Power grid dispatch decision system based on deep deterministic gradient strategy

Abstract

The present invention provides a power grid dispatching decision system based on a deep deterministic gradient strategy, which involves data processing technology and constructs a deep deterministic gradient model. The deep deterministic gradient model includes an input module, a calculation module and an output module; the input module obtains the operation information of the corresponding equipment in the power grid, and performs input analysis based on the operation information and equipment attributes to obtain the corresponding equipment operation events; the one-dimensional calculation subunit in the calculation module arranges and combines all the equipment operation events to obtain the corresponding initial event combination, and obtains a deterministic first sub-dispatching decision based on the initial event combination; the multi-dimensional calculation subunit in the calculation module is obtained, and the input first sub-dispatching decision is determined according to the structural relationship of the multi-dimensional calculation subunit, and the corresponding multi-dimensional dispatching sub-decision is obtained based on the first sub-dispatching decision; the output module outputs the determined power grid dispatching decision after verifying all the multi-dimensional dispatching sub-decisions and acts on the corresponding equipment.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to a power grid dispatching decision system based on a deep deterministic gradient strategy. Background Art

[0002] With the increasing demand for electricity and the diversification of energy structure in modern society, the operation and maintenance of power grids have become more and more complicated. The power grid includes various types of equipment, such as photovoltaic panels, batteries, wind turbines, etc. These devices are not only widely distributed, but also have complex interactive relationships between them. In order to ensure the stable operation of the power grid, it is necessary to monitor the operating status of the equipment in real time and make optimized dispatching decisions based on real-time data. This requires the power grid dispatching system to have efficient computing power and accurate decision-making capabilities to cope with the changing power grid operating environment.

[0003] At present, power grid dispatching systems usually rely on traditional rule bases and simple algorithms to dispatch equipment. However, these technologies are unable to cope with complex power grid operating environments. Traditional methods can only make simple judgments on the operating status of a single device and fail to fully consider the interaction between devices, which limits the accuracy and efficiency of dispatching decisions. For example, when photovoltaic panels cannot generate electricity normally, the system can only handle this problem alone and cannot fully consider its impact on battery storage and other equipment.

[0004] Therefore, how to automatically acquire and analyze the operating information of each device in the power grid, perform complex calculations and decision-making operations, and generate accurate scheduling decisions has become an urgent problem to be solved. Summary of the invention

[0005] The present invention provides a power grid dispatching decision system based on a deep deterministic gradient strategy, which can automatically acquire and analyze the operating information of each device in the power grid, perform complex calculations and decision-making operations, and generate accurate dispatching decisions.

[0006] A first aspect of an embodiment of the present invention provides a power grid dispatch decision system based on a deep deterministic gradient strategy, comprising:

[0007] Constructing a deep deterministic gradient model, wherein the deep deterministic gradient model includes an input module, a calculation module and an output module;

[0008] The input module obtains the operation information of the corresponding equipment in the power grid, and performs input analysis based on the operation information and equipment attributes to obtain the corresponding equipment operation events;

[0009] The one-dimensional calculation subunit in the calculation module arranges and combines all the equipment operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination;

[0010] Acquire the multidimensional computing subunits in the computing module, determine the input first sub-scheduling decision according to the structural relationship of the multidimensional computing subunits, and obtain the corresponding multidimensional scheduling sub-decision based on the first sub-scheduling decision;

[0011] After verifying all multi-dimensional dispatch sub-decisions, the output module outputs the determined power grid dispatch decision and acts on the corresponding equipment.

[0012] Optionally, in a possible implementation manner of the first aspect, the input module acquires operation information of a corresponding device in the power grid, and performs input analysis based on the operation information and device attributes to obtain a corresponding device operation event, including:

[0013] Extracting a corresponding operation-related information group based on the device attribute, wherein the operation-related information group includes multiple types of operation-related information, each device attribute has a preset related information group, and each related information group has preset operation-related information;

[0014] Matching the acquired current operation information according to the operation-related information group, and extracting the operation value of the matched operation-related information, wherein the operation information corresponds to the operation-related information;

[0015] Perform numerical analysis on the operating values ​​to obtain the corresponding equipment operating events.

[0016] Optionally, in a possible implementation manner of the first aspect, the performing numerical analysis on the operating value to obtain a corresponding device operating event includes:

[0017] Obtaining a matching algorithm for each operation-related information group, and inputting the operation value with the calculation tag in the operation-related information group into the matching algorithm to obtain a first calculation value;

[0018] Obtaining the operation value with the validation tag in the operation association information group as the second calculated value and comparing it with the first calculated value to obtain a value difference;

[0019] The corresponding device operation event is determined based on the numerical difference, and the numerical difference of each operation-related information group has a preset device operation event.

[0020] Optionally, in a possible implementation manner of the first aspect, the one-dimensional computing subunit in the computing module arranges and combines all device operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination, including:

[0021] The calculation module performs statistical processing on all first devices and second devices associated with the first device operation to obtain a corresponding device set;

[0022] The one-dimensional calculation subunit counts the equipment operation events in the equipment set, and performs permutations and combinations to obtain the corresponding initial event combination;

[0023] A deterministic first sub-scheduling decision is obtained based on the initial event combination, and the first sub-scheduling decision includes a corresponding scheduling target device.

[0024] Optionally, in a possible implementation of the first aspect, the one-dimensional calculation subunit counts device operation events in the device set, and performs permutations and combinations to obtain a corresponding initial event combination, including:

[0025] The computing module constructs corresponding one-dimensional computing subunits based on the total number of first devices, and allocates a device set of corresponding first devices to each one-dimensional computing subunit;

[0026] The one-dimensional computing subunit extracts the working relationship between the first device and the second device, and constructs a corresponding associated slot group based on the working relationship, where each working relationship corresponds to an associated slot group;

[0027] The device operation events corresponding to each first device and second device are traversed in sequence and filled into the associated slot groups to obtain initial event combinations corresponding to multiple associated slot groups.

