Trusted optimization algorithm for unit commitment based on blockchain

Abstract

This invention discloses a blockchain-based trusted optimization algorithm for unit commitment, relating to the energy blockchain field. It proposes a trusted solution method for unit commitment based on the concept of on-chain and off-chain collaboration. First, it proposes the concept of "on-chain segmentation and comparison optimization, and off-chain parallel fault-tolerant solution," establishing an on-chain and off-chain collaborative SCUC solution architecture. This overcomes the limitation of traditional blockchain technology in supporting SCUC model solution tasks. Subsequently, a fault-tolerant partitioning method for the unit commitment solution space is established to balance the reliability of the SCUC model solution results and the efficiency of the solution process in a blockchain environment. Finally, a corresponding SCUC trusted solution smart contract is designed. The proposed trusted solution method for the SCUC model has the advantages of verifiable optimality, Byzantine fault tolerance, and efficient solution.

Description

Technical Field

[0001] The present invention relates to the field of energy blockchain, and in particular to a blockchain-based unit combination trusted optimization algorithm. Background Art

[0002] Solving the security constrained unit commitment (SCUC) model, which determines the on / off status of generators, is a core component of spot market clearing. Ensuring the optimality and accuracy of SCUC solutions directly impacts the economic and security of spot market transactions.

[0003] SCUC is a large-scale mixed-integer linear programming (MILP) optimization problem. Existing research has largely focused on improving the efficiency and optimality of the centralized solution of the SCUC model. However, improving the verifiability of the SCUC model's solutions and satisfying market participants' trust in the optimality and fairness of the clearing results are equally pressing issues. The SCUC model exhibits high-dimensional discreteness, and in practice, only local optimal solutions are often found. Multiple local optimal solutions with similar objective values ​​often produce divergent unit startup and shutdown plans, and selecting different local optimal solutions can directly impact the profitability of some units. Under the centralized solution model, the SCUC model solution operates in a "black box" state, resulting in poor verifiability of the solution results and difficulty for market participants in verifying and assessing the quality of the local optimal solutions. Especially in the early stages of the development of intra- and inter-provincial electricity spot markets, spot market clearing results may deviate from the original quantity and price ranges, and unit startup and shutdown plans may deviate from expectations, potentially raising questions among some market participants about the clearing results.

[0004] Blockchain, a technology tool centered around multi-party co-governance and mutual oversight, has been widely discussed and applied in the energy sector due to its decentralized and tamper-proof nature. It possesses the inherent potential to support the trusted solution of the SCUC model. Within a blockchain environment, market operators, market regulators, and representatives of market participants can serve as consensus nodes to jointly solve and verify the SCUC model. Any attempt by any node to tamper with the solution will be detected and blocked by other nodes; only the results agreed upon on the chain can serve as the basis for market transactions. By storing rules on-chain and enabling multi-party oversight and enforcement, blockchain can break the "black box" nature of the SCUC model solution process, empowering market participants with oversight and verification rights and ensuring the fairness and transparency of transaction outcomes.

[0005] However, blockchain's multi-party synchronization and redundant computational characteristics often limit its ability to perform only simple numerical and logical calculations. Solving complex optimization tasks like the SCUC model would dramatically increase the computational complexity and runtime of traditional blockchains, making them unfeasible. Currently, scholars both domestically and internationally have proposed several blockchain solutions for complex optimization tasks. The mainstream approaches can be categorized into four main types: The first is "off-chain solution, on-chain evidence storage." This involves off-chain solution processing by a third-party, with only the results stored on-chain. This ensures consistency but struggles to ensure correctness. The second, centered on "model simplification," decomposes the model solution process into a series of simple computational steps, making it compatible with blockchain smart contract language environments. However, this approach is only applicable to scenarios where the optimization model can be easily simplified. The third, centered on "blockchain as coordination level," splits the optimization model into a high-level coordination problem and several locally solved lower-level sub-problems. The upper-level coordination problem is solved on-chain, while the lower-level sub-problems are solved locally by participating entities. However, this assumes that the original model is decomposable. The fourth category uses the Proof of Solution (PoSo) consensus mechanism as its core idea, adopting the "off-chain solution, on-chain verification" method. It performs complex optimization solution processes off-chain, sends the solution results on-chain, and each consensus node verifies the optimal solution on-chain. However, it is only applicable to scenarios with optimality criteria such as continuous optimization.

