An intelligent decision-making method and system based on artificial intelligence

Abstract

The present invention relates to the technical field of electric energy balance strategies, and discloses an intelligent decision-making method and system based on artificial intelligence, wherein an intelligent decision-making method based on artificial intelligence comprises the following steps: collecting historical hydrological and meteorological data, historical electric energy demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data of all hydropower stations in a hydropower station group for preprocessing; triggering an early warning if the average rainfall in the basin where the hydropower group is located in the next day is greater than a set rainfall threshold; encoding and combining the collected data of the basin into a first sequence of data and graph structure data; inputting the first sequence of data into a first neural network, and outputting a selected action; decoding and selecting a corresponding electric energy balance strategy according to the selected action; the present invention can issue an early warning and provide an electric energy balance strategy when extreme weather occurs.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy balance strategies, and more specifically, to an intelligent decision-making method and system based on artificial intelligence. Background Art

[0002] A river basin is divided into several small areas, each of which has a corresponding hydropower station. They are distributed in different rivers and together form a large group of hydropower stations. There are transmission lines between hydropower stations on the same river. Due to the complex and changeable climate conditions in this river basin and the uneven distribution of rainfall, the actual output of each hydropower station fluctuates greatly. At the same time, the industrial and agricultural production and urban electricity demand in the region also show obvious seasonal and uncertain characteristics. Some extreme weather may occur during power generation and power transmission. For example, heavy rain may cause floods, and lightning may cause damage to transmission lines to a certain extent, affecting the power generation and power transmission capacity of hydropower stations, which may lead to insufficient power supply in the region and power outages.

[0003] The traditional manual experience decision-making model is no longer able to cope with this changing natural and social environment. If the decision is inappropriate, the surplus power of some power stations may be idle and wasted, while other power stations are facing power supply pressure and cannot meet local electricity demand, resulting in a waste of resources. It is necessary to provide an energy balance strategy that can issue an early warning and provide an energy balance strategy in time when extreme weather occurs, take some corresponding measures to maximize the power generation efficiency and ensure that each power station in the hydropower station group reasonably dispatches the power energy. Summary of the invention

[0004] The present invention provides an intelligent decision-making method and system based on artificial intelligence, which solves the technical problem in the related art that it is difficult to make appropriate decisions in time when extreme weather occurs, resulting in power outages and waste of resources.

[0005] The present invention provides an intelligent decision-making method based on artificial intelligence, comprising the following steps:

[0006] Step 100, collecting historical hydrological and meteorological data, historical power demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data of all hydropower stations in the hydropower station group for preprocessing;

[0007] In a preferred embodiment, the collected data includes:

[0008] Historical hydrological and meteorological data include: hydrological and meteorological data recorded N times before the current moment, including average rainfall, average flow, average temperature, average humidity, and average wind speed;

[0009] The historical power demand data includes: the power demand data recorded N times before the current moment, and the power demand data includes the average power demand;

[0010] The historical hydropower station operation data includes: the hydropower station operation data recorded N times before the current time, and the hydropower station operation data includes the average reservoir water level, the average inflow flow, the average outflow flow, and the power supply of the power station;

[0011] The historical transmission line monitoring data includes: the transmission line monitoring data recorded N times before the current moment, and the transmission line monitoring data includes the average current, average voltage, average temperature, and electric energy transmission amount of the transmission line;

[0012] The historical meteorological department forecast data includes: the forecast data of the river basin where the hydropower station is located recorded N times before the current moment, and the forecast data includes the rainfall, rainfall intensity, and rainfall duration for the next day.

[0013] Step 200, if the average rainfall in the basin where the hydropower group is located in the next day is greater than the set rainfall threshold, an early warning is triggered, and step 300 is executed when the early warning is triggered;

[0014] Step 300, the collected hydrological and meteorological data of all hydropower stations in the hydropower station group in the basin, historical hydrological and meteorological data, historical electric energy demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data are coded and combined into a first sequence of data and graph structure data;

[0015] In a preferred embodiment, the first sequence data includes N first sequence units, and one sequence unit represents hydrological and meteorological data, electric energy demand data, hydropower station operation data, transmission line monitoring data and meteorological department forecast data recorded once.

[0016] In a preferred embodiment, the graph structure data includes nodes and edges. The nodes represent hydropower stations. If there is a transmission line between two hydropower stations, there is an edge between the nodes corresponding to the two hydropower stations, and there is an edge between the nodes representing the hydropower stations on the same river channel.

[0017] Step 400, inputting the first sequence data into the first neural network, and outputting the selected action, the first neural network comprising a first module, a second module and an output module;

[0018] In a preferred embodiment, the first module in the first neural network is used to take the first sequence data as input and output the first hidden state; the second module is used to take the graph structure data and the first hidden state as input and output the second hidden state; the output module is used to take the second hidden state as input and output the selected action.