[0028] Optionally, in a possible implementation manner of the first aspect, the determining first sub-scheduling decision is obtained based on the initial event combination, and the first sub-scheduling decision includes a corresponding scheduling target device, including:

[0029] Obtaining a corresponding first sub-scheduling decision based on the device running event of each slot in the associated slot group, and taking the first device corresponding to the associated slot group as the main scheduling target device;

[0030] Taking the first device and the second device in each associated group as targets, traversing the power grid and intercepting the power grid to obtain an intercepted network, and determining an impact path of the first sub-scheduling decision corresponding to the first device;

[0031] Determine a corresponding second device in the interception network based on the impact path, determine a first sub-scheduling impact of the second device based on the first sub-scheduling decision, and each second device has a preset first sub-scheduling impact with the first device;

[0032] If it is determined that the first sub-scheduling influence of the second device is not empty, the corresponding second device is used as the companion scheduling device.

[0033] Optionally, in a possible implementation manner of the first aspect, acquiring a multidimensional computing subunit in the computing module, and determining the input first sub-scheduling decision according to a structural relationship of the multidimensional computing subunit, and obtaining a corresponding multidimensional scheduling sub-decision based on the first sub-scheduling decision, includes:

[0034] The calculation module obtains the main scheduling target device and the accompanying scheduling device output by each one-dimensional calculation subunit;

[0035] If it is determined that a device is respectively used as a main scheduling target device and / or a companion scheduling device in the outputs of multiple one-dimensional computing sub-units, a multi-dimensional computing sub-unit corresponding to the corresponding device is generated, and the multi-dimensional computing sub-unit is connected to the one-dimensional computing sub-unit that outputs the main scheduling target device and / or the companion scheduling device;

[0036] The multi-dimensional computing sub-unit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the accompanying scheduling device respectively, and obtains the first sub-scheduling decision and the corresponding multi-dimensional scheduling sub-decision.

[0037] Optionally, in a possible implementation manner of the first aspect, the multidimensional computing subunit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device respectively to obtain the first sub-scheduling decision and obtain a corresponding multidimensional scheduling sub-decision, including:

[0038] If the multidimensional calculation subunit determines that the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device are non-mutually exclusive scheduling, the first sub-scheduling decision is obtained to obtain the corresponding multidimensional scheduling sub-decision;

[0039] If the multi-dimensional computing sub-unit determines that the first sub-scheduling decisions respectively corresponding to the main scheduling target device and the accompanying scheduling device are mutually exclusive scheduling, the one-dimensional computing sub-unit generating the mutually exclusive first sub-scheduling decision is selected as the computing sub-unit to be selected;

[0040] The to-be-selected computing subunit is connected to the multi-dimensional computing subunit, and automatic identification and output display interaction are performed to determine a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions.

[0041] Optionally, in a possible implementation manner of the first aspect, connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under a plurality of mutually exclusive first sub-scheduling decisions, includes:

[0042] Obtaining the main scheduling target device corresponding to the computing subunit to be selected, wherein the number of the computing subunits to be selected is at least 2;

[0043] In the decision-making process of obtaining the history, a historical combination network of the computing subunit to be selected and the multi-dimensional computing subunit appears. If there are multiple historical combination networks, the historical combination network with the closest state is determined based on the event state of the historical combination network;

[0044] The second sub-scheduling decision of the historical combination network is extracted as the second sub-scheduling decision in the current mutually exclusive multiple first sub-scheduling decision scenario.

[0045] Optionally, in a possible implementation manner of the first aspect, in the decision process of acquiring the history, a historical combination network of a computing subunit to be selected and a multidimensional computing subunit appears, and if there are multiple historical combination networks, determining a historical combination network with a closest state based on an event state of the historical combination network includes:

[0046] Acquire the event status of each historical combination network, wherein the event status at least includes a historical numerical difference;

[0047] Comparing the historical numerical value difference with the current numerical value difference to obtain a difference difference value, and adding the difference difference values ​​after weighted processing to obtain an event state similarity coefficient;

[0048] The historical combination network with the largest event state similarity coefficient is taken as the historical combination network closest to the state.

[0049] Optionally, in a possible implementation manner of the first aspect, connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under a plurality of mutually exclusive first sub-scheduling decisions, includes:

[0050] Extracting the historical combination network, and comparing each historical combination network with the current combination network, the current combination network being the current connection network between the to-be-selected computing subunit and the multi-dimensional computing subunit;

[0051] Corresponding information is added to each to-be-selected computing subunit based on the difference value, and the weight of each to-be-selected computing subunit is determined according to the mutually exclusive event, and the to-be-selected computing subunit corresponding to each mutually exclusive event has a preset weight.

[0052] According to a second aspect of the present invention, a storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the method of the first aspect of the present invention and various possible designs of the first aspect.

[0053] The beneficial effects of the present invention are as follows:

[0054] 1. This system can realize efficient and automatic grid dispatching, optimize grid operation, improve resource utilization, and ensure the stability and reliability of power supply. Among them, the grid dispatching decision system based on deep deterministic gradient strategy provided by the present invention realizes the optimization of the whole process from data acquisition, intelligent analysis to decision execution, which not only significantly improves the efficiency and accuracy of grid dispatching, but also provides a strong guarantee for the efficient and stable operation of the grid.

[0055] 2. The present invention can efficiently acquire and process the operation information of each device in the power grid by introducing an input module. The input module can not only collect basic operation data such as voltage, current, power, etc. in real time, but also extract relevant environmental information such as temperature and light intensity according to the device attributes. The input module automatically matches the current operation information with the preset associated information by extracting and matching the operation associated information group, thereby accurately extracting the operation value of the device. This not only improves the accuracy of the data, but also provides a high-quality data basis for subsequent calculations and decisions, significantly improving the real-time monitoring capability and data processing efficiency of the power grid dispatching system.

[0056] 3. The present invention not only takes into account the complex interaction between devices, but also can generate the optimal scheduling solution in a multi-dimensional complex environment, significantly improving the accuracy and adaptability of scheduling decisions. Among them, the calculation module adopts a deep deterministic gradient strategy, and realizes intelligent decision optimization and correlation analysis between devices through one-dimensional computing subunits and multi-dimensional computing subunits. The calculation module first arranges and combines the operating events of a single device through a one-dimensional computing subunit to generate an initial event combination, and then generates a first sub-scheduling decision based on the initial event combination. Subsequently, the multi-dimensional computing subunit fuses multiple first sub-scheduling decisions to generate a multi-dimensional scheduling sub-decision, realizing intelligent analysis, and significantly improving the accuracy and adaptability of scheduling decisions.

