New energy power market pricing optimization method based on improved particle swarm optimization

Abstract

The invention provides a new energy power market pricing optimization method based on an improved particle swarm algorithm, and relates to the field of power market and intelligent optimization. An electric power spot market, a carbon emission permit market and an inertia auxiliary service market are uniformly incorporated into the same optimization search space, and an updatable coupling factor matrix is introduced to describe a cross elastic relationship among the three markets. A cloud-edge-end asynchronous parallel computing architecture is adopted, a cloud side generates a globally optimal solution, an edge side performs parallel iteration, and an end side optimizes and maps a result and returns operation data to compute fitness. Iterative evidence storage, asynchronous locking and automatic settlement are realized through a block chain smart contract, and an on-chain DAO mechanism is utilized to dynamically adjust and optimize the weight to form a closed-loop optimization process. According to the method, premature convergence can be effectively inhibited, and the clearing efficiency and the operation safety of the high-proportion new energy power market are improved.

Description

A New Energy Electricity Market Pricing Optimization Method Based on Improved Particle Swarm Optimization Algorithm Technical Field

[0001] This invention relates to the field of electricity markets and intelligent optimization, specifically to a pricing optimization method for new energy electricity markets based on an improved particle swarm optimization algorithm. Background Technology

[0002] With the advancement of the global "dual-carbon" strategic goals, the proportion of new energy power generation (such as wind power and photovoltaics) in the power system is constantly increasing. According to the latest statistics, the installed capacity of new energy in some regions has exceeded 50%, and this proportion is still increasing. However, the strong randomness and volatility of new energy, as well as its low inertia, pose a huge challenge to the stability of the electricity market. The traditional electricity market pricing mechanism, which is dominated by thermal power, has shown many shortcomings in the face of this change: the fluctuation range of spot electricity prices has increased dramatically, and the peak-valley difference has frequently exceeded historical extremes; the carbon market and the electricity market operate separately, and the cost of carbon emissions has not been transmitted to the electricity production and consumption end in a timely manner, resulting in the underutilization of carbon emission reduction potential; a large number of traditional synchronous units have been replaced, resulting in a significant decrease in system inertia, frequent frequency overruns, and a significant increase in the risk of safe and stable operation of the power system.

[0003] To address these issues, existing research has attempted to introduce Particle Swarm Optimization (PSO) algorithms into the field of electricity market pricing, leveraging swarm intelligence to achieve rapid optimization. However, most current research is limited to a single electricity price dimension, with optimization objectives focusing solely on generator revenue or electricity purchase costs, failing to comprehensively consider carbon emission costs, system inertia value, and the plug-and-play nature of distributed resources. Furthermore, the centralized iterative approach of traditional PSO algorithms leads to computational bottlenecks and communication delays, limiting their application in real-time electricity markets. Additionally, standard PSO algorithms suffer from premature convergence, making it difficult to meet the high-dimensional, multi-constraint, and multi-objective optimization requirements of complex renewable energy markets.

[0004] Under the existing electricity market mechanism, electricity prices, carbon prices, and inertia compensation prices are determined independently by the spot market, carbon emission market, and ancillary services market, respectively. The lack of an effective coupling mechanism leads to frequent arbitrage among renewable energy entities across multiple markets or their passive bearing of imbalanced funding burdens, resulting in market inefficiency. Therefore, a novel electricity market pricing optimization method is urgently needed that can simultaneously consider the three-dimensional coupling of electricity prices, carbon prices, and inertia compensation, and possess strong global search and local fine-tuning capabilities to achieve economical, low-carbon, and safe coordinated operation of a high-proportion renewable energy electricity market. Summary of the Invention

[0005] The technical problem to be solved by this invention is to provide a new energy power market pricing optimization method based on an improved particle swarm optimization algorithm. This method solves the problem of comprehensive optimization of high-proportion new energy access, carbon emission costs, and system inertia value, improves the clearing efficiency of the new energy power market, reduces carbon emissions, and enhances the safety and stability of the system.

[0006] The technical solution adopted in this invention is to provide a pricing optimization method for the new energy power market based on an improved particle swarm optimization algorithm. In a power market with a high proportion of new energy power generation, this method coordinates the optimization and clearing of electricity prices, carbon prices, and inertia compensation, reducing power volatility, improving power system stability, and effectively reducing carbon emissions. By combining the particle swarm optimization algorithm with a cloud-edge-device distributed computing architecture, the limitations of existing centralized optimization algorithms are overcome, significantly improving computational efficiency and real-time performance. The blockchain notarization mechanism of this invention ensures the transparency and auditability of the optimization results.

