Typical pipeline station integrated energy system scheduling method

Abstract

The invention provides a typical pipeline station integrated energy system scheduling method. The method comprises the steps of obtaining photovoltaic output power and corresponding meteorological data; respectively predicting the wind power output power and the photovoltaic output power at the t moment by adopting an ARIMA model to obtain a first power predicted value, and respectively predicting the wind power output power and the photovoltaic output power at the t moment by adopting an ARIMA-RNN combined model to obtain a second power predicted value; carrying out optimization scheduling on the integrated energy system comprising a plurality of devices, constructing a target function, and determining constraint conditions of the devices in the integrated energy system according to the second power predicted value; and based on the constraint condition, optimizing the objective function by adopting an improved particle swarm algorithm combined with an adaptive inertia weight, and solving to obtain an optimal strategy for optimal scheduling of the integrated energy system. The method has the technical effects that the design is reasonable, and the optimal scheduling of the integrated energy system is realized on the premise of considering the operation cost and the environment cost.

Description

Technical Field

[0001] The present invention belongs to the technical field of integrated energy system scheduling, and particularly relates to a scheduling method for a typical pipeline station integrated energy system. Background Art

[0002] The integrated energy system takes the power system as the core, couples multiple energies such as cold, heat, and natural gas, and realizes the organic coordination and optimal operation of energy through various energy conversion devices and energy storage devices. The integrated energy system is an important way to improve energy utilization efficiency, and is also of great significance in improving the reliability of system energy supply, reducing the energy consumption cost of users, and reducing carbon emissions.

[0003] At present, most of the research on the optimal scheduling of integrated energy systems only considers the role of demand response in reducing the system operation cost, or considers the coordinated response of integrated demand response and a single energy storage system. The environmental protection and economy of the integrated energy system need to be balanced. However, there is still a gap in the optimal scheduling of integrated energy systems considering environmental costs, and balancing the environmental costs and economic costs of the system plays an important role in the current development situation of the goal. Summary of the Invention

[0004] The present invention aims to at least solve one of the technical problems existing in the prior art, and provides a new technical solution for a scheduling method for a typical pipeline station integrated energy system.

[0005] According to one aspect of the present invention, there is provided a scheduling method for a typical pipeline station integrated energy system, including the following steps:

[0006] Step S100: Obtain the photovoltaic output power and corresponding meteorological data through a data acquisition unit, perform a correlation analysis on each meteorological feature of the meteorological data by using the Pearson correlation coefficient, screen out the three meteorological features with the highest correlation with the photovoltaic output power, and eliminate redundant meteorological data;

[0007] Step S200: Use the ARIMA model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the first power prediction value, input the first power prediction value and the meteorological data into the RNN model for training to obtain the ARIMA-RNN combined model, and use the ARIMA-RNN combined model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the second power prediction value;

[0008] Step S300: Optimally schedule an integrated energy system including multiple devices, construct an objective function, and determine the constraint conditions of each device in the integrated energy system according to the second power prediction value;

[0009] Step S400: Based on the above constraints, use an improved particle swarm optimization algorithm combined with an adaptive inertia weight to optimize the objective function, and solve to obtain the optimal strategy for the integrated energy system's optimal scheduling.

[0010] Optionally, use the Pearson correlation coefficient to perform a correlation analysis on each meteorological feature of the meteorological data, including:

[0011] Take the photovoltaic output power as the target vector, and take the meteorological features of the meteorological data affecting photovoltaic power generation as the feature vector. Use the Pearson correlation coefficient to perform a correlation analysis on the feature vector. The calculation formula is as follows:

[0012]

[0013] In the above formula, r is the Pearson correlation coefficient, N is the number of samples, X i is the sample value of the target vector, is the mean of the target vector, Y i is the sample value of the feature vector, is the mean of the feature vector.

[0014] Optionally, before using the ARIMA model to predict the wind power output and photovoltaic power output at time t respectively to obtain the first power prediction value, it further includes:

[0015] First, use the following formula to normalize the data;

[0016]

[0017] In the above formula, X is the original time series of the data, x max is the maximum value of the data, is the average value of the data, is the time series after normalization; where the data includes power data and meteorological data:

[0018] Then, construct an ARIMA(p,d,q) model, including:

[0019] Use the unit root test method to perform a stationarity test on the power data. If the power data is a non-stationary time series, then perform differencing on the power data. After n-order differencing, the power data becomes a stationary series, and at this time d = n; if the power data is a stationary series, then d = 0; calculate the autocorrelation and partial autocorrelation functions, determine the value ranges of the autoregressive term p and the moving average term q, and use the Akaike information criterion to order p and q to obtain the ARIMA model.