[0057] 4. The present invention effectively improves the overall operating efficiency and stability of the power grid, ensures the reliable operation of the power grid in a complex environment, and avoids global problems caused by single equipment failure. Among them, after verifying all multidimensional scheduling sub-decisions, the output module can output the final power grid scheduling decision and apply it to the corresponding equipment. By simulating the operating effect of the multidimensional scheduling sub-decision in the actual power grid, the feasibility and effectiveness of the decision are ensured. In addition, the output module can not only transmit the scheduling decision to the power grid control center in real time, but also command the actual equipment to perform corresponding operations, thereby realizing the automated scheduling management of power grid equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of the structure of a deep deterministic gradient model provided by the present invention.

[0059] Figure 2A schematic diagram of an interception network provided by the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0062] like Figure 1 The above is a schematic diagram of the structure of a power grid dispatching decision system based on a deep deterministic gradient strategy provided by the invention. The implementation method of the power grid dispatching decision system based on a deep deterministic gradient strategy includes:

[0063] S1, constructing a deep deterministic gradient model, wherein the deep deterministic gradient model includes an input module, a calculation module and an output module.

[0064] It can be understood that in order to automatically generate optimized dispatching decisions based on the operating information of power grid equipment to improve the efficiency and stability of power grid operation, a deep deterministic gradient model can be constructed so that the corresponding dispatching decisions can be accurately calculated.

[0065] Among them, the deep deterministic gradient model includes an input module, a calculation module and an output module. The input module is used to obtain and process the operating information of the power grid equipment, the calculation module is used to perform complex calculations and decision-making operations, and the output module is used to output and execute the final scheduling decision.

[0066] S2, the input module obtains the operation information of the corresponding equipment in the power grid, and performs input analysis based on the operation information and equipment attributes to obtain the corresponding equipment operation events.

[0067] It can be understood that the input module obtains the operating information of each device in the power grid, such as voltage, current, power and other parameters, and performs input analysis based on this operating information and device properties to obtain corresponding device operating events. Input analysis is the basis for subsequent calculations and decision-making, ensuring that the model has sufficient high-quality data for analysis.

[0068] Among them, the operation information is the information during the operation of the equipment, which can be voltage value, electric energy value, temperature value, etc. The equipment attributes are the attributes of the power grid equipment. For example, when the equipment is a photovoltaic panel, the corresponding attribute is the power generation attribute. When the equipment is a battery, the corresponding attribute is the power storage attribute. The equipment operation event is the operation event of the equipment. For example, the equipment operation event can be that the photovoltaic panel equipment is generating electricity.

[0069] In some embodiments, the specific implementation of step S2 (the input module obtains the operation information of the corresponding device in the power grid, and performs input analysis based on the operation information and device attributes to obtain the corresponding device operation event) includes:

[0070] S21, extracting a corresponding operation-related information group based on the device attribute, wherein the operation-related information group includes multiple types of operation-related information, each device attribute has a preset related information group, and each related information group has preset operation-related information.

[0071] It is understandable that in order to subsequently obtain the operating values ​​of the corresponding device, analyze the values, and obtain accurate device operating events, the corresponding operation-related information group can be determined based on the device attributes, so that the operation-related information in the operation-related information group can be quickly obtained subsequently.

[0072] It is not difficult to understand that the operation association information group includes multiple types of operation association information. At the same time, the device attribute of each device has a preset association information group, and each association information group has preset operation association information, so that the corresponding operation association information group can be determined according to the device attribute.

[0073] Among them, the operation-related information group is a combination of multiple related information of the equipment operation in the power grid. For example, it can be a combination of related information such as voltage, electric energy, temperature, etc. The operation-related information is information related to the equipment operation information. For example, the operation-related information of photovoltaic panels may include temperature, light intensity, electric energy, etc. The related information group is an information group pre-set according to the attributes of each device. For example, when in the associated information group solar power generation, the stronger the light in the operation-related information, the more corresponding electric energy.

[0074] S22, matching the acquired current operation information according to the operation-related information group, and extracting the operation value of the matched operation-related information, wherein the operation information corresponds to the operation-related information.

[0075] It can be understood that the current operation information is obtained and matched with the operation-related information according to the operation-related information group, so that the corresponding operation values ​​can be extracted later, which is convenient for subsequent analysis of the corresponding equipment operation events, thereby improving the accuracy of the generated scheduling decision.

[0076] The operation value is the value of operation-related information, such as wind speed value, power generation value, etc.

[0077] For example, light intensity is matched with solar power generation, and wind speed is matched with wind power generation, so as to achieve matching of operation-related information groups and facilitate subsequent event analysis.

[0078] It is not difficult to understand that different devices influence each other. The operation information is matched and combined with the operation-related information to extract the corresponding operation values, so as to analyze the accurate equipment operation events later.

[0079] S23, performing numerical analysis on the operating values ​​to obtain corresponding equipment operating events.

[0080] It is understandable that the extracted operating values ​​are analyzed to obtain corresponding equipment operating events.

[0081] For example, by analyzing the light intensity and output voltage of a photovoltaic panel, it can be determined that the photovoltaic panel device is generating electricity.

[0082] In some embodiments, the specific implementation of step S23 (the step of performing numerical analysis on the operating value to obtain the corresponding device operating event) includes:

[0083] S231, obtaining a matching algorithm for each operation-related information group, and inputting the operation value with a calculation tag in the operation-related information group into the matching algorithm to obtain a first calculation value.

[0084] It is understandable that since different operation-related information groups correspond to different operation information, such as multiple factors such as power and environmental information, a corresponding matching algorithm needs to be adopted to determine the first calculated value in order to subsequently determine the corresponding device operation event.

[0085] Among them, the matching algorithm is an algorithm for matching the corresponding operation-related information group. For example, when the operation-related information group is the operation information group corresponding to photovoltaic power generation, which includes light intensity, line length, etc., the operation values ​​that affect the calculation of power generation can be input into the matching algorithm, and the corresponding calculation formula can be called to obtain the corresponding power generation.

[0086] Furthermore, the calculation label is a label of operating information that affects the result. For example, in the solar power generation information group, since the light intensity will affect the power generation, the corresponding calculation label can be the light intensity, so as to retrieve the corresponding calculation formula and calculate the corresponding power generation information. The first calculated value is the calculated information value, such as power generation, power storage, etc.