[0007] In a preferred embodiment, the present invention provides a three-dimensional coupled particle coding model that maps electricity price, carbon price, and inertia compensation price to the same particle search space. Optimization is performed through three-dimensional coupled particle coding. The electricity price vector, carbon price vector, and inertia compensation price vector are concatenated to form the particle position. The cross-elasticity relationship between electricity price, carbon price, and inertia is quantified through a coupling factor matrix, ensuring that the optimization process takes into account economy, low carbon emissions, and safety.

[0008] In a preferred embodiment, the present invention provides a cloud-edge-device asynchronous parallel computing framework. The cloud side generates a global optimal solution and performs global initialization. The edge side uses a hybrid update strategy of "quantum-chaos-leapfrog" to perform local search and quickly lock in a local optimal solution. The device side performs power and inertia response through distributed wind and solar power, energy storage and virtual synchronous machine, and transmits data back in real time to calculate fitness.

[0009] In a preferred embodiment, this invention further provides a dynamic weighting method for blockchain smart contracts and DAOs, introducing a fitness function based on returns, carbon emissions, and frequency deviation, with weights dynamically adjusted by a voting mechanism on the blockchain. By setting criteria such as the global optimal particle remaining unchanged through continuous iteration and the edge node locking depth reaching a certain threshold, convergence of the global optimization solution is achieved. Attached Figure Description

[0010] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 is an architecture diagram of a new energy power market pricing optimization method based on an improved particle swarm optimization algorithm according to the present invention; Figure 2 is a flowchart of a new energy power market pricing optimization method based on an improved particle swarm optimization algorithm according to the present invention. Detailed Implementation

[0011] To better understand the purpose, system architecture, and functional implementation of this embodiment, the embodiments and features described herein can be combined with each other without conflict. The exemplary embodiments disclosed herein will be described below with reference to the accompanying drawings, including specific technical details disclosed to aid understanding; however, these details should be considered exemplary rather than restrictive. Therefore, those skilled in the art should understand that various improvements and adjustments can be made to the embodiments described herein without departing from the scope and core ideas of the invention. Similarly, for clarity, detailed descriptions of well-known technologies, functions, and structures are omitted in the following description.

[0012] Example 1: Figure 1 is an architecture diagram of a new energy power market pricing optimization method based on an improved particle swarm optimization algorithm according to the present invention.

[0013] As shown in Figure 1, a new energy power market pricing optimization method 100 based on an improved particle swarm optimization algorithm consists of steps S110 to S150, including: in step S110, three-dimensional coupled particle encoding and initialization, splicing compensation price generation encoding, initializing velocity and coupling factor, and generating initial factors through dimensionality reduction and clustering; in step S120, cloud-side global iterative contract release, deployment of global smart contract, push of hash to edge nodes, and contract state variables; in step S130, edge-side asynchronous parallel update and locking, execution of hybrid velocity update, invocation of asynchronous locking contract, triggering on-chain game sub-contract re-evaluation of returns, and calculation of local optimum; in step S140, edge-side physical mapping and actual measurement feedback, measurement terminal data collection, and feedback of data to cloud-side fitness calculation; in step S150, collaborative convergence and on-chain settlement, and continuous iteration to complete joint clearing.

[0014] In an embodiment of the present invention, in step S110, a unified search space that can simultaneously reflect the value of electricity price, carbon price, and inertia is constructed, and a high-quality initial population is generated. First, prediction data for 24 time periods of the following day is acquired, including spot electricity price vector, carbon price vector, and inertia compensation price vector. These three vectors are vertically concatenated to form a particle position vector that represents a complete curve of electricity price, carbon price, and inertia price over 24 hours. To quantify the mutual influence between electricity price, carbon price, and inertia price, a 3×3 updatable coupling factor matrix is ​​introduced, derived through regression analysis of historical market data, and can be dynamically adjusted by the on-chain DAO in subsequent iterations. To avoid the algorithm getting trapped in local optima, an intelligent initialization method based on historical data is adopted. Cloud testing collects historical "electricity price-carbon price-inertia" data over a period of time, uses the t-SNE algorithm to reduce its dimensionality to a low-dimensional space, and then applies the K-means++ algorithm for clustering in the low-dimensional space to obtain multiple cluster centers. These cluster centers are then mapped back from the low-dimensional space to the original 72-dimensional space as the positions of the initial particles.

