A power system scheduling decision support system based on the temporal difference algorithm

Abstract

A power system scheduling decision support system based on the temporal difference algorithm, including a parameter acquisition module, an optimization objective module, a resource aggregation module, an algorithm optimization module, a scheduling decision output module, a policy simulation module, and a decision deployment module; the power system scheduling decision support system based on the temporal difference algorithm proposed by the present invention aggregates power resources based on power system data, uses an improved algorithm to optimize the power system scheduling, outputs power scheduling decisions, and conducts decision simulation.

Description

Technical Field

[0001] The present invention relates to the field of power system decision-making, and particularly to a power system scheduling decision support system based on the temporal difference algorithm. Background Art

[0002] With the continuous development and complexity of power systems, traditional power system scheduling faces more and more challenges. Traditional methods mainly rely on static models and empirical rules, and cannot effectively adapt to the rapid changes in the power market, the increasing integration of renewable energy, and the dynamics of power loads. In addition, existing power system scheduling support systems often show certain limitations when dealing with large-scale systems and variable environments, and it is difficult to achieve efficient and real-time power system scheduling decisions. Existing scheduling systems mainly rely on rule-based static methods and cannot make full use of time series data for real-time scheduling decisions, resulting in the system being difficult to respond promptly to the rapidly changing power market and power loads. Therefore, in order to overcome the various limitations of traditional power system scheduling, this patent proposes a power system scheduling decision support system based on the temporal difference algorithm. This system combines the superior performance of the temporal difference algorithm and the requirements of power system real-time performance, aiming to make new breakthroughs in the field of power system scheduling. By making full use of time series data, optimizing parameter acquisition, and setting flexible optimization goals, this system will provide a more advanced and intelligent solution for power system scheduling, and promote the development of power systems towards a more sustainable and efficient direction. Summary of the Invention

[0003] The object of the present invention is to provide a power system scheduling decision support system based on the temporal difference algorithm, aiming to solve the problems mentioned in the above background.

[0004] To achieve the above object, the present invention proposes a power system scheduling decision support system based on the temporal difference algorithm, including a parameter acquisition module, an optimization goal module, a resource aggregation module, an algorithm optimization module, a scheduling decision output module, a strategy simulation module, and a decision deployment module; the parameter acquisition module collects grid operation parameters in the power system; the optimization goal module designs decision optimization goals according to the expectations of power system scheduling, and the resource aggregation module proposes a resource aggregation classification algorithm based on power resource constraints to aggregate and classify power resources; the algorithm optimization module proposes an optimization algorithm and optimizes the scheduling strategy according to the optimization goal; the scheduling decision output module outputs the final power system scheduling decision according to the output of the algorithm optimization module; the strategy simulation module conducts computer simulation on the power system scheduling decision; the decision deployment module arranges and deploys the power system scheduling decision passed by the simulation.

[0005] Further, the parameter acquisition module collects and obtains the operation parameters of the power grid system under various operating conditions, including power load, motor status, market price, environmental conditions, and electricity demand.

[0006] Further, the optimization target module designs decision optimization targets according to the power system scheduling expectations, considering efficiency, cost, system stability, resource utilization rate, and renewable energy ratio, and designs decision optimization targets.

[0007] Further, the resource aggregation module proposes a resource aggregation classification algorithm considering the constraints of the power resource cluster network and equipment conditions, and aggregates and classifies different types of power resources.

[0008] Further, the detailed process of the resource aggregation classification algorithm is as follows:

[0009] The power system resources include a variety of different devices, and there are differences in the operating parameters between different devices. To improve the efficiency of resource scheduling, first, multiple power device resources are aggregated into different power device clusters according to parameter changes, and the total power resources are defined Denoted as where P1, P2, P i , P n respectively represent the first power device, the second power device, the i-th power device, and the n-th power device in the power resources. The operating parameters of each power device are different, and their roles in power scheduling are different. For the power device P i , the operating parameter is represented as P′ i = [p {i,1} , p {i,2} , …, p {i,i} , …, p {i,n} , where p {i,1} , p {i,2} , p {i,i} , p {i,n} respectively represent the first operating parameter, the second operating parameter, the i-th operating parameter, and the n-th operating parameter. Based on the operating parameters of the power devices, the power devices are aggregated into different power clusters, and the constraint value vector η i of the power devices is defined. The time sequence difference of each operating parameter has a range, which is expressed as:

[0010]