[0006] In summary, existing blockchain-based trusted optimization methods cannot support the solution of the SCUC model, a large-scale, discrete model. There is an urgent need to propose an efficient and trusted solution algorithm that is suitable for it. Summary of the Invention

[0007] The purpose of the present invention is to provide a blockchain-based unit combination trusted optimization algorithm to solve the problems raised in the above background technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] The blockchain-based unit commitment trusted optimization algorithm includes the following steps:

[0010] (1) Establishing the SCUC model and the on-chain-off-chain collaborative SCUC solution architecture, which includes an on-chain optimization layer and an off-chain solution layer. The on-chain optimization layer is composed of a group of blockchain consensus nodes, which is responsible for dividing the SCUC model solution space into several redundant sub-solution spaces, distributing them to the off-chain solution layer, and obtaining the optimal solution of the original model by comparing the correct solution results submitted by the off-chain solution layer; the off-chain solution layer is composed of a group of solution nodes, each of which searches for the local optimal solution of the original model in the corresponding sub-solution space distributed by the on-chain optimization layer, and submits it to the on-chain optimization layer;

[0011] (2) Establish a fault-tolerant partitioning method for the unit combination solution space to balance the reliability of the SCUC model solution results and the efficiency of the solution process in the blockchain environment;

[0012] (3) Design the corresponding SCUC trusted solution smart contract;

[0013] The steps of establishing the SCUC model in step (1) are as follows:

[0014] The SCUC model aims to minimize the total power generation cost, and its objective function is expressed as:

[0015]

[0016] Where T is the total number of time periods in the scheduling cycle; I is the total number of units that can be started and stopped; are the electricity cost, startup cost and shutdown cost of the i-th unit in time period t respectively;

[0017] Electricity cost

[0018]

[0019] Where K is the number of segmented quotations of unit i in time period t; and are the bid and winning bid of unit i in the kth segment within time period t; α i,t is the online status of unit i in time period t (1 means the unit is online, 0 means the unit is not online); is the minimum technical output cost of unit i in period t;

[0020] Startup costs

[0021]

[0022] Among them, β i,t Indicates whether unit i performs the startup action within time period t (1 indicates the startup action is performed, 0 indicates the startup action is not performed); is the startup cost of unit i;

[0023] Shutdown costs

[0024]

[0025] Among them, γ i,t Indicates whether unit i performs shutdown action in time period t (1 indicates shutdown action is performed, 0 indicates no shutdown action is performed); is the shutdown cost of unit i;

[0026] Constraints

[0027] Power balance constraints

[0028]

[0029] Among them, P i,t is the output of unit i in time period t; N is the number of nodes; D n,t is the load of node n in time period t; is the minimum technical output of unit i in time period t;

[0030] System spare capacity constraints

[0031]

[0032] in, is the maximum output of unit i in time period t; M is the total number of inter-provincial tie lines; is the transmission power of inter-provincial tie line m in time period t; η i is the unit confidence factor; P t up 、P t down are the system positive reserve constraint relaxation factor and the system negative reserve constraint relaxation factor for period t, respectively; are the system positive reserve demand and system negative reserve demand in period t respectively;

[0033] Constraints for winning bids for unit segments

[0034]

[0035] in, is the declared quantity of the kth segment quotation of unit i in time period t;

[0036] Unit ramp rate constraint

[0037]

[0038] in, and are the maximum ramp-up rate and maximum ramp-down rate of unit i in time period t, respectively;

[0039] Minimum continuous start and stop time constraints for units

[0040]

[0041] in, and are the continuous start-up and shutdown times of unit i in time period t, respectively; and are the minimum continuous start time and minimum continuous stop time of unit i respectively;

[0042] Energy Constraint

[0043]

[0044] Among them, E n,t is the net injected power of node n at time period t; n is the set of all units on node n; F l,t is the branch power flow of line l at time period t; SF n,l is the output power transfer distribution factor of node n to line l; The upper and lower limits of the transmission power of line l in time period t respectively;

[0045] The steps of establishing the fault-tolerant partitioning method of the unit combination solution space in step (2) are as follows: first, a 0-1 variables are selected in the unit combination model, and the complete solution space Ω is partitioned into X=2 a Segmentation Unit Then, the segmented units are reassembled into N sub-solution spaces Each sub-solution space contains [X(f+1) / N] partition units, which are distributed to the solving nodes for optimization. Each solving node performs optimization in parallel based on the divided sub-solution space and submits the solution results to the chain. The consensus node compares all the solution results through cross-communication verification and selects the solution that meets the constraints and minimizes the total power generation cost as the final optimal solution.