[0019] In a preferred embodiment, the action is selected from an action space, which includes all possible selected actions. An action is represented by an action matrix. The elements in the i-th row and j-th column of the action matrix represent the amount of electric energy transmitted from the i-th power station to the j-th power station, and the elements in the i-th row and i-th column represent the amount of electric energy supplied to the i-th power station.

[0020] In a preferred embodiment, the calculation formula of the first module is as follows:

[0021] ;

[0022] ;

[0023] ;

[0024] ;

[0025] in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, where represents the tth first sequence unit of the sequence data, and denote the t-th and t-1-th first hidden states respectively, and the n-th sequence hidden state is the first hidden state of the node. , N represents the total number of sequence units of the sequence data, when t=1 , is the hyperbolic tangent function, Represents a sigmoid function.

[0026] In a preferred embodiment, the calculation formula of the second module is as follows:

[0027] ;

[0028] in represents the second hidden state of the i-th node, represents the aggregation coefficient of the i-th node, represents the set of nodes that have edges with the i-th node, represents the activation function, represents the state weight coefficient, represents the first hidden state of the jth node;

[0029] ;

[0030] ;

[0031] ;

[0032] in and Respectively represent and The first hidden state of a node, represents the aggregation weight coefficient, represents the splicing weight coefficient, represents transpose, represents an exponential function with a natural constant as the base, LeakyRelu represents a modified linear unit function, Indicates The set of nodes with which there are edges.

[0033] In a preferred embodiment, the output module calculation formula is as follows:

[0034] ;

[0035] in represents the control vector, a component of the control vector represents the probability value of an action, and the set of all actions is called the action space, which contains all possible actions that may be selected. represents the second hidden state of the i-th node, represents the concatenation of the hidden states of all nodes, U represents the set of all nodes, is the fourth weight parameter, is the fourth bias parameter, Represents a sigmoid function.

[0036] Step 500 , selecting a corresponding power balancing strategy according to the selected action decoding, the power balancing strategy including the power transmission amount transmitted between power stations and the power supply amount of each power station.

[0037] The beneficial effects of the present invention are:

[0038] The present invention can issue an early warning when extreme weather occurs and provide an electric energy balance strategy, through which the power generation efficiency can be maximized and the reasonable scheduling of electric energy by each power station in the hydropower station group can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of an intelligent decision-making method based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0040] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Various examples may omit, replace or add various processes or components as needed. In addition, the features described in some examples may also be combined in other examples.

[0041] At least one embodiment of the present invention discloses an intelligent decision-making method based on artificial intelligence, such as Figure 1 As shown, the following steps are included:

[0042] Step 100, collecting historical hydrological and meteorological data, historical power demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data of all hydropower stations in the hydropower station group for preprocessing;

[0043] The historical hydrological and meteorological data include: hydrological and meteorological data recorded N times before the current moment, including average rainfall, average flow, average temperature, average humidity, and average wind speed;

[0044] The historical power demand data includes: the power demand data recorded N times before the current moment, and the power demand data includes the average power demand;

[0045] The historical hydropower station operation data includes: the hydropower station operation data recorded N times before the current time, and the hydropower station operation data includes the average reservoir water level, the average inflow flow, the average outflow flow, and the power supply of the power station;

[0046] The historical transmission line monitoring data includes: the transmission line monitoring data recorded N times before the current moment, and the transmission line monitoring data includes the average current, average voltage, average temperature, and electric energy transmission amount of the transmission line.

[0047] The historical meteorological department forecast data includes: the forecast data of the basin where the hydropower station is located recorded N times before the current moment, and the forecast data includes the rainfall, rainfall intensity, and rainfall duration in the next day;

[0048] In one embodiment of the present invention, the time interval between two adjacent records is The average rainfall, average flow, average temperature, average humidity, and average wind speed recorded at the Nth time are the time before the Nth record. The average value within the time period.

[0049] The N records before the current moment include the record at the current moment;

[0050] In one embodiment of the present invention, the time period The default value of is 10 days, and the default value of N is 20.

[0051] Step 200, if the average rainfall in the basin where the hydropower group is located in the next day is greater than the set rainfall threshold, an early warning is triggered, and step 300 is executed when the early warning is triggered;

[0052] In one embodiment of the present invention, the default value of the rainfall threshold is 150 mm.