[0011] where η′ i,1 , η′ i,2 , η′ i,j , η′ i,n respectively represent the power device P iThe first constraint value of the i-th operating parameter, the second constraint value of the i-th operating parameter, the j-th constraint value of the i-th operating parameter, the n-th constraint value of the i-th operating parameter, Δη′ i,1,t , Δη′ i,2,t , Δη′ i,j,t , Δη′ i,n,t respectively represent the time series difference of the first constraint value of the i-th operating parameter, the time series difference of the second constraint value of the i-th operating parameter, the time series difference of the j-th constraint value of the i-th operating parameter, the time series difference of the n-th constraint value of the i-th operating parameter, α i,1 , α i,2 , α i,j , α i,n respectively represent the lower limit of the cluster network constraint of the first constraint value of the i-th operating parameter, the lower limit of the cluster network constraint of the second constraint value of the i-th operating parameter, the lower limit of the cluster network constraint of the j-th constraint value of the i-th operating parameter, the lower limit of the cluster network constraint of the n-th constraint value of the i-th operating parameter, β i,1 , β i,2 , β i,j , β i,n respectively represent the upper limit of the cluster network constraint of the first constraint value of the i-th operating parameter, the upper limit of the cluster network constraint of the second constraint value of the i-th operating parameter, the upper limit of the cluster network constraint of the j-th constraint value of the i-th operating parameter, the upper limit of the cluster network constraint of the n-th constraint value of the i-th operating parameter. Based on the constraint value vectors of each power device, the power devices are combined and aggregated, and each power device is aggregated into different aggregated power clusters. The aggregated power clusters are classified, and the total category U is defined, expressed as U = [u1, u2, …, u k , …, u n , where u1, u2, u k , u n respectively represent the first category, the second category, the k-th category, the n-th category. There is a membership relationship between the constraint values of each operating parameter of each device and each category. For the category u k , there is an aggregation representation as follows:

[0012]

[0013] where F(·) is the aggregation representation, γ k,i,j represents the membership degree between the j-th data in the i-th power cluster and the category u k . The j-th data of the i-th aggregated power cluster is expressed as μ i,j , which consists of all constraint values in this aggregated power cluster. Define the measure relationship D(u k , μ i,j ), which represents the data μ i,j and the category uk The correlation relationship between them is calculated as follows:

[0014] D(u k , μ i,j ) = 1 - (F(u k ) - ∑p(η″ i,j |u k ))

[0015] where η″ i,j represents the j-th constraint value in the i-th aggregated power cluster, p(·) represents probability, the maximum measure between the data and all categories is the fuzzy measure of the data, and based on the fuzzy measure, the aggregated power clusters are classified fuzzily. The resource aggregation classification algorithm proposed by the present invention aggregates discrete power devices based on the constraint value vector of the power equipment, and then, based on the membership relationship between the constraint values of each operating parameter and each category, obtains the probability representation of each category. Through the measure relationship between the data and the categories, based on the fuzzy measure, the fuzzy classification of the power clusters is performed. The power equipment after the aggregation classification can improve the efficiency of the scheduling strategy during the power system scheduling.

[0016] Furthermore, the algorithm optimization module proposes an improved optimization algorithm and performs strategy optimization based on the optimization objective.

[0017] Furthermore, the detailed process of the improved optimization algorithm is as follows:

[0018] Based on all the classified aggregated power clusters, a power equipment state space is constructed, which is expressed as:

[0019]

[0020] where S1, S2, S t represent the state vectors in the 1st time period, the 2nd time period, and the t-th time period, and s 1,1 , s 1,2 , s 1,n , s 2,1 , s 2,n , s t,1 , s t,2 , s t,n represent the states of the 1st aggregated power cluster in the 1st time period, the 2nd aggregated power cluster in the 1st time period, the n-th aggregated power cluster in the 1st time period, the 1st aggregated power cluster in the 2nd time period, the n-th aggregated power cluster in the 2nd time period, the 1st aggregated power cluster in the t-th time period, the 2nd aggregated power cluster in the t-th time period, and the n-th aggregated power cluster in the t-th time period respectively. Based on the historical scheme, the initial state of the aggregated power cluster in the first time period is generated, and an objective function Y is designed, and the function is as follows:

[0021]

[0022] where represents the efficiency of the i-th aggregated power cluster at time period t, ρ t,i represents the revenue function of the i-th aggregated power cluster at time period t, θ t,i represents the operating loss of the i-th aggregated power cluster at time period t, ε t represents the total cost of the aggregated power cluster at time period t. To reduce the error caused by the different probabilities of various aggregated power cluster states, an important factor χ is defined, and the formula is as follows:

[0023]

[0024] where π(S t ) represents the probability density of the aggregated power cluster state S t , h(S t |Y) represents the probability density of the aggregated power cluster state S t under the condition of the objective function, E(·) represents the expectation. Considering the cumulative influence of historical states, an improved temporal difference algorithm is proposed, and the state update function is as follows:

[0025]

[0026] where V(·) represents the state value function, ζ represents the state step factor, ψ represents the data sampling rate, ξ i represents the i-th state discount factor, R i represents the reward, σ i represents the reward coefficient. Through state update, the strategy is continuously optimized, and the optimal strategy is output. The optimization algorithm proposed by the present invention constructs the state space of power equipment, designs the objective function based on the optimization goal, updates the state function through the improved temporal difference algorithm, and continuously iterates to optimize the strategy. The optimization algorithm proposed by the present invention can better match the power system scheduling, improving the optimization degree of the strategy.

[0027] Further, the scheduling decision output module outputs the final power system scheduling decision according to the optimization goal output by the algorithm optimization module.

[0028] Further, the strategy simulation module, based on the output power system scheduling decision, conducts power dispatching simulation through a computer to simulate the grid dispatching operation under the optimal decision, and judges the power system scheduling decision.

[0029] Further, the decision deployment module conducts actual scheduling deployment on the power system scheduling decision passed by the strategy simulation module.

[0030] Beneficial effects

[0031] The present invention provides a power system scheduling decision support system based on the temporal difference algorithm, which includes a parameter acquisition module, an optimization objective module, a resource aggregation module, an algorithm optimization module, a scheduling decision output module, a strategy simulation module, and a decision deployment module. The parameter acquisition module collects the grid operation parameters in the power system, including power load, motor status, market price, environmental conditions, and power consumption demand. The optimization objective module designs the decision optimization objective according to the power system scheduling expectation. The resource aggregation module proposes a resource aggregation classification algorithm based on the power resource constraints to aggregate and classify the power resources. The resource aggregation classification algorithm proposed by the present invention aggregates and matches the discrete power devices based on the constraint value vector of the power equipment, and then obtains the probability representation of each category based on the membership relationship between the constraint values of each operating parameter and each category. Through the measure relationship between the data and the categories, based on the fuzzy measure, the fuzzy classification of the power clusters is carried out. After the aggregation and classification of the power equipment, the efficiency of the scheduling strategy can be improved during the power system scheduling. The algorithm optimization module proposes an optimization algorithm to optimize the scheduling strategy according to the optimization objective. The optimization algorithm proposed by the present invention constructs the state space of the power equipment, designs the objective function based on the optimization objective, updates the state function through the improved temporal difference algorithm, and continuously iterates to optimize the strategy. The optimization algorithm proposed by the present invention can better match the power system scheduling and improve the optimization degree of the strategy. The scheduling decision output module outputs the final power system scheduling decision according to the algorithm optimization module. The strategy simulation module conducts computer simulation on the power system scheduling decision. The decision deployment module arranges and deploys the power system scheduling decision passed by the simulation. The power system scheduling decision support system based on the temporal difference algorithm proposed by the present invention is expected to significantly improve the overall efficiency and economy of the power system through more accurate real-time data and flexible optimization objective setting. The system can better cope with the fluctuations in the power market, reasonably allocate resources, reduce operating costs, and improve the competitiveness and sustainability of the power system. Combining the temporal difference algorithm with the actual needs of power system scheduling provides an innovative direction for the intelligent development of the power system. Through more intelligent data processing and scheduling decisions, the system will promote the power system to develop towards a more intelligent and adaptive direction to adapt to the new challenges of the future power industry. Brief description of the drawings

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0033] Figure 1 is a structural schematic diagram of the present invention;

[0034] Figure 2 is a schematic diagram of resource aggregation of the present invention. Detailed implementation manners

[0035] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present invention.

[0036] To achieve the above object, the present invention proposes a power system scheduling decision support system based on the temporal difference algorithm, including a parameter acquisition module, an optimization objective module, a resource aggregation module, an algorithm optimization module, a scheduling decision output module, a strategy simulation module, and a decision deployment module; the parameter acquisition module collects the grid operation parameters in the power system; the optimization objective module designs decision optimization objectives according to the power system scheduling expectations, and the resource aggregation module proposes a resource aggregation classification algorithm based on power resource constraints to aggregate and classify power resources; the algorithm optimization module proposes an optimization algorithm and optimizes the scheduling strategy according to the optimization objective; the scheduling decision output module outputs the final power system scheduling decision according to the algorithm optimization module; the strategy simulation module conducts computer simulation on the power system scheduling decision; the decision deployment module arranges and deploys the power system scheduling decision passed by the simulation.