[0046] In the on-chain optimization layer, based on the classic branch-and-bound idea of ​​discrete optimization problems, the original solution space is divided into a set of segmentation units using the 0-1 variables in the unit commitment model. The segmentation units are then combined into N sub-solution spaces, which are distributed to the solving nodes for optimization, where N is the number of solving nodes. To ensure that the proposed method can still obtain the optimal solution even if some solving nodes tamper with the solution results or are offline, and to ensure optimization efficiency, the sub-solution space segmentation meets the following conditions:

[0047] Load balancing: The sub-solution spaces that each solving node is responsible for optimizing are of the same size, as shown in formula (12):

[0048]

[0049] Among them, J is the number of nodes to be solved, To solve the sub-solution space that node j is responsible for searching, |·| represents the number of split units contained in the solution space;

[0050] Byzantine fault tolerance: Assuming that there are no more than f cheating nodes among the solving nodes, which may disrupt the solving process by submitting incorrect solution results or not submitting solution results, then any f+1 solving nodes should restore the complete solution space, as shown in formula (13):

[0051]

[0052] Where, Ω is the solution space of the unit commitment model;

[0053] Minimum redundancy: The sub-solution space that each solution node is responsible for optimizing should minimize redundancy, as shown in formula (14):

[0054]

[0055] The step (3) of designing the corresponding SCUC trusted solution smart contract includes three stages: SCUC model preparation, fault-tolerant segmentation of solution space, and optimal solution verification and comparison, including the following functions:

[0056] 1) SCUC model preparation stage

[0057] ① Model data upload function: The market operation organization collects the relevant parameters and market boundary conditions of each generator set, establishes the SCUC model, converts the model into JSON format and uploads it to the blockchain, including the objective function and various constraints, as shown in Equations (1) to (11), for each solving node to access and download, ensuring the consistency of the subsequent solving process;

[0058] 2) Solution space segmentation stage

[0059] ②Solution space partition function: select a 0-1 variable in the SCUC model and partition the solution space Ω into X=2 a Segmentation Unit The segmentation units are then reassembled into N sub-solution spaces to form the allocation scheme M as shown in formula (15):

[0060]

[0061] The number of rows N of the matrix M is the number of sub-solution spaces, and the number of columns X is the number of segmentation units; the element M ij =1 represents the subsolution space Contains split units M ij =0 means subsolution space Does not contain split units

[0062] 3) Optimal solution verification and comparison stage

[0063] ③ Local optimal solution publishing function: Each solving node independently searches for the local optimal solution in the corresponding sub-solution space off-chain, and publishes the calculated local optimal solution L i Submit to blockchain;

[0064] ④ Constraint verification function: collect the local optimal solution L of each solution node iAfter that, verify whether each local optimal solution meets the constraint conditions such as formula (5)-formula (11), and remove the local optimal solution that does not meet the constraint conditions;

[0065] ⑤ Final optimal solution comparison function: compare all local optimal solutions that meet the constraints, and take the local optimal solution with the minimum total power generation cost as the final optimal solution G.

[0066] Compared with the prior art, the present invention has the following beneficial effects:

[0067] The present invention proposes the idea of ​​"on-chain segmentation and comparison optimization, and off-chain parallel fault-tolerant solution", establishes an on-chain and off-chain collaborative SCUC solution architecture, transfers the discrete optimization model solution process to the off-chain, and reduces the computational burden of the blockchain; the present invention considers Byzantine fault tolerance, load balancing and minimum redundancy, and establishes a fault-tolerant segmentation method for the unit combination solution space to take into account the reliability of the solution results and the efficiency of the solution process; the present invention designs a SCUC trusted solution smart contract and provides a practical SCUC trusted solution solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 Schematic diagram of the SCUC solution architecture based on on-chain and off-chain collaboration in the present invention.