[0015] In an embodiment of the present invention, in step S120, key aspects of establishing a distributed computing framework and a trusted environment are achieved by utilizing blockchain technology to realize transparent management and synchronization of the global state. By deploying the GlobalIteration smart contract, the global optimal solution and its related information are maintained and managed. State variables are defined including gbestVector (a 72-dimensional position vector storing the current global optimal particle), gbestHash (a cryptographic hash value storing gbestVector for fast verification and comparison), gbestFit (a fitness value storing the corresponding global optimal particle), and iterationCounter (a record of the current iteration count). The initial global optimal particle generated in step S110, along with its hash value, fitness, and the initial state written into the contract, are initialized. After the contract deployment is complete, the current gbestHash and iterationCounter are pushed to all registered edge computing nodes to ensure that all nodes have the latest global optimal information.

[0016] In an embodiment of the present invention, in step S130, the algorithm is computed in parallel by edge nodes distributed throughout the power grid, significantly improving computational efficiency and scalability. Each edge node obtains the latest gbestHash and iterationCounter from the cloud or blockchain network, and maintains its own small particle swarm. These particles represent the optimization scheme for that region. An improved PSO velocity update formula is executed on the local particle swarm, incorporating three strategies: quantum potential well, chaotic reinitialization, and frog-jump local search. After completing one velocity update and position update, the fitness of the local particle swarm is evaluated, and the particle with the highest fitness is found. The hash value and fitness value of the locally computed particle are uploaded to the asynchronous locking contract on the blockchain. When the hash of the particle is uploaded, if it is found that the gbestHash in the cloud has been updated, an on-chain game sub-contract is triggered. This sub-contract automatically re-evaluates the potential gains of the particle in the new global environment to decide whether to adopt it.

[0017] In an embodiment of the present invention, in step S140, the virtual optimization world and the physical power system are connected to realize digital twin and closed-loop feedback. When the particle swarm of the edge node is adopted, its corresponding spot electricity price vector, carbon price vector and inertia compensation price vector curves are sent to the specific end-side distributed energy unit. The new energy unit performs charging and discharging scheduling according to the spot electricity price vector curve, and the virtual synchronous machine provides virtual inertia support according to the inertia compensation price vector curve. The smart meters, sensors and other terminal devices deployed on the edge side collect the measured data of the physical system operation in real time, including grid frequency deviation, actual carbon emissions and actual benefits of the participants. The collected measured data is transmitted back to the cloud-side master station through the 5G network. After receiving the measured data, the cloud-side master station calculates the fitness of the current iteration.

[0018] In an embodiment of the present invention, in step S150, collaborative convergence and on-chain settlement are completed iteratively to achieve joint clearing, at which point the optimization process ends, and final market settlement is initiated. The cloud-side main station continuously monitors the changes in gbestHash recorded on the blockchain and the lock depth of each side node. When gbestHash remains unchanged for 30 consecutive iterations and the lock depth of all side nodes is ≥90%, the optimization process is deemed to have converged. The final global optimal particle corresponds to the spot electricity price vector, carbon price vector, and inertia compensation price vector curves, which represent the joint clearing results of the spot market, carbon market, and inertia auxiliary service market for the next day. This triggers the settlement contract on the blockchain, automatically calculating the amounts due and payable for each party based on the final determined clearing price and the actual declaration / execution status of each participant, driving the flow of funds, and completing the entire value loop.

[0019] Figure 2 is a flowchart of a new energy power market pricing optimization method based on an improved particle swarm optimization algorithm according to the present invention.

[0020] As shown in Figure 2, a unified mathematical model that can simultaneously optimize electricity price, carbon price and inertia price is constructed. The formula for encoding particle position is shown in Equation (1) below.

[0021] (1) Among them, The 72-dimensional particle position vector represents the joint market decision variables. This represents the 24-hour spot electricity price vector. This represents the carbon price vector over a 24-hour period. The 24-hour inertia compensation price vector maps three independent market variables to a high-dimensional search space, achieving a dimensionality-reduced expression for joint optimization of electricity, carbon, and inertia, thus eliminating the market fragmentation problem.