[0087] S232, obtaining an operation value with a validation tag in the operation association information group as a second calculated value and comparing it with the first calculated value to obtain a value difference.

[0088] It can be understood that in order to determine accurate grid dispatching decisions, the actual operating value can be compared with the calculated normal first calculated value to obtain the corresponding numerical difference, and then the corresponding equipment operating status can be determined based on the numerical difference.

[0089] The second calculated value is the operation value with the verification tag in the operation-related information group, and the value difference is the difference between the second calculated value and the first calculated value.

[0090] S233, determining a corresponding device operation event based on the numerical difference, wherein the numerical difference of each operation-related information group has a preset device operation event.

[0091] It can be understood that the numerical differences between different operation-related information groups have preset device operation events, so that the device operation events that match the calculated numerical differences can be determined.

[0092] For example, when the difference between the second calculated value and the first calculated value is small, that is, the actual power generation is slightly different from the calculated theoretical power generation, it can be explained that the power generation equipment is in normal power loss, so that the relevant equipment does not need to be scheduled. When the difference is large, that is, the difference between the actual power generation and the theoretical power generation is large, it can be explained that there may be an interruption in the power generation line, and thus, a decision to inspect or repair the relevant equipment may be required. Therefore, the corresponding equipment operation event can be determined by the numerical difference, so as to facilitate the subsequent generation of the corresponding scheduling decision.

[0093] S3, the one-dimensional computing subunit in the computing module arranges and combines all the equipment operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination.

[0094] It is understandable that in order to obtain an accurate first sub-dispatching decision and realize efficient and automated power grid dispatching, it is necessary to obtain the equipment operation events of all devices in the power grid. For example, the corresponding equipment may be short of power or over-powered, the equipment may be working normally or abnormally, etc., so as to take corresponding strategies for different devices according to different events. Therefore, the one-dimensional calculation subunit will arrange and combine all equipment operation events to obtain the corresponding initial event combination.

[0095] Among them, the one-dimensional computing subunit is a computing unit that associates a single device and an event, the initial event combination is an event combination after associating and combining the equipment operation events, the first sub-dispatching decision is the dispatching decision of the power grid equipment corresponding to the initial event combination, and each initial event combination has a corresponding deterministic first sub-dispatching decision.

[0096] For example, photovoltaic panels are related to batteries. The electric energy generated by photovoltaic panels will have a certain impact on the power storage of batteries. Since the devices influence each other, the corresponding device operation events also influence each other. Therefore, the corresponding device operation events can be combined according to the photovoltaic panels and batteries to obtain a deterministic first sub-scheduling decision.

[0097] In some embodiments, the specific implementation of step S3 (the one-dimensional computing subunit in the computing module arranges and combines all device operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination) includes:

[0098] S31, the calculation module performs statistical processing on all first devices and second devices associated with the first devices to obtain a corresponding device set.

[0099] It is understandable that in order to perform correlation analysis on multiple devices so as to improve the accuracy of decision making, relevant devices may be statistically aggregated to facilitate subsequent rapid selection of corresponding devices for analysis.

[0100] The first device is the main device currently being analyzed, the second device is a device associated with the first device, and the device set is a set after all devices are classified.

[0101] For example, when performing the first sub-scheduling decision analysis on solar power generation, the corresponding first device is a photovoltaic panel, and the corresponding operation-related battery, light intensity monitor, etc. can be used as the second device, thereby combining the related devices.

[0102] S32, the one-dimensional calculation subunit counts the device operation events in the device set, and performs permutations and combinations to obtain a corresponding initial event combination.

[0103] It is understandable that the one-dimensional calculation subunit can be used to count the related device operation events in the device set, so that a plurality of device operation events can be arranged and combined to obtain a corresponding initial event combination.

[0104] In some embodiments, the specific implementation of step S32 (the one-dimensional calculation subunit counts the device operation events in the device set, and performs permutations and combinations to obtain the corresponding initial event combination) includes:

[0105] S321, the computing module constructs corresponding one-dimensional computing subunits based on the total number of first devices, and allocates a device set of corresponding first devices to each one-dimensional computing subunit.

[0106] It can be understood that in order to enable each first device to have a corresponding one-dimensional computing subunit, so as to facilitate the subsequent statistics and analysis of device operation events, the total number of first devices can be used to construct the same number of one-dimensional computing subunits as the total number. At the same time, each first device is used as the main body to assign a corresponding device set of the first device to each corresponding one-dimensional computing subunit.

[0107] Through the above implementation, each device has a corresponding one-dimensional computing subunit and a corresponding device set, so as to subsequently determine the corresponding working relationship, thereby generating an initial event combination, thereby improving the accuracy of subsequent decisions.

[0108] S322, the one-dimensional computing subunit extracts the working relationship between the first device and the second device, and constructs corresponding associated slot groups based on the working relationship, where each working relationship corresponds to an associated slot group.

[0109] It is understandable that clarifying the working relationship between different devices helps to build a more accurate initial event combination. By building associated slot groups, the working relationship between devices can be intuitively displayed so that the corresponding initial event combination can be obtained later, providing a basis for subsequent scheduling decisions.

[0110] The working relationship is the functional relationship between different devices, and the associated slot group is a combination of slots that demonstrate the working relationship.

[0111] For example, when the first device is a photovoltaic panel and the second device is a battery, since the photovoltaic panel can generate electricity and then transmit it to the battery for storage, the corresponding working relationship is energy generation-energy storage, so that a corresponding associated slot group can be constructed, and when the energy generation is 1000 watts and the energy storage is 10 watts, the corresponding operation is abnormal. When the energy generation is 1000 watts and the energy storage is 999 watts, of which 1 watt is normal loss, the relevant equipment is operating normally. Therefore, relevant scheduling decisions can be determined based on the corresponding working relationship to ensure the stability of the power grid operation.

[0112] S323, traversing the device running events corresponding to each first device and second device in turn and filling them into the associated slot groups, to obtain initial event combinations corresponding to multiple associated slot groups.

[0113] It can be understood that by traversing and filling the equipment operation events, a comprehensive initial event combination can be generated to provide sufficient information for subsequent decisions. At the same time, traversing and filling the operation events makes the initial event combination include all possible equipment states and working relationships, thereby improving the accuracy of subsequent decisions.