[0069] Figure 2 This is a flowchart of the SCUC trusted solution based on smart contracts in the present invention.

[0070] Figure 3 This is a comparison chart of the unit start-stop status results of the centralized normal solution and the centralized cheating solution in the present invention.

[0071] Figure 4 This is a comparison chart of the solution results of different scenarios based on the SCUC trusted solution method in the present invention.

[0072] Figure 5 This is the unit start-stop status diagram for the algorithm solution and the centralized normal solution in the present invention. DETAILED DESCRIPTION

[0073] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0074] In an embodiment of the present invention, a blockchain-based unit commitment trusted optimization algorithm includes the following steps:

[0075] (1) Establish the SCUC model and the on-chain-off-chain collaborative SCUC solution architecture, which includes an on-chain optimization layer and an off-chain solution layer, such as Figure 1 As shown in the figure, the on-chain optimization layer consists of a group of blockchain consensus nodes, which is responsible for dividing the SCUC model solution space into several redundant sub-solution spaces, distributing them to the off-chain solution layer, and obtaining the optimal solution of the original model by comparing the correct solution results submitted by the off-chain solution layer; the off-chain solution layer consists of a group of solution nodes, each of which searches for the local optimal solution of the original model in the corresponding sub-solution space distributed by the on-chain optimization layer and submits it to the on-chain optimization layer;

[0076] (2) Establish a fault-tolerant partitioning method for the unit combination solution space to balance the reliability of the SCUC model solution results and the efficiency of the solution process in the blockchain environment;

[0077] (3) Design the corresponding SCUC trusted solution smart contract;

[0078] The steps of establishing the SCUC model in step (1) are as follows:

[0079] The SCUC model aims to minimize the total power generation cost, and its objective function is expressed as:

[0080]

[0081] Where T is the total number of time periods in the scheduling cycle; I is the total number of units that can be started and stopped; are the electricity cost, startup cost and shutdown cost of the i-th unit in time period t respectively;

[0082] Electricity cost

[0083]

[0084] Where K is the number of segmented quotations of unit i in time period t; and are the bid and winning bid of unit i in the kth segment within time period t; α i,t is the online status of unit i in time period t (1 means the unit is online, 0 means the unit is not online); is the minimum technical output cost of unit i in period t;

[0085] Startup costs

[0086]

[0087] Among them, β i,t Indicates whether unit i performs the startup action within time period t (1 indicates the startup action is performed, 0 indicates the startup action is not performed); is the startup cost of unit i;

[0088] Shutdown costs

[0089]

[0090] Among them, γ i,t Indicates whether unit i performs shutdown action in time period t (1 indicates shutdown action is performed, 0 indicates no shutdown action is performed); is the shutdown cost of unit i;

[0091] Constraints

[0092] Power balance constraints

[0093]

[0094] Among them, P i,t is the output of unit i in time period t; N is the number of nodes; D n,t is the load of node n in time period t; is the minimum technical output of unit i in time period t;

[0095] System spare capacity constraints

[0096]

[0097] in, is the maximum output of unit i in time period t; M is the total number of inter-provincial tie lines; is the transmission power of inter-provincial tie line m in time period t; η i is the unit confidence factor; P t up 、P t down are the system positive reserve constraint relaxation factor and the system negative reserve constraint relaxation factor for period t, respectively; are the system positive reserve demand and system negative reserve demand in period t respectively;

[0098] Constraints for winning bids for unit segments

[0099]

[0100] in, is the declared quantity of the kth segment quotation of unit i in time period t;

[0101] Unit ramp rate constraint

[0102]

[0103] in, and are the maximum ramp-up rate and maximum ramp-down rate of unit i in time period t, respectively;

[0104] Minimum continuous start and stop time constraints for units

[0105]

[0106] in, and are the continuous start-up and shutdown times of unit i in time period t, respectively; and are the minimum continuous start time and minimum continuous stop time of unit i respectively;

[0107] Energy Constraint

[0108]

[0109] Among them, E n,t is the net injected power of node n at time period t; n is the set of all units on node n; F l,t is the branch power flow of line l at time period t; SF n,l is the output power transfer distribution factor of node n to line l; The upper and lower limits of the transmission power of line l in time period t respectively;