[0020] Optionally, use the ARIMA model to predict the wind power output and photovoltaic power output at time t respectively to obtain the first power prediction value, including:

[0021] Train the ARIMA model using historical data, where the historical data includes historical data of wind power output and historical data of photovoltaic power output;

[0022] Use the trained ARIMA model to predict the wind power output and photovoltaic power output at time t respectively, and obtain the first wind power prediction value at time t and the first photovoltaic power prediction value

[0023] Optionally, input the first power prediction value and meteorological data into the RNN model for training, including:

[0024] For the photovoltaic power output, use the data of the three meteorological characteristics and the first photovoltaic power prediction value as input features and input them into the RNN model for training;

[0025] For the wind power output, use the wind speed as the data of the meteorological characteristic and the first wind power prediction value as input features and input them into the RNN model for training;

[0026] Among them, the RNN model is constructed using the following formula:

[0027]

[0028] In the above formula, is the predicted value of wind power output or photovoltaic power output, g(·) is the softmax function, V is the parameter matrix of the hidden layer and output layer of the recurrent neural network, c is the bias term, and h t is the state information of the hidden layer, and h is defined using the following formula t :

[0029] h t = f(Wh t-1 + Ux t + b)

[0030] In the above formula, f(·) is a non-linear activation function, that is, the sigmoid activation function, W and U are the parameter matrices of the previous hidden layer and the current input respectively, h t-1 is the state information of the previous moment, x t is the input of the current moment, which is composed of the first wind power prediction value, the first photovoltaic power prediction value and the data of the corresponding meteorological characteristics, and b is the bias term.

[0031] Optionally, optimize the scheduling of an integrated energy system including multiple devices, and construct an objective function, including:

[0032] First, determine the operating cost \(C\) of the integrated energy system IES as follows:

[0033] \(\min C\) IES \(=C\) EB \(-C\) ES \(+C\) G

[0034] In the above formula, \(C\) EB is the cost required for the integrated energy system to purchase electricity from the power grid when wind power generation and photovoltaic power generation are insufficient, and \(C\) ES is the income from selling electricity from the integrated energy system to the power grid, and \(C\) G is the cost of purchasing the required natural gas for the integrated energy system;

[0035] At the same time, determine the environmental cost \(C\) of the integrated energy system env as follows:

[0036]

[0037] In the above formula, is the penalty coefficient for the integrated energy system to generate \(CO_2\), \(\theta\) e and \(\theta\) g are the \(CO_2\) emissions per unit of natural gas and per unit of electric power respectively, and \(V\) MT 、\(V\) FC and \(V\) GB are the amounts of natural gas consumed by the gas turbine, fuel cell and gas boiler respectively;

[0038] Secondly, construct the following objective function according to the operating cost and environmental cost of the integrated energy system:

[0039] \(\min C = C\) IES \(+C\) env

[0040] In the above formula, \(\min C\) is the minimum value of the objective function.

[0041] Optionally, the constraint conditions of each device in the integrated energy system are as follows:

[0042] The electric power balance constraint is as follows:

[0043] \(P\) MT \(+P\) PV \(+P\) W \(+P\) FC \(+P\) grid \(=P\) ER \(+P\) EB \(+P\) user

[0044] In the above formula, \(P\) MT 、\(P\) FC 、\(P\)ER and P EB are the electric powers of the gas turbine, fuel cell, electric chiller, and electric boiler respectively, and P PV is the second power prediction value of the photovoltaic output power of the photovoltaic module, and P PW is the second power prediction value of the wind power output, and P grid is the electricity quantity purchased by the integrated energy system from the power grid. If P grid is negative, it means that the integrated energy system sells electricity to the power grid, and P user is the electricity load demand of users after the integrated energy system is optimized and adjusted;

[0045] The gas supply balance constraint is as follows:

[0046] V g = V user + V MT + V FC + V GB

[0047] In the above formula, V g is the natural gas quantity obtained by the system from the natural gas network, and V user is the required quantity for the gas load on the user side;

[0048] The heat energy balance constraint is as follows:

[0049] H user = H EB + H GB

[0050] In the above formula, H user is the heat load of users in the system, and H EB and H GB are the heat generation powers of the electric boiler and the gas boiler respectively.