[0053] Step 300, the collected hydrological and meteorological data of all hydropower stations in the hydropower station group in the basin, historical hydrological and meteorological data, historical electric energy demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data are coded and combined into a first sequence of data and graph structure data;

[0054] The first sequence data includes N first sequence units, and one sequence unit represents hydrological and meteorological data, electric energy demand data, hydropower station operation data, transmission line monitoring data and meteorological department forecast data recorded once;

[0055] Generate graph structure data based on the collected data. The graph structure data includes nodes and edges. Nodes represent hydropower stations. If there is a transmission line between two hydropower stations, there is an edge between the nodes corresponding to the two hydropower stations. This means there is an edge between the nodes of the hydropower stations on the same river.

[0056] Step 400, inputting the first sequence data into the first neural network, and outputting the selected action, the first neural network comprising a first module, a second module and an output module;

[0057] The first module is used to take the first sequence data as input and output the first hidden state;

[0058] The second module is used to take the graph structure data and the first hidden state as input and output the second hidden state;

[0059] The output module is used to take the second hidden state as input and output the selected action;

[0060] In one embodiment of the present invention, the action is selected from an action space, the action space includes all possible selected actions, an action is represented by an action matrix, the element of the i-th row and j-th column of the action matrix represents the amount of electric energy transmitted from the i-th power station to the j-th power station, and the element of the i-th row and i-th column represents the amount of electric energy supplied by the i-th power station;

[0061] In one embodiment of the present invention, the power supply amount and the power transmission amount are mean discretized within their respective value ranges to obtain multiple point values, and the actions in the action space are obtained by randomly combining the point values ​​of the power supply amount and the power transmission amount.

[0062] In one embodiment of the present invention, the calculation formula of the first module is as follows:

[0063] ;

[0064] ;

[0065] ;

[0066] ;

[0067] in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, where represents the tth first sequence unit of the sequence data, and denote the t-th and t-1-th first hidden states respectively, and the n-th sequence hidden state is the first hidden state of the node. , N represents the total number of sequence units of the sequence data, when t=1 , is the hyperbolic tangent function, Represents a sigmoid function.

[0068] In one embodiment of the present invention, the calculation formula of the second module is as follows:

[0069] ;

[0070] in represents the second hidden state of the i-th node, represents the aggregation coefficient of the i-th node, represents the set of nodes that have edges with the i-th node, represents the activation function, represents the state weight coefficient, represents the first hidden state of the jth node;

[0071] ;

[0072] ;

[0073] ;

[0074] in and Respectively represent and The first hidden state of a node, represents the aggregation weight coefficient, represents the splicing weight coefficient, represents transpose, represents an exponential function with a natural constant as the base, LeakyRelu represents a modified linear unit function, Indicates The set of nodes with which there are edges.

[0075] In one embodiment of the present invention, the output module calculation formula is as follows:

[0076] ;

[0077] in represents the control vector, a component of the control vector represents the probability value of an action, and the set of all actions is called the action space, which contains all possible actions that may be selected. represents the second hidden state of the i-th node, represents the concatenation of the hidden states of all nodes, U represents the set of all nodes, is the fourth weight parameter, is the fourth bias parameter, Represents a sigmoid function.

[0078] Step 500 , selecting a corresponding power balancing strategy according to the selected action decoding, the power balancing strategy including the power transmission amount transmitted between power stations and the power supply amount of each power station.

[0079] In one embodiment of the present invention, the training method of the first neural network comprises the following steps:

[0080] Step 101, constructing a target network, wherein the target network replicates the first neural network;

[0081] Step 102, randomly obtain an experience from the experience pool, the experience includes the state at time t , the action performed at time t , the state at time t+1 , perform the action Rewards ;

[0082] The states at time t and t+1 represent the first sequence data and graph structure data at time t and t+1 of the first neural network input respectively;

[0083] Then calculate the state The corresponding output of the target network The probability value ;

[0084] ;

[0085] express The maximum probability value of the target network output;

[0086] Step 103, according to and Calculate the loss value loss and update the first neural network;

[0087] ;

[0088] express The output of the first neural network corresponds to The probability value of

[0089] is the discount factor, a value between 0 and 1;

[0090] Step 104, updating the target network every fixed number of network updates so that its parameters are the same as those of the current first neural network;

[0091] Step 105, until the number of collected experiences reaches a set threshold or the first neural network converges, the step is terminated;

[0092] award The calculation formula is as follows:

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] Where m is the number of power stations, is the electricity price of the ith power station at the tth moment, is the power generation cost of the ith power station, is the output of the ith power station at time t+1, is the reward value that maximizes the power generation revenue; is the adjusted reservoir water level of the i-th power station at time t+1, is the safe water level of the i-th power station. When the water level is too high or too low, it will be punished. It is the reward value when the reservoir water level deviates from the safe water level during extreme weather; is the total load of the power grid at time t+1, is the reward value when the output and load are unbalanced, , , Represents the first, second, and third weight coefficients.