[0022] To quantify the interaction between electricity price, carbon price and inertia price, a 3×3 updatable coupling factor matrix is ​​introduced, as shown in equation (2) below.

[0023] (2) Among them, This is the coupling factor matrix. This represents the change in carbon price per unit of electricity price increase. The sensitivity of system inertia demand to increases per unit of electricity price. The change in inertia compensation price caused by the increase in unit carbon price is used to quantify the inherent correlation of the ternary market, so that the particle search process naturally meets the economic-low-carbon-safety linkage constraint and avoids invalid solutions.

[0024] The initial particle positions are obtained through clustering and historical market data, as shown in the following formula.

[0025] (3) Among them, The initial position of the particle. For k-clustering methods, For the t-SNE algorithm, This is a collection of 72 dimensions of historical market data over 90 days. To determine the initial number of particles, we extract historical data features through dimensionality reduction clustering to generate a high-quality initial population, thereby improving the diversity index and accelerating convergence.

[0026] During the asynchronous update and locking process on the edge, a globally iterative smart contract is deployed on the blockchain. The state variables include: the 72-dimensional vector of the current globally optimal particle gbest, its SHA-256 hash, its fitness F(gbest), and the iteration counter. At the start of each iteration, the contract automatically pushes the gbest hash to each edge node. The edge node uses this hash to determine whether its local particle needs to be updated. After the local particle swarm accepts gbest, it uses the previous local best particle lbest as a benchmark to perform a mixed velocity update of quantum, chaotic and leapfrog. The quantum potential well update formula is shown in equation (4) below.

[0027] (4) Among them, For the first Individual particles The velocity vector at time t, For the first Individual particles The velocity vector at time t, For the first Particle position vectors and As a learning factor, and Uniformly distributed random numbers, For the first The local optimal position of each particle. The global optimal particle position. For quantum terms, Let be the dynamic quantum trap radius, as shown in equation (5).

[0028] (5) Among them, As the reference radius, As a benchmark carbon emission intensity, For the first Real-time carbon emission intensity of each edge node The maximum particle position vector, The minimum particle position vector, the higher the carbon emission intensity, the larger the search radius, which enhances the global exploration capability of highly polluted areas and avoids getting trapped in local low-carbon optima.

[0029] Chaotic reinitialization selects the 5% of particles with the worst fitness in each generation, generates new positions using Logistic mapping, and then maps them back to the feasible region, as shown in equation (6) below.

[0030] (6) Among them, The initial value is the value of the nth chaotic iteration. The 5% of particles with the worst adaptability are randomly selected as the initial value. By using the chaotic characteristics of the Logistic mapping, the particles can escape local optima and enhance population diversity.

[0031] The frog-jump local search finds the period with the largest inertia deficit and performs perturbation on the inertia compensation price of the period, as shown in equation (7).

[0032] (7) Among them, This is the index for the time period with the largest system inertia deficit. For time period The price of inertia compensation. The perturbation amplitude is uniformly distributed random quantity, and the compensation price is specifically corrected during the period when inertia is most scarce. The 2-opt perturbation improves the local fine search capability and increases the frequency qualification rate.

[0033] The asynchronous locking mechanism calculates and updates the local optimal hash H(lbest) and fitness F(lbest) of the lbest on the edge, and calls the contract interface asyncLock(bytes32 hash, int256 fit). If the cloud gbestHash matches the local cache, the contract records and returns success. If they do not match, an on-chain game subcontract is triggered to re-evaluate the payoff according to Stackelberg order and decide whether to retain or replace the lbest. The average locking success rate is 96%.

[0034] To determine whether the global optimal solution has changed due to the contributions of other edge nodes, the gbestHash received by the current edge node is compared with the one received previously. When gbest has been updated, the global environment changes, triggering the on-chain game subcontract to re-evaluate the current local optimal lbest, thus preventing the adoption of an "outdated" local optimal lbest. If gbest has not been updated, the local optimal lbest is locked directly, and the H(lbest) and F(lbest) of lbest are written to the blockchain as a valid result record of this round of iteration, ensuring the immutability and traceability of the result.

