters of the wind farm are obtained as nodes of the input layer of the neural network model; At least one hidden layer is set, and the number of nodes in the hidden layer is determined; the activation function of the hidden layer is a ReLU function; the variable in the ReLU function is wind speed, wind direction, air pressure or temperature; An output layer is set; the number of nodes in the output layer is 1, and the output is the predicted wind power; the activation function of the output layer is selected according to the range of the prediction target, and if the predicted power is a positive value, a linear activation function is selected; Divide the processed historical wind power generation data into a training set, a validation set, and a test set; Take the training set as the input vector and calculate the predicted output through the forward propagation of the neural network model; The mean square error is used as the loss function, and the gradient of the loss function to the weights and biases of each layer is calculated through the back propagation algorithm; The gradient from the output layer to the hidden layer weights is calculated according to the chain rule. Update weights and biases using gradient descent algorithm; Execute the loop until the loss function converges or the preset number of training times is reached; The trained neural network model is used as the first prediction model.

9. The method according to claim 1, characterized in that: The method further comprises: Obtain historical light intensity data, solar radiation angle data, cloud coverage data, ambient temperature data and characteristic parameters of photovoltaic modules as historical photovoltaic power generation data; Build a support vector machine model; The kernel function of the support vector machine model will be determined; Construct the optimization objectives and constraints of the support vector machine model; Training the support vector machine model based on the historical photovoltaic power generation data; The trained support vector machine model is used as the second prediction model.

10. A maintenance plan optimization device for a new energy consumption power system, characterized in that: include: Processor and memory; The processor and the memory are connected via a communication bus: Wherein, the processor is used to call and execute the program stored in the memory; The memory is used to store a program, and the program is used at least to execute a maintenance plan optimization method for a new energy consumption power system as described in any one of claims 1-9.

Images