[0051] Optionally, an improved particle swarm optimization algorithm combined with an adaptive inertia weight is used to optimize the objective function, including:

[0052] First, set the scale of the particle swarm and the initial state of the particles. Each particle updates its velocity and position according to the following formula to search for the individual optimal and global optimal particles:

[0053] v ij (t + 1)= ω(t)v ij (t)+ c1r1(t)[p ij - x ij (t)]+ c2r2(t)[p gj - x ij (t)]

[0054] x ij (t + 1)= x ij(t) + v ij (t + 1)

[0055] In the above formula, v ij (t) and x ij (t) are the position and velocity of the particle respectively; c1 and c2 are learning factors; r1 and r2 are random numbers within the range of [0, 1]; p ij is the individual optimal value obtained by iteration up to the current moment, and p gj is the global optimal value; ω(t) is the inertia weight coefficient, as follows:

[0056]

[0057] In the above formula, ω ini is the initial inertia weight, ω end is the termination inertia weight, ω ini = 0.9, ω end = 0.4; t max is the maximum number of iterations set initially;

[0058] Secondly, perturb the current particle swarm, calculate the individual optimal particle and the global optimal particle after perturbation. If the global optimal particle after re - search is better than the global optimal particle before perturbation, and the solution after perturbation satisfies the constraint conditions of each device, then accept this solution as the new individual optimal particle;

[0059] Finally, the end condition of the improved particle swarm algorithm is reaching the maximum number of iterations or the algorithm converges. If the end condition is reached, output the optimal solution, otherwise continue to update the particle state and search for the global optimal solution; the optimal solution obtained by the improved particle swarm algorithm is the corresponding optimal scheduling result of the integrated energy system.

[0060] Optionally, in step S100, the meteorological data includes solar irradiance, temperature, rainfall, humidity, and air pressure.

[0061] Optionally, in step S300, the multiple devices include wind turbines, photovoltaic systems, fuel cells, gas turbines, and electric chillers.

[0062] One technical effect of the present invention is that:

[0063] In the embodiment of the present application, first, the wind power output power and photovoltaic power generation output power of the integrated energy system are predicted to accurately obtain the predicted value of the wind power output power and the predicted value of the photovoltaic power generation output power; then, the integrated energy system including multiple devices is optimized and scheduled, a target function is constructed, and the constraint conditions of each device in the integrated energy system are determined according to the predicted value of the wind power output power and the predicted value of the photovoltaic power generation output power; finally, based on the constraint conditions, an improved particle swarm optimization algorithm combined with an adaptive inertia weight is used to optimize the target function, and the optimal strategy for optimizing the scheduling of the integrated energy system is obtained by solving.

[0064] Therefore, under the premise of considering the operating cost and environmental cost, the typical pipeline station integrated energy system scheduling method inputs the prediction results of the wind power output power and the photovoltaic power generation output power into the improved particle swarm optimization algorithm combined with the adaptive inertia weight for optimization, so as to obtain the optimal strategy for optimizing the scheduling of the integrated energy system, thereby realizing the optimization and scheduling of the integrated energy system, making the integrated energy system more economical and more environmentally friendly, and further balancing the environmental protection and economic needs of the integrated energy system. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic flowchart of a typical pipeline station integrated energy system scheduling method according to an embodiment of the present invention;

[0066] Figure 2 It is a schematic diagram of the integrated energy system architecture according to an embodiment of the present invention;

[0067] Figure 3 It is a flowchart of predicting the wind power output power and photovoltaic output power using an ARIMA-RNN combined model according to an embodiment of the present invention;

[0068] Figure 4 It is a topological structure diagram of the ARIMA-RNN combined model according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0069] Now, various exemplary embodiments of the present application will be described in detail with reference to the accompanying drawings. It should be noted that: Unless otherwise specifically stated, the relative arrangements, numerical expressions, and numerical values of the components and steps described in these embodiments do not limit the scope of the present application.

[0070] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present application and should not be construed as a limitation of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts belong to the scope protected by the present application.

[0071] The terms "first" and "second" in the description and claims of the present application may explicitly or implicitly include one or more of such features. In the description of the present application, unless otherwise specified, the meaning of "a plurality" is two or more. In addition, "and / or" in the description and claims means at least one of the connected objects. The character " / " generally indicates an "or" relationship between the associated objects before and after.