[0114] It is not difficult to understand that there can be multiple slots in the associated slot group, for example, the power relationship between two devices has one slot, the temperature relationship has one slot, etc., so that the relevant working relationship between the corresponding devices can be fully counted to improve the accuracy of decision-making.

[0115] S33: Obtain a deterministic first sub-scheduling decision based on the initial event combination, where the first sub-scheduling decision includes a corresponding scheduling target device.

[0116] It can be understood that when the first sub-dispatching decision is determined through the initial event combination, when the relevant equipment needs to be repaired or replaced in the first sub-dispatching decision, the corresponding decision information will also include the dispatching target equipment, so that the dispatching target equipment can be determined in time and the corresponding dispatch can be applied to the equipment to realize automatic dispatching of power grid equipment.

[0117] The dispatch target device is the power grid device that needs to be dispatched.

[0118] In some embodiments, the specific implementation of step S33 (obtaining a deterministic first sub-scheduling decision based on the initial event combination, wherein the first sub-scheduling decision includes a corresponding scheduling target device) includes:

[0119] S331, obtaining a corresponding first sub-scheduling decision based on the device running event of each slot in the associated slot group, and taking the first device corresponding to the associated slot group as the main scheduling target device.

[0120] It can be understood that the corresponding first sub-scheduling decision can be determined by associating the equipment operation events of each slot in the slot group. For example, when the power differences in the slots are large, the corresponding first sub-scheduling decision can be to detect the power storage lines to find out the reasons for the low power storage capacity.

[0121] Among them, the main scheduling target device is the first device corresponding to the associated slot group. For example, when the device operation event in the currently analyzed slot is an equipment operation event determined by the first device photovoltaic panel and the second device battery, the corresponding main scheduling target device is the photovoltaic panel. Similarly, when the first device is a battery and the second device is a generator, the corresponding main scheduling target device will be the battery, that is, the main scheduling target device corresponds to the first device applied to the one-dimensional computing sub-unit.

[0122] S332, traversing the power grid with the first device and the second device in each associated group as the target and intercepting the power grid to obtain an intercepted network, and determining the impact path of the first sub-scheduling decision corresponding to the first device.

[0123] It can be understood that by traversing the actual power grid network with the first device and the second device as the target, it is possible to intercept according to the positional relationship of the devices in the actual power grid network, and then obtain the intercepted network, and further determine the impact path of the first sub-scheduling decision corresponding to the first device in the power grid.

[0124] The intercepted network is a power grid network intercepted in an actual power grid according to the first device and the associated second device, and the impact path is a network path involved in the first sub-scheduling decision.

[0125] For example, Figure 2 As shown, when the first device is a photovoltaic panel and the second device has a battery and an electrical appliance (battery A, battery B and an electrical appliance), after the actual power grid is intercepted by the first device and the second device, the corresponding interception network can be determined, so as to obtain the impact path to facilitate the subsequent determination of the corresponding first sub-dispatching impact.

[0126] Through the above-mentioned implementation mode, the present invention can clearly understand the scheduling decision impact path of the first device by traversing the power grid, which is helpful to globally optimize the operating status of the power grid. At the same time, it can determine the overall impact of the scheduling decision on the power grid and ensure the comprehensiveness of the scheduling decision.

[0127] S333, determine the corresponding second device in the intercepted network based on the impact path, determine the first sub-scheduling impact of the second device based on the first sub-scheduling decision, and each second device has a preset first sub-scheduling impact with the first device.

[0128] It is understandable that since the devices in the power grid are interconnected, the scheduling of one device will affect another device. Therefore, it is necessary to determine the first sub-scheduling impact of the second device in the impact path to ensure that the scheduling impact of each device is taken into account, making the scheduling decision more targeted, thereby improving the integrity and accuracy of the decision.

[0129] The first sub-scheduling impact is the impact of the scheduling of the first device on the second device, which is preset.

[0130] For example, when maintenance on photovoltaic panels results in a reduction in the battery charge, this effect needs to be considered in the decision.

[0131] S334: If it is determined that the first sub-scheduling influence of the second device is not empty, the corresponding second device is used as a companion scheduling device.

[0132] It can be understood that when it is determined that the first sub-dispatching impact of the second device is not empty, it can be explained that the second device is affected by the dispatch of the first device, so that the corresponding second device can be used as a companion dispatching device, so that when the first device is subsequently dispatched, the corresponding companion dispatching device can also participate in the dispatching, ensuring that any affected equipment can be dispatched in time, avoiding chain reactions caused by dispatching decisions of a single device, improving the overall coordination and stability of power grid dispatching, and avoiding the occurrence of sudden problems.

[0133] The accompanying scheduling device is a device that also participates in the scheduling along with the scheduling of the first device, that is, the second device whose first sub-scheduling influence is not empty.

[0134] S4, obtaining the multidimensional computing sub-units in the computing module, and determining the input first sub-scheduling decision according to the structural relationship of the multidimensional computing sub-units, and obtaining the corresponding multidimensional scheduling sub-decision based on the first sub-scheduling decision.

[0135] It should be noted that since the same device may be affected by multiple devices, for example, both photovoltaic power generation equipment and wind power generation equipment will affect the battery's power storage. For example, when the photovoltaic panel cannot generate electricity normally, the corresponding battery may turn off the power storage function connected to the photovoltaic power generation, but when the wind power generation is normal, the battery can turn on the power storage function, and then the different first sub-dispatching decisions can be integrated to generate a multi-dimensional scheduling sub-decision, that is, the photovoltaic power storage branch is closed and the wind power storage branch is opened, thereby integrating multiple factors to realize the integrated dispatching of power grid equipment, so that it can perform more refined power grid dispatching, so as to improve the stability of power grid operation.

[0136] It can be understood that the multi-dimensional calculation sub-unit is a sub-unit that performs fusion calculation on multiple first sub-scheduling decisions, and the multi-dimensional sub-decision is a scheduling decision generated by fusing multi-dimensional factors.

[0137] It is not difficult to understand that the multi-dimensional computing subunit can handle more complex scheduling decision situations and generate more optimized scheduling solutions through multi-dimensional calculations.

[0138] In some embodiments, the specific implementation of step S4 (obtaining the multidimensional computing subunits in the computing module, determining the input first sub-scheduling decision according to the structural relationship of the multidimensional computing subunits, and obtaining the corresponding multidimensional scheduling sub-decision based on the first sub-scheduling decision) includes:

[0139] S41, the calculation module obtains the main scheduling target device and the accompanying scheduling device output by each one-dimensional calculation sub-unit.