[0110] The steps of establishing the fault-tolerant partitioning method of the unit combination solution space in step (2) are as follows: first, a 0-1 variables are selected in the unit combination model, and the complete solution space Ω is partitioned into X=2 a Segmentation Unit Then, the segmented units are reassembled into N sub-solution spaces Each sub-solution space contains [X(f+1) / N] partition units, which are distributed to the solving nodes for optimization. Each solving node performs optimization in parallel based on the divided sub-solution space and submits the solution results to the chain. The consensus node compares all the solution results through cross-communication verification and selects the solution that meets the constraints and minimizes the total power generation cost as the final optimal solution.

[0111] In the on-chain optimization layer, based on the classic branch-and-bound idea of ​​discrete optimization problems, the original solution space is divided into a set of segmentation units using the 0-1 variables in the unit commitment model. The segmentation units are then combined into N sub-solution spaces, which are distributed to the solving nodes for optimization, where N is the number of solving nodes. To ensure that the proposed method can still obtain the optimal solution even if some solving nodes tamper with the solution results or are offline, and to ensure optimization efficiency, the sub-solution space segmentation meets the following conditions:

[0112] Load balancing: The sub-solution spaces that each solving node is responsible for optimizing are of the same size, as shown in formula (12):

[0113]

[0114] Among them, J is the number of nodes to be solved, To solve the sub-solution space that node j is responsible for searching, |·| represents the number of split units contained in the solution space;

[0115] Byzantine fault tolerance: Assuming that there are no more than f cheating nodes among the solving nodes, which may disrupt the solving process by submitting incorrect solution results or not submitting solution results, then any f+1 solving nodes should restore the complete solution space, as shown in formula (13):

[0116]

[0117] Where, Ω is the solution space of the unit commitment model;

[0118] Minimum redundancy: The sub-solution space that each solution node is responsible for optimizing should minimize redundancy, as shown in formula (14):

[0119]

[0120] The step (3) of designing the corresponding SCUC trusted solution smart contract includes three stages: SCUC model preparation, fault-tolerant segmentation of solution space, and optimal solution verification and comparison, including the following functions:

[0121] 1) SCUC model preparation stage

[0122] ① Model data upload function: The market operation organization collects the relevant parameters and market boundary conditions of each generator set, establishes the SCUC model, converts the model into JSON format and uploads it to the blockchain, including the objective function and various constraints, as shown in Equations (1) to (11), for each solving node to access and download, ensuring the consistency of the subsequent solving process;

[0123] 2) Solution space segmentation stage

[0124] ②Solution space partition function: select a 0-1 variable in the SCUC model and partition the solution space Ω into X=2 a Segmentation Unit The segmentation units are then reassembled into N sub-solution spaces to form the allocation scheme M as shown in formula (15):

[0125]

[0126] The number of rows N of the matrix M is the number of sub-solution spaces, and the number of columns X is the number of segmentation units; the element M ij =1 represents the subsolution space Contains split units M ij =0 means subsolution space Does not contain split units

[0127] 3) Optimal solution verification and comparison stage

[0128] ③ Local optimal solution publishing function: Each solving node independently searches for the local optimal solution in the corresponding sub-solution space off-chain, and publishes the calculated local optimal solution L i Submit to blockchain;

[0129] ④ Constraint verification function: collect the local optimal solution L of each solution node i After that, verify whether each local optimal solution meets the constraint conditions such as formula (5)-formula (11), and remove the local optimal solution that does not meet the constraint conditions;

[0130] ⑤ Final optimal solution comparison function: compare all local optimal solutions that meet the constraints, and take the local optimal solution with the minimum total power generation cost as the final optimal solution G.

[0131] Case Analysis

[0132] 1Environmental testing and node selection

[0133] To validate the effectiveness of the proposed SCUC trusted solution method, we tested it using actual data from a provincial power grid (containing 756 nodes and 120 units). We built a Hyperledger Fabric blockchain environment within an Ubuntu virtual machine, and wrote the SCUC trusted solution smart contract in Go. Off-chain, we used GAMS software, calling CPLEX, as the solver. The computer configuration was Windows 10, a 13th-generation i5-13400F processor, an RTX 4060 GPU, an 8GB discrete graphics card, 16GB of RAM, and a 1TB SSD.