[0072] According to one aspect of the present invention, referring to Figure 1 , a typical scheduling method for an integrated energy system of a pipeline station yard is provided, including the following steps:

[0073] Step S100: Obtain the photovoltaic output power and corresponding meteorological data through a data acquisition unit, perform a correlation analysis on each meteorological feature of the meteorological data using the Pearson correlation coefficient, screen out the three meteorological features with the highest correlation with the photovoltaic output power, and eliminate redundant meteorological data.

[0074] For example, calculate using the photovoltaic output power of a certain 110 kV photovoltaic power station. The Pearson correlation coefficients of the three meteorological features with the highest degree of correlation are shown in Table 1. The larger the absolute value of the correlation coefficient, the higher the degree of correlation, and the negative value indicates a negative correlation.

[0075] Table 1 Pearson coefficients of photovoltaic output power and meteorological factors

[0076] Factor Solar irradiance Temperature Humidity Correlation coefficient 0.955 -0.619 -0.550

[0077] As can be seen from Table 1 above, solar irradiance is the most important factor affecting photovoltaic output power. The higher the irradiance, the greater the photovoltaic output power; temperature and humidity are negatively correlated with photovoltaic output power, and an increase in temperature and humidity will lead to a decrease in photovoltaic power.

[0078] Step S200: Use the ARIMA (Autoregressive Integrated Moving Average) model to predict the wind power output and photovoltaic power output at time t respectively to obtain the first power prediction values. Input the first power prediction values and meteorological data into the RNN (Recurrent Neural Network) model for training to obtain the ARIMA-RNN combined model. Use the ARIMA-RNN combined model to predict the wind power output and photovoltaic power output at time t respectively to obtain the second power prediction values. The ARIMA-RNN combined model helps to accurately predict the wind power output and photovoltaic power output at time t (i.e., the current time).

[0079] Step S300: Optimize the scheduling of the integrated energy system including multiple devices to minimize the operating cost and environmental cost. Construct the objective function and determine the constraint conditions of each device in the integrated energy system according to the second power prediction values.

[0080] Step S400: Based on the constraint conditions, use the improved particle swarm optimization algorithm combined with the adaptive inertia weight to optimize the objective function, and solve to obtain the optimal scheduling strategy for the integrated energy system optimization. Among them, the optimal scheduling strategy for the integrated energy system optimization takes into account both the operating cost and the environmental cost.

[0081] In the embodiment of the present application, this typical pipeline station integrated energy system scheduling method, on the premise of considering the operating cost and environmental cost, inputs the prediction results of the wind power output and the photovoltaic power generation output into the improved particle swarm optimization algorithm combined with the adaptive inertia weight for optimization, so as to obtain the optimal scheduling strategy for the integrated energy system optimization, thereby realizing the optimization scheduling of the integrated energy system, making the integrated energy system more economical and more environmentally friendly, and further balancing the environmental protection and economic needs of the integrated energy system.

[0082] Optionally, use the Pearson correlation coefficient to perform correlation analysis on each meteorological feature of the meteorological data, including:

[0083] Take the photovoltaic power output as the target vector, and take the meteorological features of the meteorological data affecting photovoltaic power generation as the feature vector. Use the Pearson correlation coefficient to perform correlation analysis on the feature vector. Its calculation formula is as follows:

[0084]

[0085] In the above formula, r is the Pearson correlation coefficient, N is the number of samples, X i is the sample value of the target vector, is the mean value of the target vector, Y i is the sample value of the feature vector, is the mean value of the feature vector.

[0086] In the above embodiment, the Pearson correlation coefficient is used to perform correlation analysis on each meteorological feature of the meteorological data, so as to screen out the three meteorological features with the highest correlation with the photovoltaic output power, which helps to accurately and quickly eliminate redundant meteorological data.

[0087] Optionally, before using the ARIMA model to predict the wind power output and photovoltaic output at time t respectively to obtain the first power prediction value, it further includes:

[0088] First, the data is normalized using the following formula. For example, the data is mapped to [-1, 1], that is, the mapping range is [-1, 1], to improve the convergence speed of the subsequent prediction process;

[0089]

[0090] In the above formula, X is the original time series of the data, x max is the maximum value of the data, is the average value of the data, is the time series after normalization; where the data includes power data and meteorological data:

[0091] Then, an ARIMA(p, d, q) model is constructed, including:

[0092] The unit root test method is used to perform a stationarity test on the power data. If the power data is a non-stationary time series, the power data is differenced. After n-order differencing, the power data becomes a stationary series, and at this time d = n; if the power data is a stationary series, then d = 0; calculate the autocorrelation and partial autocorrelation functions, determine the value ranges of the autoregressive term p and the moving average term q, and use the Akaike information criterion to order p and q to obtain the ARIMA model.