[0140] It can be understood that when there is no abnormality in the running equipment event, the corresponding one-dimensional calculation unit will not output the corresponding main scheduling target device and the accompanying scheduling device. When the running equipment event has an abnormality, the corresponding one-dimensional calculation unit will output the corresponding main scheduling target device and the accompanying scheduling device. Therefore, the main scheduling target device and the accompanying scheduling device corresponding to the output of the one-dimensional calculation sub-unit are obtained through the calculation module, so as to facilitate the subsequent fusion analysis of related equipment, so as to make the decision of power grid scheduling more accurate and realize the refined power grid scheduling.

[0141] S42, if it is determined that a device serves as a main scheduling target device and / or a companion scheduling device in the output of multiple one-dimensional computing sub-units, a multi-dimensional computing sub-unit corresponding to the corresponding device is generated, and the multi-dimensional computing sub-unit is connected to the one-dimensional computing sub-unit that outputs the main scheduling target device and / or the companion scheduling device.

[0142] It can be understood that when a device is used as the main scheduling target device and the accompanying scheduling device in multiple different one-dimensional computing sub-units, that is, because the same device will be affected by multiple dispatches, the device will be ignored in multiple scenarios, which may reduce the adaptability of power grid dispatch. Therefore, a decision-making judgment will be made on the device. In order to make the dispatch decision generated by the device more accurate, a multi-dimensional computing sub-unit can be generated for the corresponding device, and the multi-dimensional computing sub-unit can be connected with the related one-dimensional computing sub-unit to generate a multi-dimensional scheduling sub-decision for the corresponding device, thereby improving the accuracy of equipment scheduling.

[0143] S43, the multidimensional computing subunit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device respectively, and obtains the corresponding multidimensional scheduling sub-decision of the first sub-scheduling decision.

[0144] It is understandable that for the same device, due to the different scenarios in which it is located, the scheduling method when the device is the main scheduling target device will conflict with the scheduling method when the device is the accompanying scheduling device. For example, for power generation equipment, the more electric energy is generated, the temperature of the corresponding equipment may increase accordingly. Therefore, when power generation equipment 1 is the main scheduling target device and needs to generate more electric energy, the corresponding temperature will increase. However, when power generation equipment 1 is the accompanying scheduling device and the corresponding main scheduling target device in its scenario is device 2, in order to improve the working efficiency of device 2, the temperature of the accompanying scheduling device, i.e., power generation equipment 1, needs to be reduced, and the power generation of the corresponding device needs to be reduced accordingly. Therefore, the first sub-scheduling decision of the same device in different scenarios is conflicting, that is, when the first sub-scheduling decision is to increase the power generation, it will be beneficial to one party and harmful to the other party. Therefore, a fusion analysis is needed to generate the optimal decision to make the power grid scheduling more stable.

[0145] It is worth mentioning that since the deterministic gradient strategy model is used in this scheme, its output is a deterministic output, and the result produced will be beneficial to one party and harmful to the other party. Therefore, the first sub-scheduling decision will be conflict-checked in order to generate the most adaptable deterministic decision.

[0146] In some embodiments, the specific implementation of step S43 (the multidimensional computing subunit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device, respectively, to obtain the first sub-scheduling decision and obtain the corresponding multidimensional scheduling sub-decision) includes:

[0147] S431, if the multidimensional computing subunit determines that the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device are non-mutually exclusive scheduling, then the first sub-scheduling decision is obtained to obtain the corresponding multidimensional scheduling sub-decision.

[0148] It can be understood that when the multidimensional computing sub-unit determines that the first sub-scheduling decision generated by the device under the main scheduling target device and the accompanying scheduling device will not have opposite effects on each other, the current first sub-scheduling decision can be used as the multi-dimensional scheduling sub-decision of the device, so as to facilitate subsequent application of the device to realize the scheduling of the power grid.

[0149] Among them, non-mutually exclusive scheduling is a scheduling decision in which the first sub-scheduling decision has no mutually exclusive impact.

[0150] S432, if the multi-dimensional computing sub-unit determines that the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device are mutually exclusive scheduling, then the one-dimensional computing sub-unit that generates the mutually exclusive first sub-scheduling decision is selected as the computing sub-unit to be selected.

[0151] It can be understood that when the multidimensional computing sub-unit determines that the first sub-dispatching decision generated by the device under the main scheduling target device and the accompanying scheduling device will have opposite effects on each other, the one-dimensional computing sub-unit corresponding to the mutually exclusive first sub-dispatching decision can be used as the computing sub-unit to be selected, so as to be subsequently connected to the multidimensional computing sub-unit, so that the multidimensional computing sub-unit receives the output decision of the computing sub-unit to be selected for fusion analysis, and generates the corresponding second sub-dispatching decision, so as to optimize the power grid scheduling.

[0152] The mutually exclusive scheduling is a scheduling decision in which the first sub-scheduling decision has a mutually exclusive impact, and the calculation subunit to be selected is a one-dimensional calculation subunit that generates the mutually exclusive first sub-scheduling decision.

[0153] S433, connecting the to-be-selected computing subunit with the multi-dimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions.

[0154] It can be understood that after the multidimensional computing subunit is connected to the computing subunit to be selected, it can receive the output result of the computing subunit to be selected for automatic identification, thereby determining a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions for output display.

[0155] The second sub-scheduling decision is an equipment scheduling decision determined by analyzing and integrating multiple mutually exclusive first sub-scheduling decisions.

[0156] In some embodiments, the specific implementation of step S433 (connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions) includes:

[0157] S4331, obtaining the main scheduling target device corresponding to the computing subunit to be selected, wherein there are at least two computing subunits to be selected.

[0158] It can be understood that since only the first sub-scheduling decisions corresponding to two or more computing sub-units to be selected will produce mutually exclusive results, at least two computing sub-units are selected, and the main scheduling target device corresponding to the computing sub-unit to be selected is obtained to subsequently generate the corresponding second sub-scheduling decision.

[0159] S4332, obtaining the historical combination network of the to-be-selected computing subunit and the multidimensional computing subunit that appear in the historical decision-making process, and if there are multiple historical combination networks, determining the historical combination network with the closest state based on the event states of the historical combination networks.