[0134] In this example, we set a = 3, N = 4, and f = 1, and divide the complete solution space into 8 partition units. As shown in Section 3.1, each sub-solution space should have no fewer than 4 partition units, and the allocation matrix is ​​shown in Equation (16).

[0135]

[0136] 2 Fault Tolerance Testing

[0137] To verify the fault tolerance of the proposed SCUC trustworthy solution method, we compared and analyzed scenarios involving centralized cheating and our method with varying numbers of cheating nodes. The cheating party manipulated the online status of some units to affect their revenue. For example, they could force some units to shut down during certain periods, excluding them from the power generation plan, thereby increasing the revenue of other units.

[0138] In the case of centralized cheating solution, it is assumed that units 100 and 101 are forced to shut down during the daytime, that is, α 100,t ,α 101,t =0(t∈[20,64]), the solution is shown in Table 1. The total power generation cost is RMB 195.46 million, which is about 0.2% higher than the normal solution.

[0139] Table 1 Comparison between centralized normal solution and centralized cheating solution

[0140] Concentrated normal solution Centralized cheating solution Total power generation cost (10,000 yuan) 19508 19546 Computation time (s) 40.04 41.13

[0141] Figure 3 The online status of all units in time period 2 is listed for both the centralized normal solution and the centralized cheating solution. A comparison shows that during the cheating solution, due to the forced shutdown of units 100 and 101, units that should have been shut down can be started up to increase output. This shows that while cheating in the centralized solution method has a minimal impact on total power generation costs, it still significantly changes unit output, affecting fair market competition and the rational allocation of resources.

[0142] In order to test the fault tolerance of the method proposed in this invention, 6 test scenarios are set. The cheating node numbers and cheating behaviors of each scenario are shown in Table 2 and Table 3 respectively. The solution results of each scenario are shown in Table 3. Figure 4 shown.

[0143] Table 2. Cheating solution node numbers in each scenario

[0144] Scene number Cheat Solver Node Scenario 1 Node 1 Scenario 2 Node 2 Scenario 3 Node 1, Node 2 Scene 4 Node 1, Node 4 Scene 5 Node 1, Node 2, Node 4 Scene 6 Node 1, Node 2, Node 3

[0145] Table 3 Node numbers and cheating behaviors

[0146] Solving node number Cheating Node 1 Set units 31, 37, and 45 to be built-in 0 during the period of 12:00-24:00 Node 2 Set units 108, 109, and 110 to be built-in between 10:00 and 16:00. Node 3 Set units 56 and 57 to be built-in between 11:00 and 24:00. Node 4 Set units 1, 34, and 35 to be built-in 0 during the period of 14:00-16:00

[0147] Depend on Figure 4 It can be seen that when only one solving node cheats, as shown in scenarios 1 and 2, the solution obtained by the proposed method is consistent with the normal solution result of the centralized solution because the other nodes have traversed the entire solution space. Therefore, the proposed method can meet the 1 / 3 fault tolerance.

[0148] When multiple solvers are cheating, if the subsolution spaces managed by the remaining solvers contain the partitioned unit where the global optimal solution resides (as in scenarios 3 and 6), the proposed method can still produce results consistent with the normal centralized solution. Otherwise (as in scenarios 4 and 5), the proposed method's results deviate from the normal centralized solution. Despite this, using the fault-tolerant solution space partitioning method, the solution remains close to the normal solution, with a deviation of less than 2%, demonstrating the high robustness of the proposed method.

[0149] Since the final results of scenarios 1, 2, 3 and 6 are consistent with the correct solution results, the present invention compares the start and stop states of each unit in scenario 4 with those in the centralized normal solution, such as Figure 5 As shown in the figure, in scenario 4, compared to the centralized cheating solution, only unit 57 switches from the on-state to the off-state, while the optimal solutions of the original model for the start-stop states of the remaining units remain consistent. This shows that this method can effectively reduce unit output variations in the presence of node cheating.

[0150] Computational efficiency test

[0151] To test the computational efficiency of the proposed method, this section compares the solution time of the proposed method with that of the centralized normal solution, as shown in Table 4. The solution time is the average of 10 experiments.