[0093] In the above embodiment, normalizing the data helps to improve the convergence speed of the subsequent prediction process. At the same time, the construction of the ARIMA(p, d, q) model is relatively reasonable, and it can quickly predict the wind power output and photovoltaic output at time t.

[0094] Optionally, using the ARIMA model to predict the wind power output and photovoltaic output at time t respectively to obtain the first power prediction value includes:

[0095] The ARIMA model is trained using historical data; where the historical data includes the historical data of wind power output and the historical data of photovoltaic output;

[0096] The trained ARIMA model is used to predict the wind power output and photovoltaic output at time t respectively to obtain the first wind power prediction value at time t and the first photovoltaic power prediction value

[0097] In the above embodiment, the ARIMA model is trained first, and then the trained ARIMA model is used to predict the wind power output and photovoltaic power output at time t respectively, which helps to improve the accuracy of the prediction.

[0098] Optionally, inputting the first power prediction value and meteorological data into the RNN model for training includes:

[0099] For the photovoltaic power output, the data of the three meteorological characteristics and the first photovoltaic power prediction value are jointly used as input features and input into the RNN model for training;

[0100] For the wind power output, the data of the wind speed as the meteorological characteristic and the first wind power prediction value are used as input features and input into the RNN model for training;

[0101] Among them, the RNN model is constructed by the following formula:

[0102]

[0103] In the above formula, is the predicted value of the wind power output or photovoltaic power output, g(·) is the softmax function, V is the parameter matrix of the hidden layer and output layer of the recurrent neural network, c is the bias term, and h t is the state information of the hidden layer, and h is defined by the following formula t :

[0104] h t = f(Wh t-1 + Ux t + b)

[0105] In the above formula, f(·) is a non-linear activation function, that is, the sigmoid activation function, W and U are the parameter matrices of the hidden layer at the previous moment and the parameter matrix of the current moment input respectively, h t-1 is the state information at the previous moment, x t is the input at the current moment, which is composed of the first wind power prediction value, the first photovoltaic power prediction value and the data of the corresponding meteorological characteristics, and b is the bias term.

[0106] In the above embodiment, inputting the first power prediction value and meteorological data into the RNN model for training helps to ensure the accuracy of the ARIMA-RNN combined model in predicting the wind power output and photovoltaic power output at time t.

[0107] In a specific embodiment, refer toFigure 3 , first, on the one hand, obtain wind power and photovoltaic power data, that is, historical data, construct and train an ARIMA model; then, output the power prediction value through the ARIMA model, that is, the first power prediction value; on the other hand, obtain the corresponding meteorological data; then, conduct a correlation analysis of the meteorological data. Secondly, input the power prediction value output by the ARIMA model and the correlation analysis result of the meteorological data into an RNN (recurrent neural network) model. Finally, output the wind power and photovoltaic power prediction values, that is, the second power prediction value.

[0108] In the embodiment of the present application, the topological structure diagram of the ARIMA-RNN combined model is as Figure 4 , input the power prediction value at time t, the meteorological data at time t-1, the meteorological data at time t-2,... the meteorological data at time t-n based on the ARIMA model into the corresponding input layer respectively, and input the results of the input layer into the hidden layer, and obtain the output layer according to the output result of the hidden layer, that is, the prediction value at time t.

[0109] Optionally, the integrated energy system architecture is as Figure 2 shown, optimize the scheduling of the integrated energy system including multiple devices, and construct an objective function, including:

[0110] First, determine the operating cost C of the integrated energy system IES as follows:

[0111] minC IES = C EB - C ES + C G

[0112] In the above formula, C EB is the cost required for the integrated energy system to purchase electricity from the power grid when the wind power generation and photovoltaic power generation are insufficient, C ES is the income obtained by the integrated energy system from selling electricity to the power grid, C G is the cost for the integrated energy system to purchase the required natural gas;

[0113] At the same time, determine the environmental cost C of the integrated energy system env as follows:

[0114]

[0115] In the above formula, is the penalty coefficient for the integrated energy system to generate CO2, θ e and θ g are the CO2 emissions per unit of natural gas and per unit of electric power respectively, V MT , V FC and V GBThe natural gas consumption of the gas turbine, fuel cell, and gas boiler respectively;

[0116] Secondly, the following objective function is constructed according to the operating cost and environmental cost of the integrated energy system:

[0117] minC = C IES + C env

[0118] In the above formula, minC is the minimum value of the objective function.