[0160] It can be understood that the historical combination network is a combination network of the to-be-selected computing subunits and the multi-dimensional computing subunits that appeared in the past time period.

[0161] It is not difficult to understand that based on the event status of these combined networks, the historical combined network closest to the current state is determined, where the event status may include historical numerical differences, operating parameters, environmental factors, load conditions, etc. of the power grid equipment. The historical combined network closest to the current state is determined so that the scheduling decisions of the power grid equipment can be effectively optimized later. This process can not only improve the accuracy of decision-making, but also use the experience in historical data to deal with current practical problems.

[0162] In some embodiments, the specific implementation of step S4332 (in the decision process of obtaining the history, a historical combination network of a computing subunit to be selected and a multidimensional computing subunit appears, and if there are multiple historical combination networks, the historical combination network with the closest state is determined based on the event state of the historical combination network) includes:

[0163] S43321, obtaining the event status of each historical combination network, wherein the event status at least includes a historical numerical difference.

[0164] It can be understood that by acquiring the time state of each historical combination network, a historical combination network closest to the current event state can be subsequently determined to obtain the corresponding second sub-scheduling decision.

[0165] Among them, the historical numerical difference is the numerical difference of the corresponding historical combination network.

[0166] S43322, comparing the historical numerical value difference with the current numerical value difference to obtain a difference value, performing weighted processing on the difference value and adding them together to obtain an event state similarity coefficient.

[0167] It can be understood that the difference difference value is the difference between the historical numerical difference and the current numerical difference, so that corresponding weights can be assigned to the importance of different parameters, and the event state similarity coefficient can be obtained by weighted summation of the difference difference values.

[0168] The event state similarity coefficient is a coefficient value for calculating the similarity of the time state, so as to determine the historical combination network that is closest to the current event state.

[0169] S43323, the historical combination network with the largest event state similarity coefficient is taken as the historical combination network closest to the state.

[0170] It can be understood that the larger the time state similarity coefficient is, the higher the similarity between the event state corresponding to the corresponding historical combination network and the current time state is, so the historical combination network with the largest event state similarity coefficient can be used as the historical combination network closest to the state.

[0171] S4333: extract the second sub-scheduling decision of the historical combination network as the second sub-scheduling decision in the current mutually exclusive multiple first sub-scheduling decision scenario.

[0172] It can be understood that by using the second sub-dispatching decision of the historical combined network as the second sub-dispatching decision in the current mutually exclusive multiple first sub-dispatching decision scenarios, the optimal second sub-dispatching decision for the corresponding equipment in multiple different first sub-dispatching decision scenarios can be determined in a timely manner, thereby facilitating the dispatching of power grid equipment.

[0173] In some embodiments, the specific implementation of step S433 (connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions) includes:

[0174] S4334, extracting the historical combination network, and comparing each historical combination network with the current combination network, the current combination network being the current connection network of the to-be-selected computing subunit and the multi-dimensional computing subunit.

[0175] It is understandable that the historical combined network is compared with the current combined network in order to effectively obtain potential useful information in the historical data and provide a reference for the current scheduling decision.

[0176] S4335, adding corresponding information to each to-be-selected computing subunit based on the difference value, and determining the weight of each to-be-selected computing subunit according to the mutually exclusive event, and the to-be-selected computing subunit corresponding to each mutually exclusive event has a preset weight.

[0177] It can be understood that adding the information of the difference difference value in the computing subunit to be selected helps to accurately evaluate the status and changes of the current network, determine the weight under mutually exclusive events, optimize the scheduling decision, make resource allocation more reasonable, and improve the efficiency and performance of the overall system, wherein the weight of the computing subunit to be selected is pre-set.

[0178] S5, the output module outputs the determined power grid dispatching decision after verifying all multi-dimensional dispatching sub-decisions and acts on the corresponding equipment.

[0179] It is understandable that the output module can verify all multi-dimensional sub-decisions, thereby outputting the final power grid dispatching decision and applying it to the corresponding equipment to facilitate cluster control of power grid equipment.

[0180] Among them, the output module is used to verify all multi-dimensional scheduling sub-decisions to ensure their feasibility and effectiveness. The verified scheduling decisions will be output and acted on the corresponding power grid equipment to achieve optimized scheduling of the power grid.

[0181] For example, the output module may simulate the operating effects of multi-dimensional scheduling sub-decisions in the actual power grid to verify their feasibility. If the decision passes the verification, the output module will transmit it to the power grid control center to instruct the actual equipment to perform the corresponding scheduling operations.

[0182] The present invention also provides a storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided by the various embodiments described above.

[0183] Among them, the storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, the storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the storage medium can also exist in a communication device as discrete components. The storage medium can be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0184] The present invention also provides a program product, which includes an execution instruction, which is stored in a storage medium. At least one processor of a device can read the execution instruction from the storage medium, and at least one processor executes the execution instruction so that the device implements the methods provided in the above various embodiments.