[0037] Specifically, the parameter acquisition module collects and obtains the operation parameters of the grid system under various operating conditions, including power load, motor status, market price, environmental conditions, and electricity demand.

[0038] Specifically, the optimization objective module designs decision optimization objectives according to the power system scheduling expectations, considering efficiency, cost, system stability, resource utilization rate, and renewable energy ratio, and designs decision optimization objectives.

[0039] Specifically, the resource aggregation module proposes a resource aggregation classification algorithm considering the constraints of the power resource cluster network and equipment conditions, and aggregates and classifies different types of power resources.

[0040] Specifically, the detailed process of the resource aggregation classification algorithm is as follows: The power system resources include various different devices, and there are differences in the operating parameters between different devices. To improve the efficiency of resource scheduling, first aggregate multiple power device resources into different power device clusters according to parameter changes, and define the total power resources is expressed as Among them, P1, P2, P i , P n respectively represent the first power device, the second power device, the i-th power device, and the n-th power device in the power resources. The operating parameters of each power device are different, and the roles played in power dispatching are different. For the power device P i , the operating parameters are expressed as P′ i = [p {i,1} , p {i,2} , …, p {i,i} , …, p {i,n} , where p {i,1} , p {i,2} , p {i,i} , p {i,n} respectively represent the first operating parameter, the second operating parameter, the i-th operating parameter, and the n-th operating parameter. Based on the operating parameters of the power devices, the power devices are aggregated into different power clusters, and the constraint value vector η i of the power devices is defined. There is a range for the time series difference of each operating parameter, which is expressed as:

[0041]

[0042] Among them, η′ i,1 , η′ i,2 , η′ i,j , η′ i,n respectively represent the first constraint value of the i-th operating parameter, the second constraint value of the i-th operating parameter, the j-th constraint value of the i-th operating parameter, and the n-th constraint value of the i-th operating parameter of the power device P i . Δη′ i,1,t , Δη′ i,2,t , Δη′ i,j,t , Δη′ i,n,t respectively represent the time series difference of the first constraint value of the i-th operating parameter, the time series difference of the second constraint value of the i-th operating parameter, the time series difference of the j-th constraint value of the i-th operating parameter, and the time series difference of the n-th constraint value of the i-th operating parameter. α i,1 , α i,2 , α i,j , α i,n respectively represent the lower limit of the cluster network constraint of the first constraint value of the i-th operating parameter, the lower limit of the cluster network constraint of the second constraint value of the i-th operating parameter, the lower limit of the cluster network constraint of the j-th constraint value of the i-th operating parameter, and the lower limit of the cluster network constraint of the n-th constraint value of the i-th operating parameter. β i,1 , β i,2 , β i,j , β i,nrespectively represent the upper limit of the cluster network constraint for the first constraint value of the i-th operating parameter, the upper limit of the cluster network constraint for the second constraint value of the i-th operating parameter, the upper limit of the cluster network constraint for the j-th constraint value of the i-th operating parameter, and the upper limit of the cluster network constraint for the n-th constraint value of the i-th operating parameter. Based on the constraint value vectors of each power device, the power devices are grouped and aggregated, and each power device is aggregated into different aggregated power clusters. The aggregated power clusters are classified, and the total number of categories U is defined, expressed as U = [u1, u2, …, u k , …, u n , where u1, u2, u k , u n respectively represent the first category, the second category, the k-th category, and the n-th category. There is a membership relationship between the constraint values of each operating parameter of each device and each category. For the category u k , there is an aggregated representation as follows:

[0043]

[0044] where F(·) is the aggregated representation, γ k,i,j represents the membership degree between the j-th data in the i-th power cluster and the category u k . The j-th data of the i-th aggregated power cluster is expressed as μ i,j , which is composed of all the constraint values in this aggregated power cluster. The measure relationship D(u k , μ i,j ) is defined to represent the correlation relationship between the data μ i,j and the category u k . The calculation formula is as follows:

[0045] D(u k , μ i,j ) = 1 - (F(u k ) - ∑p(η″ i,j |u k ))

[0046] where η″ i,j represents the j-th constraint value in the i-th aggregated power cluster, p(·) represents probability, and the maximum measure between the data and all categories is the fuzzy measure of this data. Based on the fuzzy measure, the aggregated power clusters are fuzzily classified. The resource aggregation classification algorithm proposed by the present invention, based on the constraint value vectors of power devices, groups and aggregates discrete power devices, and then, based on the membership relationship between the constraint values of each operating parameter and each category, obtains the probability representation of each category. Through the measure relationship between the data and the category, based on the fuzzy measure, the fuzzy classification of power clusters is carried out. The power devices after aggregation classification can improve the efficiency of the dispatching strategy during the power system dispatching.