[0119] In the above embodiment, constructing the objective function on the premise of comprehensively balancing the operating cost and environmental cost helps to ensure the effect of the optimal dispatching of the integrated energy system.

[0120] Optionally, the active power of each device and the load consumption in the integrated energy system need to be balanced. The constraint conditions of each device in the integrated energy system are as follows:

[0121] The power balance constraint of electricity is as follows:

[0122] P MT + P PV + P W + P FC + P grid = P ER + P EB + P user

[0123] In the above formula, P MT 、P FC 、P ER and P EB are the electric powers of the gas turbine, fuel cell, electric chiller, and electric boiler respectively, P PV is the second power prediction value of the photovoltaic output power of the photovoltaic module, P PW is the second power prediction value of the wind power output, P grid is the electricity quantity purchased by the integrated energy system from the power grid. If P grid is negative, it means that the integrated energy system sells electricity to the power grid, and P user is the electricity load demand of users after the integrated energy system is optimized and adjusted;

[0124] The gas supply balance constraint conditions are as follows:

[0125] V g = V user + V MT + V FC + V GB

[0126] In the above formula, V g is the natural gas quantity obtained by the system from the natural gas network, Vuser is the required amount of gas load on the user side;

[0127] The heat energy balance constraint conditions are as follows:

[0128] H user = H EB + H GB

[0129] In the above formula, H user is the heat load of the user in the system, H EB and H GB are the heat production powers of the electric boiler and the gas boiler respectively.

[0130] In the above embodiment, the constraint conditions are set reasonably, which helps to maintain the balance of the active power of each device and the load consumption in the integrated energy system, thus contributing to ensuring the optimization scheduling effect of the integrated energy system.

[0131] Optionally, an improved particle swarm optimization algorithm combined with an adaptive inertia weight is used to optimize the objective function, including:

[0132] First, set the scale of the particle swarm and the initial state of the particles. Each particle updates its velocity and position according to the following formula to search for the individual optimal and global optimal particles:

[0133] v ij (t + 1)= ω(t)v ij (t)+ c1r1(t)[p ij - x ij (t)]+ c2r2(t)[p gj - x ij (t)]

[0134] x ij (t + 1)= x ij (t)+ v ij (t + 1)

[0135] In the above formula, v ij (t) and x ij (t) are the position and velocity of the particle respectively; c1 and c2 are learning factors; r1 and r2 are random numbers in the range of [0, 1]; p ij is the individual optimal value obtained by iterating to the current moment, p gj is the global optimal value; ω(t) is the inertia weight coefficient, as follows:

[0136]

[0137] In the above formula, ω ini is the initial inertia weight, ω end is the termination inertia weight, ω ini= 0.9, ω end = 0.4; t max is the maximum number of initial set iterations;

[0138] Secondly, perturb the current particle swarm, calculate the individual optimal particle and the global optimal particle after perturbation. If the globally optimal particle after rediscovery is better than the globally optimal particle before perturbation, and the solution after perturbation satisfies the constraint conditions of each device, then accept this solution as the new individual optimal particle;

[0139] Finally, the end condition of the improved particle swarm algorithm is to reach the maximum number of iterations or the algorithm converges. If the end condition is reached, the optimal solution is output, otherwise, continue to update the particle state and search for the global optimal solution; The optimal solution obtained by the improved particle swarm algorithm is the optimal scheduling result of the corresponding integrated energy system.

[0140] In the above embodiment, the method of optimizing the objective function is relatively reasonable, which helps to obtain the optimal solution through the improved particle swarm algorithm, that is, the optimal scheduling result of the corresponding integrated energy system, and the solution result is relatively accurate.

[0141] Optionally, in step S100, the meteorological data includes solar irradiance, temperature, rainfall, humidity, and air pressure.

[0142] In the above embodiment, using the Pearson correlation coefficient to analyze the correlation of the above multiple meteorological characteristics of the meteorological data helps to accurately screen out the three meteorological characteristics with the highest correlation with the photovoltaic output power.

[0143] Optionally, in step S300, the multiple devices include wind turbines, photovoltaic systems, fuel cells, gas turbines, and electric chillers.