[0185] In the above-mentioned terminal or server embodiments, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0186] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power grid dispatching decision system based on deep deterministic gradient strategy, characterized in that: include: Constructing a deep deterministic gradient model, wherein the deep deterministic gradient model includes an input module, a calculation module and an output module; The input module obtains the operation information of the corresponding equipment in the power grid, and performs input analysis based on the operation information and equipment attributes to obtain the corresponding equipment operation events; The one-dimensional calculation subunit in the calculation module arranges and combines all the equipment operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination; Acquire the multidimensional computing subunits in the computing module, determine the input first sub-scheduling decision according to the structural relationship of the multidimensional computing subunits, and obtain the corresponding multidimensional scheduling sub-decision based on the first sub-scheduling decision; The output module outputs the determined power grid dispatch decision after verifying all multi-dimensional dispatch sub-decisions and acts on the corresponding equipment; The one-dimensional calculation subunit in the calculation module arranges and combines all the equipment operation events to obtain a corresponding initial event combination, and obtains a deterministic first sub-scheduling decision based on the initial event combination, including: The calculation module performs statistical processing on all first devices and second devices associated with the first device operation to obtain a corresponding device set; The one-dimensional calculation subunit counts the equipment operation events in the equipment set, and performs permutations and combinations to obtain the corresponding initial event combination; A deterministic first sub-scheduling decision is obtained based on the initial event combination, wherein the first sub-scheduling decision includes a corresponding scheduling target device; The one-dimensional calculation subunit counts the device operation events in the device set, and performs permutations and combinations to obtain corresponding initial event combinations, including: The computing module constructs corresponding one-dimensional computing subunits based on the total number of first devices, and allocates a device set of corresponding first devices to each one-dimensional computing subunit; The one-dimensional computing subunit extracts the working relationship between the first device and the second device, and constructs a corresponding associated slot group based on the working relationship, where each working relationship corresponds to an associated slot group; Traversing the device running events corresponding to each first device and second device in turn and filling them into the associated slot groups, to obtain initial event combinations corresponding to multiple associated slot groups; The deterministic first sub-scheduling decision is obtained based on the initial event combination, and the first sub-scheduling decision includes a corresponding scheduling target device, including: Obtaining a corresponding first sub-scheduling decision based on the device running event of each slot in the associated slot group, and taking the first device corresponding to the associated slot group as the main scheduling target device; Taking the first device and the second device in each associated group as targets, traversing the power grid and intercepting the power grid to obtain an intercepted network, and determining an impact path of the first sub-scheduling decision corresponding to the first device; Determine a corresponding second device in the interception network based on the impact path, determine a first sub-scheduling impact of the second device based on the first sub-scheduling decision, and each second device has a preset first sub-scheduling impact with the first device; If it is determined that the first sub-scheduling influence of the second device is not empty, the corresponding second device is used as a companion scheduling device; The multidimensional computing subunits in the computing module are obtained, and the input first sub-scheduling decision is determined according to the structural relationship of the multidimensional computing subunits, and the corresponding multidimensional scheduling sub-decision is obtained based on the first sub-scheduling decision, including: The calculation module obtains the main scheduling target device and the accompanying scheduling device output by each one-dimensional calculation subunit; If it is determined that a device is respectively used as a main scheduling target device and / or a companion scheduling device in the outputs of multiple one-dimensional computing sub-units, a multi-dimensional computing sub-unit corresponding to the corresponding device is generated, and the multi-dimensional computing sub-unit is connected to the one-dimensional computing sub-unit that outputs the main scheduling target device and / or the companion scheduling device; The multi-dimensional computing sub-unit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the accompanying scheduling device respectively, and obtains the first sub-scheduling decision and the corresponding multi-dimensional scheduling sub-decision.

2. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 1 is characterized in that: The input module obtains the operation information of the corresponding equipment in the power grid, and performs input analysis based on the operation information and equipment attributes to obtain corresponding equipment operation events, including: Extracting a corresponding operation-related information group based on the device attribute, wherein the operation-related information group includes multiple types of operation-related information, each device attribute has a preset related information group, and each related information group has preset operation-related information; Matching the acquired current operation information according to the operation-related information group, and extracting the operation value of the matched operation-related information, wherein the operation information corresponds to the operation-related information; Perform numerical analysis on the operating values ​​to obtain the corresponding equipment operating events.

3. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 2 is characterized in that: The performing numerical analysis on the operating values ​​to obtain corresponding equipment operating events includes: Obtaining a matching algorithm for each operation-related information group, and inputting the operation value with the calculation tag in the operation-related information group into the matching algorithm to obtain a first calculation value; Obtaining the operation value with the validation tag in the operation association information group as the second calculated value and comparing it with the first calculated value to obtain a value difference; The corresponding device operation event is determined based on the numerical difference, and the numerical difference of each operation-related information group has a preset device operation event.

4. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 1 is characterized in that: The multidimensional calculation subunit performs conflict verification on the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device, respectively, to obtain the first sub-scheduling decision and obtain the corresponding multidimensional scheduling sub-decision, including: If the multidimensional calculation subunit determines that the first sub-scheduling decisions corresponding to the main scheduling target device and the companion scheduling device are non-mutually exclusive scheduling, the first sub-scheduling decision is obtained to obtain the corresponding multidimensional scheduling sub-decision; If the multi-dimensional computing sub-unit determines that the first sub-scheduling decisions respectively corresponding to the main scheduling target device and the accompanying scheduling device are mutually exclusive scheduling, the one-dimensional computing sub-unit generating the mutually exclusive first sub-scheduling decision is selected as the computing sub-unit to be selected; The to-be-selected computing subunit is connected to the multi-dimensional computing subunit, and automatic identification and output display interaction are performed to determine a second sub-scheduling decision under multiple mutually exclusive first sub-scheduling decisions.

5. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 4 is characterized in that: The connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under a plurality of mutually exclusive first sub-scheduling decisions, includes: Obtaining the main scheduling target device corresponding to the computing subunit to be selected, wherein the number of the computing subunits to be selected is at least 2; In the decision-making process of obtaining the history, a historical combination network of the computing subunit to be selected and the multi-dimensional computing subunit appears. If there are multiple historical combination networks, the historical combination network with the closest state is determined based on the event state of the historical combination network; The second sub-scheduling decision of the historical combination network is extracted as the second sub-scheduling decision in the current mutually exclusive multiple first sub-scheduling decision scenario.

6. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 5 is characterized in that: In the decision-making process of acquiring the history, a historical combination network of a computing subunit to be selected and a multi-dimensional computing subunit appears. If there are multiple historical combination networks, the historical combination network with the closest state is determined based on the event state of the historical combination network, including: Acquire the event status of each historical combination network, wherein the event status at least includes a historical numerical difference; Comparing the historical numerical value difference with the current numerical value difference to obtain a difference difference value, and adding the difference difference values ​​after weighted processing to obtain an event state similarity coefficient; The historical combination network with the largest event state similarity coefficient is taken as the historical combination network closest to the state.

7. The power grid dispatching decision system based on deep deterministic gradient strategy according to claim 6 is characterized in that: The connecting the to-be-selected computing subunit with the multidimensional computing subunit, and performing automatic identification and output display interaction, and determining a second sub-scheduling decision under a plurality of mutually exclusive first sub-scheduling decisions, includes: Extracting the historical combination network, and comparing each historical combination network with the current combination network, the current combination network being the current connection network between the to-be-selected computing subunit and the multi-dimensional computing subunit; Corresponding information is added to each to-be-selected computing subunit based on the difference value, and the weight of each to-be-selected computing subunit is determined according to the mutually exclusive event, and the to-be-selected computing subunit corresponding to each mutually exclusive event has a preset weight.

Images