[0152] Table 4 Comparison of normal solution results of centralized normal solution and SCUC trusted solution method

[0153]

[0154] As shown in Table 4, each solving node searches for the optimal solution within its own sub-solution space, with a maximum solution time of 36.55 seconds, an 8.7% reduction compared to the centralized solution. This improvement in computational efficiency stems from the fact that the centralized solution requires searching for the optimal solution within the entire solution space. The proposed method partitions the solution space, allowing each solving node to search for the optimal solution only within a subset of the complete solution space, thus reducing solution time.

[0155] This paper addresses the verifiability and trustworthiness issues of traditional centralized solution methods for the SCUC model. It proposes a trusted solution method for unit commitment based on the concept of on-chain and off-chain collaboration. It builds a trusted SCUC solution architecture based on the concept of "on-chain segmentation and comparison optimization, and off-chain parallel fault-tolerant solution." It also proposes a fault-tolerant partitioning method for the unit commitment solution space and designs a trusted SCUC solution smart contract. Simulation results based on the Hyperledger Fabric blockchain platform demonstrate that this method offers the following advantages over traditional centralized solution methods:

[0156] 1) Verifiable optimality: The proposed method breaks the “black box” nature of the SCUC model solution process, enabling all participants to participate in and verify the model solution process and results, thus achieving transparency in the solution process and verifiability of the results.

[0157] 2) Byzantine fault tolerance: The proposed method can tolerate no more than 1 / 3 of the solving nodes cheating, and still ensure the accuracy of the final solution of the SCUC model under cheating;

[0158] 3) Solution efficiency: The proposed method partitions the SCUC model solution space so that each solution node only needs to search for the optimal solution within a subset of the complete solution space, reducing the solution time by 8.7% compared with centralized solution.

[0159] The proposed SCUC model trusted solution method solves the problem that traditional "blockchain +" trusted optimization methods are difficult to support the solution of discrete optimization models. In the future, we can further explore the application of the proposed method in other scenarios involving discrete optimization problems.