[0144] In the above embodiment, accurately optimizing the scheduling of the integrated energy system including multiple devices makes the integrated energy system more economical and more environmentally friendly, and further balances the environmental protection and economic needs of the integrated energy system.

[0145] It can be understood that the above embodiments are only exemplary embodiments adopted to illustrate the principle of the present invention. However, the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also regarded as the protection scope of the present invention.

Claims

1. A scheduling method for an integrated energy system of a typical pipeline station yard, characterized in that, It includes the following steps: Step S100: Obtain the photovoltaic output power and corresponding meteorological data through the data acquisition unit, perform correlation analysis on each meteorological feature of the meteorological data using the Pearson correlation coefficient, screen out the three meteorological features with the highest correlation with the photovoltaic output power, and eliminate redundant meteorological data; Step S200: Use the ARIMA model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the first power prediction value, input the first power prediction value and meteorological data into the RNN model for training to obtain the ARIMA-RNN combined model, and use the ARIMA-RNN combined model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the second power prediction value; Step S300: Optimize and dispatch the integrated energy system including multiple devices, construct the objective function, and determine the constraint conditions of each device in the integrated energy system according to the second power prediction value; Step S400: Based on the constraint conditions, use the improved particle swarm optimization algorithm combined with the adaptive inertia weight to optimize the objective function, and solve to obtain the optimal strategy for the optimization dispatch of the integrated energy system.

2. The scheduling method for the integrated energy system of a typical pipeline station yard according to claim 1, wherein Performing correlation analysis on each meteorological feature of the meteorological data using the Pearson correlation coefficient includes: Taking the photovoltaic output power as the target vector, taking the meteorological features of the meteorological data affecting photovoltaic power generation as the feature vector, and performing correlation analysis on the feature vector using the Pearson correlation coefficient. The calculation formula is as follows: In the above formula, r is the Pearson correlation coefficient, N is the number of samples, X i is the sample value of the target vector, is the mean of the target vector, Y i is the sample value of the feature vector, is the mean of the feature vector.

3. The scheduling method for the integrated energy system of a typical pipeline station yard according to claim 2, wherein Before using the ARIMA model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the first power prediction value, it also includes: First, normalize the data using the following formula; In the above formula, X is the original time series of the data, and x max is the maximum value of the data, is the average value of the data, is the time series after normalization processing; where the data includes power data and meteorological data: Then, construct the ARIMA(p,d,q) model, including: Use the unit root test method to perform stationarity test on the power data. If the power data is a non-stationary time series, perform differencing on the power data. After n-order differencing, the power data becomes a stationary series, and at this time d = n; if the power data is a stationary series, then d = 0; calculate the autocorrelation and partial autocorrelation functions, determine the value ranges of the autoregressive term p and the moving average term q, and use the Akaike information criterion to order p and q to obtain the ARIMA model.

4. The dispatching method for the typical pipeline station integrated energy system according to claim 3, characterized in that Using the ARIMA model to predict the wind power output power and photovoltaic output power at time t respectively to obtain the first power prediction value includes: Train the ARIMA model using historical data; among them, the historical data includes the historical data of wind power output power and the historical data of photovoltaic output power; Use the trained ARIMA model to predict the wind power output and photovoltaic power output at time t respectively, and obtain the first wind power prediction value at time t and the first photovoltaic power prediction value 5. The dispatching method for the integrated energy system of a typical pipeline station yard according to claim 1, characterized in that Inputting the first power prediction value and meteorological data into the RNN model for training includes: For the photovoltaic output power, the data of the three meteorological characteristics and the first photovoltaic power prediction value are jointly used as input features and input into the RNN model for training; For the wind power output, the wind speed, which is the data of meteorological characteristics, and the first wind power prediction value are used as input features and input into the RNN model for training; Among them, construct the RNN model using the following formula: In the above formula, is the predicted value of wind power output or photovoltaic power output, g(·) is the softmax function, V is the parameter matrix of the hidden layer and output layer of the recurrent neural network, c is the bias term, and h t is the state information of the hidden layer, and h is defined by the following formula t : h t = f(Wh t-1 + Ux t + b) In the above formula, f(·) is a non-linear activation function, namely the sigmoid activation function. W and U are the parameter matrices of the hidden layer at the previous moment and the parameter matrix of the current input respectively, and h t-1 is the state information at the previous moment, and x t is the input at the current moment, which is composed of the first wind power prediction value, the first photovoltaic power prediction value, and the data of the corresponding meteorological characteristics. b is the bias term.