[0098] At least one embodiment of the present invention discloses an artificial intelligence-based intelligent decision-making system, which is used to execute the aforementioned artificial intelligence-based intelligent decision-making method.

[0099] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation mode. The above-mentioned specific implementation mode is merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make more forms of equivalent embodiments, all of which are within the protection of this embodiment.

Claims

1. An intelligent decision-making method based on artificial intelligence, characterized in that: The following steps are involved: Step 100, collecting historical hydrological and meteorological data, historical power demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data of all hydropower stations in the hydropower station group for preprocessing; Step 200, if the average rainfall in the basin where the hydropower group is located in the next day is greater than the set rainfall threshold, an early warning is triggered, and step 300 is executed when the early warning is triggered; Step 300, the collected hydrological and meteorological data of all hydropower stations in the hydropower station group in the basin, historical hydrological and meteorological data, historical electric energy demand data, historical hydropower station operation data, historical transmission line monitoring data and historical meteorological department forecast data are coded and combined into a first sequence of data and graph structure data; The first sequence data includes N first sequence units, and one sequence unit represents hydrological and meteorological data, electric energy demand data, hydropower station operation data, transmission line monitoring data and meteorological department forecast data recorded once; Graph structure data includes nodes and edges. Nodes represent hydropower stations. If there is a transmission line between two hydropower stations, there will be an edge between the nodes corresponding to the two hydropower stations. This means there is an edge between the nodes of the hydropower stations on the same river. Step 400, inputting the first sequence data into the first neural network, and outputting the selected action, the first neural network comprising a first module, a second module and an output module; The first module is used to take the first sequence data as input and output the first hidden state; The second module is used to take the graph structure data and the first hidden state as input and output the second hidden state; The output module is used to take the second hidden state as input and output the selected action; The action is selected from the action space, which includes all possible selected actions. An action is represented by an action matrix. The element in the i-th row and j-th column of the action matrix represents the amount of electric energy transmitted from the i-th power station to the j-th power station, and the element in the i-th row and i-th column represents the amount of electric energy supplied by the i-th power station. Step 500 , selecting a corresponding power balancing strategy according to the selected action decoding, the power balancing strategy including the power transmission amount transmitted between power stations and the power supply amount of each power station.

2. The intelligent decision-making method based on artificial intelligence according to claim 1, characterized in that: The calculation formula for the first module is as follows: ； ； ； ； in, , , represents the first, second, and third weight parameters, , , represents the first, second and third bias parameters, represents the dot product, where represents the tth first sequence unit of the sequence data, and denote the t-th and t-1-th first hidden states respectively, and the n-th sequence hidden state is the first hidden state of the node. , N represents the total number of sequence units of the sequence data, when t=1 , is the hyperbolic tangent function, represents the sigmoid function; The calculation formula for the second module is as follows: ； in represents the second hidden state of the i-th node, represents the aggregation coefficient of the i-th node, represents the set of nodes that have edges with the i-th node, represents the S-type function, represents the state weight coefficient, represents the first hidden state of the jth node; ； ； ； in and Respectively represent and The first hidden state of a node, represents the aggregation weight coefficient, represents the splicing weight coefficient, represents transpose, represents an exponential function with a natural constant as the base, and LeakyRelu represents a modified linear unit function; The output module calculation formula is as follows: ； in represents the control vector, a component of the control vector represents the probability value of an action, and the set of all actions is called the action space, which contains all possible actions that may be selected. represents the second hidden state of the i-th node, represents the concatenation of the hidden states of all nodes, U represents the set of all nodes, is the fourth weight parameter, is the fourth bias parameter, Represents a sigmoid function.

3. The intelligent decision-making method based on artificial intelligence according to claim 1, characterized in that: The collected data include: Historical hydrological and meteorological data include: hydrological and meteorological data recorded N times before the current moment, including average rainfall, average flow, average temperature, average humidity, and average wind speed; The historical power demand data includes: the power demand data recorded N times before the current moment, and the power demand data includes the average power demand; The historical hydropower station operation data includes: the hydropower station operation data recorded N times before the current time, and the hydropower station operation data includes the average reservoir water level, the average inflow flow, the average outflow flow, and the power supply of the power station; The historical transmission line monitoring data includes: the transmission line monitoring data recorded N times before the current moment, and the transmission line monitoring data includes the average current, average voltage, average temperature, and electric energy transmission amount of the transmission line; The historical meteorological department forecast data includes: the forecast data of the river basin where the hydropower station is located recorded N times before the current moment, and the forecast data includes the rainfall, rainfall intensity, and rainfall duration for the next day.

4. An intelligent decision-making system based on artificial intelligence, characterized in that: It is used to execute the intelligent decision-making method based on artificial intelligence as described in any one of claims 1-3.

Images