[0035] After the judgment is completed, the end-side mapping is performed. The energy storage charges and discharges according to the spot electricity price vector curve, with a power deviation of ≤±2% of the rated capacity. The virtual synchronous machine outputs virtual inertia according to the inertia compensation price vector curve, which is equivalent to replacing the inertia of the traditional synchronous machine. The wind and solar units actively curtail wind and solar power during periods of high carbon prices to reduce the carbon price vector curve. The terminal PMU and smart meter collect data such as frequency deviation, actual carbon emissions and revenue, package the data and upload it back to the cloud side through the 5G network. The cloud side then calculates the final fitness, as shown in the following formula.

[0036] (8) Among them, , and For weight parameters, For profit, For actual carbon emissions, To account for frequency deviation, the weight parameters are adjusted by a second vote on the on-chain DAO every 24 hours. When the vote exceeds the threshold of 60%, the weight parameters are further adjusted.

[0037] If the blockchain's gbestHash remains unchanged for 30 consecutive iterations and all edge lock depths are ≥90%, then global convergence is considered achieved. If convergence is not achieved, the entire process is rerun until convergence is reached. In on-chain computation, the GlobalIteration contract triggers settlement(), writing the final global particle gbest's corresponding spot electricity price vector curve, carbon price vector curve, and inertia compensation price vector curve into the block, and driving the electricity price, carbon price, and inertia compensation price settlement contracts to complete fund allocation.

[0038] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0039] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A pricing optimization method for the new energy electricity market based on an improved particle swarm optimization algorithm, characterized in that, Includes the following steps: S1. Construct a three-dimensional coupled particle code for electricity price, carbon price, and inertia compensation price; S2. Execute particle swarm optimization based on a cloud-edge-device asynchronous parallel architecture. The cloud side generates and publishes the global optimal solution, the edge side executes a hybrid local search including quantum, chaotic, and leapfrog strategies, and the device side executes power and inertia response and feeds back measured data. S3. Utilize blockchain smart contracts to realize particle state storage, asynchronous locking, and on-chain game coordination. S4. Determine convergence based on the consistency of the global optimal particle hash value recorded on the chain and the locking depth of edge nodes. After convergence, complete the joint clearing and settlement of electricity price, carbon price, and inertia compensation price through on-chain contracts.

2. The method according to claim 1, characterized in that, step In S1, the three-dimensional coupled particle encoding concatenates the spot electricity price vector, carbon price vector, and inertia compensation price vector for a preset time period to form a particle position vector, and introduces a coupling factor matrix to characterize the elastic correlation between the three.

3. The method according to claim 1, characterized in that, In step S2, the generation of the cloud-side global optimal solution includes: performing dimensionality reduction and clustering on historical market data to generate a diverse initial particle swarm, and publishing the hash value of the global optimal particle through a blockchain global iterative contract.

4. The method according to claim 1, characterized in that, In step S2, the edge-side hybrid local search includes quantum potential well perturbation, particle re-initialization based on chaotic mapping, and leapfrog local perturbation of the inertia price during the period of maximum inertia deficit.

5. The method according to claim 1, characterized in that, In step S2, the end-side execution and feedback includes: the distributed energy storage device charging and discharging according to the electricity price curve, the virtual synchronizing machine providing virtual inertia support according to the inertia price curve, and real-time collection of system frequency deviation, actual carbon emissions and market revenue data and transmitting them back to the cloud side.

6. The method according to claim 1, characterized in that, Step S3 also includes constructing a fitness function that integrates actual benefits, carbon emissions, and frequency deviation, with the weights of each objective dynamically adjusted through voting by a decentralized autonomous organization on the blockchain.

7. The method according to claim 1, characterized in that, In step S4, the convergence criterion is that the global optimal particle hash value remains unchanged for multiple consecutive iterations, and the particle locking depth of all edge nodes reaches a preset threshold.

8. The method according to claim 1, characterized in that, In step S4, the on-chain settlement includes: automatically performing joint clearing and fund allocation of electrical energy, carbon emission rights and inertia auxiliary services based on the electricity price, carbon price and inertia price curves corresponding to the converged global optimal particle.

9. The method according to claim 2, characterized in that, The elements in the coupling factor matrix are estimated based on historical data regression and can be updated during the runtime through an on-chain voting mechanism.

10. The method according to claim 5, characterized in that, The virtual inertia provided by the virtual synchronizer is calculated based on the inertia price curve, the rated capacity of the equipment, and the system reference frequency.

Images