[0047] Specifically, the algorithm optimization module proposes an improved optimization algorithm and conducts policy optimization based on the optimization objective.

[0048] Specifically, the detailed process of the improved optimization algorithm is as follows:

[0049] Based on all classified aggregated power clusters, construct the power equipment state space, which is expressed as:

[0050]

[0051] where S1, S2, S t represent the state vectors in the first time period, the second time period, and the t-th time period, and s 1,1 , s 1,2 , s 1,n , s 2,1 , s 2,n , s t,1 , s t,2 , s t,n respectively represent the state of the first aggregated power cluster in the first time period, the state of the second aggregated power cluster in the first time period, the state of the n-th aggregated power cluster in the first time period, the state of the first aggregated power cluster in the second time period, the state of the n-th aggregated power cluster in the second time period, the state of the first aggregated power cluster in the t-th time period, the state of the second aggregated power cluster in the t-th time period, and the state of the n-th aggregated power cluster in the t-th time period. Based on the historical plan, generate the initial state of the aggregated power cluster in the first time period and design the objective function Y, and the function is as follows:

[0052]

[0053] where represents the efficiency of the i-th aggregated power cluster in the t-th time period, ρ t,i represents the revenue function of the i-th aggregated power cluster in the t-th time period, θ t,i represents the working loss of the i-th aggregated power cluster in the t-th time period, ε t represents the total cost of the aggregated power cluster in the t-th time period. To reduce the error caused by the different probabilities of various aggregated power cluster states, define the important factor χ, and the formula is as follows:

[0054]

[0055] where π(S t ) represents the probability density of the aggregated power cluster state S t , and h(S t |Y) represents the aggregated power cluster state S t under the condition of the objective function.The probability density, E(·) represents the expectation. Considering the cumulative impact of historical states, an improved temporal difference algorithm is proposed, and the state update function is as follows:

[0056]

[0057] Where V(·) represents the state value function, ζ represents the state step factor, ψ represents the data sampling rate, ξ i represents the i-th state discount factor, R i represents the reward, σ i represents the reward coefficient. Through state updates, the strategy is continuously optimized to output the optimal strategy. The optimization algorithm proposed in the present invention constructs the state space of power equipment. Based on the optimization objective, the objective function is designed, and the state function is updated through the improved temporal difference algorithm, and iteration is continuously carried out to optimize the strategy. The optimization algorithm proposed in the present invention can better match the power system scheduling, improving the degree of strategy optimization.

[0058] Specifically, the scheduling decision output module outputs the final power system scheduling decision according to the optimization objective output by the algorithm optimization module.

[0059] Specifically, the strategy simulation module, based on the output power system scheduling decision, conducts power dispatching simulation through a computer to simulate the grid dispatching operation under the optimal decision, and judges the power system scheduling decision.

[0060] Specifically, the decision deployment module performs actual scheduling deployment on the power system scheduling decision passed through the strategy simulation module.