6. The scheduling method for the integrated energy system of a typical pipeline station yard according to claim 5, characterized in that, Optimizing and dispatching the integrated energy system including multiple devices and constructing the objective function includes: First, determine the operating cost C of the integrated energy system IES as follows: min C IES = C EB - C ES + C G In the above formula, C EB is the required cost for the integrated energy system to purchase electricity from the power grid when wind power generation and photovoltaic power generation are insufficient. C ES is the income from selling electricity from the integrated energy system to the power grid. C G is the cost for the integrated energy system to purchase the required natural gas; Meanwhile, determine the environmental cost C of the integrated energy system env as follows: In the above formula, is the penalty coefficient for CO2 generated by the integrated energy system, and θ e and θ g are the CO2 emissions per unit of natural gas and per unit of electric power respectively. V MT , V FC and V GB are the amounts of natural gas consumed by the gas turbine, fuel cell, and gas boiler respectively; Secondly, construct the following objective function according to the operating cost and environmental cost of the integrated energy system: min C = C IES + C env In the above formula, minC is the minimum value of the objective function.

7. The scheduling method for the integrated energy system of a typical pipeline station according to claim 6, wherein The constraint conditions of each device in the integrated energy system are as follows: The electric power balance constraint is as follows: P MT +P PV +P W +P FC +P grid =P ER +P EB +P user In the above formula, P MT , P FC , P ER and P EB are the electric powers of the gas turbine, fuel cell, electric chiller, and electric boiler respectively, P PV is the second power prediction value of the photovoltaic output power of the photovoltaic module, P PW is the second power prediction value of the wind power output, P grid is the electricity quantity purchased by the integrated energy system from the power grid. If P grid is negative, it means that the integrated energy system sells electricity to the power grid, and P user is the electric load demand of the user after the integrated energy system is optimized and adjusted; The gas supply balance constraint condition is as follows: V g = V user + V MT + V FC + V GB In the above formula, V g is the amount of natural gas obtained by the system from the natural gas network, and V user is the amount required for the gas consumption load on the user side; The heat energy balance constraint condition is as follows: H user = H EB + H GB In the above formula, H user is the heat load of the user in the system, H EB and H GB are the heat production powers of the electric boiler and the gas boiler respectively.

8. The scheduling method for the integrated energy system of a typical pipeline station according to claim 7, characterized in that Optimize the objective function by using an improved particle swarm optimization algorithm combined with an adaptive inertia weight, including: First, set the scale of the particle swarm and the initial state of the particles. Each particle updates its velocity and position according to the following formula to search for the individual optimal and global optimal particles: v ij (t + 1) = ω(t)v ij (t) + c1r1(t)[p ij -x ij (t)] + c2r2(t)[p gj -x ij (t)] x ij (t + 1)=x ij (t)+v ij (t + 1) In the above formula, v ij (t) and x ij (t) are the position and velocity of the particle respectively; c1 and c2 are learning factors; r1 and r2 are random numbers in the range of [0, 1]; p ij is the individual optimal value obtained by iteration up to the current moment, and p gj is the global optimal value; ω(t) is the inertia weight coefficient, as follows: In the above formula, ω ini is the initial inertia weight, ω end is the termination inertia weight, ω ini = 0.9, ω end = 0.4; t max is the maximum number of iterations set initially; Second, perturb the current particle swarm, calculate the individual optimal particle and the global optimal particle after perturbation. If the global optimal particle after re-search is better than the global optimal particle before perturbation and the solution after perturbation satisfies the constraint conditions of each device, then accept this solution as the new individual optimal particle; Finally, the termination condition of the improved particle swarm optimization algorithm is to reach the maximum number of iterations or the algorithm converges. If the termination condition is reached, output the optimal solution; otherwise, continue to update the particle state and search for the global optimal solution; the optimal solution obtained by the improved particle swarm optimization algorithm is the corresponding optimal scheduling result of the integrated energy system.

9. The scheduling method of the typical pipeline station integrated energy system according to claim 1, characterized in that In step S100, the meteorological data includes solar irradiance, temperature, rainfall, humidity, and air pressure.

10. The scheduling method of the typical pipeline station integrated energy system according to claim 1, characterized in that, In step S300, the multiple devices include wind turbines, photovoltaic systems, fuel cells, gas turbines, and electric chillers.

Images