[0160] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

[0161] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. The blockchain-based unit commitment trusted optimization algorithm is characterized by: The steps include: (1) Establishing the SCUC model and the on-chain-off-chain collaborative SCUC solution architecture, which includes an on-chain optimization layer and an off-chain solution layer. The on-chain optimization layer is composed of a group of blockchain consensus nodes, which is responsible for dividing the SCUC model solution space into several redundant sub-solution spaces, distributing them to the off-chain solution layer, and obtaining the optimal solution of the original model by comparing the correct solution results submitted by the off-chain solution layer; the off-chain solution layer is composed of a group of solution nodes, each of which searches for the local optimal solution of the original model in the corresponding sub-solution space distributed by the on-chain optimization layer, and submits it to the on-chain optimization layer; (2) Establish a fault-tolerant partitioning method for the unit combination solution space to balance the reliability of the SCUC model solution results and the efficiency of the solution process in the blockchain environment; (3) Design the corresponding SCUC trusted solution smart contract; The steps of establishing the SCUC model in step (1) are as follows: The SCUC model aims to minimize the total power generation cost, and its objective function is expressed as: Where T is the total number of time periods in the scheduling cycle; I is the total number of units that can be started and stopped; are the electricity cost, startup cost and shutdown cost of the i-th unit in time period t respectively; Electricity cost Where K is the number of segmented quotations of unit i in time period t; and are the bid and winning bid of unit i in the kth segment within time period t; α i,t is the online status of unit i in time period t, 1 means the unit is online, 0 means the unit is not online; is the minimum technical output cost of unit i in period t; Startup costs Among them, β i,t Indicates whether unit i performs the startup action within time period t (1 indicates the startup action is performed, 0 indicates the startup action is not performed); is the startup cost of unit i; Shutdown costs Among them, γ i,t Indicates whether unit i performs shutdown action in time period t (1 indicates shutdown action is performed, 0 indicates no shutdown action is performed); is the shutdown cost of unit i; Constraints Power balance constraints Among them, P i,t is the output of unit i in time period t; N is the number of nodes; D n,t is the load of node n in time period t; is the minimum technical output of unit i in time period t; System spare capacity constraints in, is the maximum output of unit i in time period t; M is the total number of inter-provincial tie lines; is the transmission power of inter-provincial tie line m in time period t; η i is the unit confidence factor; P t up 、P t down are the system positive reserve constraint relaxation factor and the system negative reserve constraint relaxation factor for period t, respectively; are the system positive reserve demand and system negative reserve demand in period t respectively; Constraints for winning bids for unit segments in, is the declared quantity of the kth segment quotation of unit i in time period t; Unit ramp rate constraint in, and are the maximum ramp-up rate and maximum ramp-down rate of unit i in time period t, respectively; Minimum continuous start and stop time constraints for units in, and are the continuous start-up and shutdown times of unit i in time period t, respectively; and are the minimum continuous start time and minimum continuous stop time of unit i respectively; Energy Constraint Among them, E n,t is the net injected power of node n at time period t; n is the set of all units on node n; F l,t is the branch power flow of line l at time period t; SF n,l is the output power transfer distribution factor of node n to line l; The upper and lower limits of the transmission power of line l in time period t respectively; The steps of establishing the fault-tolerant partitioning method of the unit combination solution space in step (2) are as follows: first, a 0-1 variables are selected in the unit combination model, and the complete solution space Ω is partitioned into X=2 a Segmentation Unit Then, the segmented units are reassembled into N sub-solution spaces Each sub-solution space contains [X(f+1) / N] partition units, which are distributed to the solving nodes for optimization. Each solving node performs optimization in parallel based on the divided sub-solution space and submits the solution results to the chain. The consensus node compares all the solution results through cross-communication verification and selects the solution that meets the constraints and minimizes the total power generation cost as the final optimal solution. In the on-chain optimization layer, based on the classic branch-and-bound idea of ​​discrete optimization problems, the original solution space is divided into a set of segmentation units using the 0-1 variables in the unit commitment model. The segmentation units are then combined into N sub-solution spaces, which are distributed to the solving nodes for optimization, where N is the number of solving nodes. The sub-solution space segmentation satisfies the following conditions: Load balancing: The sub-solution spaces that each solving node is responsible for optimizing are of the same size, as shown in formula (12): Among them, J is the number of nodes to be solved, To solve the sub-solution space that node j is responsible for searching, |·| represents the number of split units contained in the solution space; Byzantine fault tolerance: Assuming that there are no more than f cheating nodes among the solving nodes, which disrupt the solving process by submitting incorrect solution results or not submitting solution results, then any f+1 solving nodes should restore the complete solution space, as shown in formula (13): Where, Ω is the solution space of the unit commitment model; Minimum redundancy: The sub-solution space that each solution node is responsible for optimizing should minimize redundancy, as shown in formula (14): The step (3) of designing the corresponding SCUC trusted solution smart contract includes three stages: SCUC model preparation, fault-tolerant segmentation of solution space, and optimal solution verification and comparison, including the following functions: 1) SCUC model preparation stage ① Model data upload function: The market operation organization collects the relevant parameters and market boundary conditions of each generator set, establishes the SCUC model, converts the model into JSON format and uploads it to the blockchain, including the objective function and various constraints, as shown in Equations (1) to (11), for each solving node to access and download, ensuring the consistency of the subsequent solving process; 2) Solution space segmentation stage ②Solution space partition function: select a 0-1 variable in the SCUC model and partition the solution space Ω into X=2 a Segmentation Unit The segmentation units are then reassembled into N sub-solution spaces to form the allocation scheme M as shown in formula (15): The number of rows N of the matrix M is the number of sub-solution spaces, and the number of columns X is the number of segmentation units; the element M ij =1 represents the subsolution space Contains split units M ij =0 means subsolution space Does not contain split units 3) Optimal solution verification and comparison stage ③ Local optimal solution publishing function: Each solving node independently searches for the local optimal solution in the corresponding sub-solution space off-chain, and publishes the calculated local optimal solution L i Submit to blockchain; ④ Constraint verification function: collect the local optimal solution L of each solution node i After that, verify whether each local optimal solution meets the constraint conditions such as formula (5)-formula (11), and remove the local optimal solution that does not meet the constraint conditions; ⑤ Final optimal solution comparison function: compare all local optimal solutions that meet the constraints, and take the local optimal solution with the minimum total power generation cost as the final optimal solution G.

Images