Claims

1. A power system scheduling decision support system based on the temporal difference algorithm, including a parameter acquisition module, an optimization objective module, a resource aggregation module, an algorithm optimization module, a scheduling decision output module, a strategy simulation module, and a decision deployment module; the parameter acquisition module collects the grid operation parameters in the power system; the optimization objective module designs the decision optimization objective according to the power system scheduling expectation, the resource aggregation module proposes a resource aggregation classification algorithm based on the power resource constraint, and aggregates and classifies the power resources; the algorithm optimization module proposes an optimization algorithm and optimizes the scheduling strategy according to the optimization objective; the scheduling decision output module outputs the final power system scheduling decision according to the output of the algorithm optimization module; the strategy simulation module conducts computer simulation on the power system scheduling decision; the decision deployment module arranges and deploys the power system scheduling decision passed by the simulation. The resource aggregation module considers the power resource cluster network constraint and the equipment condition constraint, proposes a resource aggregation classification algorithm, and aggregates and classifies different types of power resources. The detailed process of the resource aggregation classification algorithm is as follows: Power system resources include a variety of different devices, and there are differences in the operating parameters between different devices. To improve the efficiency of resource scheduling, first, multiple power device resources are aggregated into different power device clusters according to parameter changes, and the total power resources are defined. The operating parameters of each power device are different, and their roles in power scheduling are also different. For power device Pi, the operating parameters are expressed as Pi′ = [p{i,1}, p{i,2}, …, p{i,i}, …, p{i,n}], where p{i,1}, p{i,2}, p{i,i}, and p{i,n} represent the first operating parameter, the second operating parameter, the i-th operating parameter, and the n-th operating parameter respectively. Based on the operating parameters of the power devices, the power devices are aggregated into different power clusters, and the constraint value vector ηi of the power devices is defined. There is a range for the time series difference of each operating parameter. Based on the parameter time series difference, the power devices are paired and aggregated, and each power device is aggregated into different aggregated power clusters. The aggregated power clusters are classified, and the total category U is defined. There is a membership relationship between the constraint values of each operating parameter of each device and each category, and the probability representation of the category is obtained. The j-th data of the i-th aggregated power cluster is expressed as μi,j, which is composed of all the constraint values in this aggregated power cluster. The measure relationship D(uk, μi,j) is defined, which represents the correlation relationship between the data μi,j and the category uk. The calculation formula is as follows: D(uk,μi,j)=1-(F(uk)-∑p(ηi″,j|uk)) Where ηi″,j represents the j-th constraint value in the i-th aggregated power cluster, p(·) represents probability, the maximum measure between the data and all categories is the fuzzy measure of the data, and based on the fuzzy measure, the aggregated power cluster is fuzzily classified.

2. The power system dispatching decision support system based on the temporal difference algorithm according to claim 1, characterized in that The parameter acquisition module collects and obtains the operation parameters of the power grid system under various operating conditions.

3. The power system scheduling decision support system based on the temporal difference algorithm according to claim 2, characterized in that, The optimization objective module designs the decision optimization objective according to the power system scheduling expectation, and considers efficiency, cost, system stability, resource utilization rate, and renewable energy ratio to design the decision optimization objective.

4. A power system scheduling decision support system based on the temporal difference algorithm according to claim 1, characterized in that, The algorithm optimization module proposes an improved optimization algorithm and conducts strategy optimization based on the optimization objective.

5. The power system scheduling decision support system based on the temporal difference algorithm according to claim 4, characterized in that, The detailed process of the improved optimization algorithm is as follows: Based on all the aggregated power clusters after classification, construct the power equipment state space, generate the initial aggregated power cluster state in the first time period based on the historical scheme, and design the objective function Y, and the function is as follows: Among them represents the efficiency of the i-th aggregated power cluster in the t-th time period, ρt,i represents the revenue function of the i-th aggregated power cluster in the t-th time period, θt,i represents the working loss of the i-th aggregated power cluster in the t-th time period, and εt represents the total cost of the aggregated power cluster in the t-th time period. To reduce the error caused by the probability of different states of various aggregated power clusters, an important factor χ is defined, and the formula is as follows: Where π(St) represents the probability density of the aggregated power cluster state St, h(St|Y) represents the probability density of the aggregated power cluster state St under the condition of the objective function, E(·) represents expectation, considering the cumulative influence of the historical state, propose an improved temporal difference algorithm, and the state update function is as follows: Where V(·) represents the state value function, ζ represents the state step factor, ψ represents the data sampling rate, ξi represents the i-th state discount factor, Ri represents the reward, σi represents the reward coefficient, and through state update, continuously optimize the strategy and output the optimal strategy.

6. The power system scheduling decision support system based on the temporal difference algorithm according to claim 1, characterized in that, The scheduling decision output module outputs the final power system scheduling decision according to the optimization objective output by the algorithm optimization module.

7. A power system scheduling decision support system based on the temporal difference algorithm according to claim 1, characterized in that, The strategy simulation module conducts power scheduling simulation through a computer based on the output power system scheduling decision, simulates the grid scheduling operation situation under the optimal decision, and judges the power system scheduling decision.

8. A power system scheduling decision support system based on the temporal difference algorithm according to claim 1, characterized in that, The decision-making and deployment module actually schedules and deploys the power system scheduling decisions passed through the policy simulation module.

Images