Intelligent lease management method and system based on shared charge pal

Abstract

The invention discloses an intelligent lease management method and system based on shared power banks, and relates to the technical field of intelligent scheduling of the shared power banks, and the method comprises the steps: collecting lease data of the shared power banks, obtaining a total lease demand time sequence through employing a VMD algorithm, generating a station position list through employing DBSCAN clustering based on the time sequence, and carrying out the lease management of the shared power banks. Obtaining a total capacity demand based on the time sequence; calculating the degradation cost of the power bank, obtaining the lease net income based on the degradation cost, obtaining a long-term income estimated value based on the lease net income by using a reinforcement learning algorithm, obtaining the real-time residual capacity based on the income estimated value and the total capacity demand, and generating a candidate scheduling scheme based on the station position list; and calculating the charging time required by the power bank, generating the candidate scheduling scheme fitness based on the charging time, and optimizing the candidate scheduling scheme based on the fitness. According to the method, the demand response precision, the resource configuration efficiency and the service credibility of the shared power bank system in a complex scene are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent scheduling technology for shared power banks, and in particular to an intelligent rental management method and system based on shared power banks. Background Technology

[0002] With the continuous deepening of the sharing economy model, shared power banks, as an important infrastructure for urban mobile consumer electronic devices, have been widely deployed in high-traffic scenarios such as shopping malls, transportation hubs, and restaurants and entertainment venues. The core of its operation model lies in achieving efficient allocation and dynamic scheduling of power bank resources in the spatial and temporal dimensions to meet users' immediate and fluctuating power needs. In recent years, the development of related technologies has mainly revolved around data-driven operation management. Typical methods include time series prediction models (such as ARIMA and LSTM) based on historical rental data for demand forecasting, combined with clustering algorithms (such as K-means) for hotspot area identification, and supplemented by static pricing or simple threshold-triggered scheduling strategies. Meanwhile, some systems have introduced IoT technologies (such as RFID and Wi-Fi communication modules) to achieve real-time monitoring of equipment status, initially establishing a closed-loop management system of "collection-analysis-scheduling". However, existing technologies still have significant limitations in dealing with highly nonlinear and multi-scale demand fluctuations in complex urban environments. First, traditional time series analysis methods are difficult to effectively separate stable baseline demand from sudden peak demand, leading to an over-reliance on historical peaks in capacity planning and resulting in the dual dilemma of equipment idleness or insufficient service capacity. Second, site layout often adopts fixed geographical division or simple density clustering, lacking analysis of demand structure (stable / fluctuating). The lack of detailed consideration of components and insufficient integration of the impact of external environmental factors (such as weather and holidays) on demand leads to low site coverage efficiency. Furthermore, in terms of revenue management, existing systems generally neglect the quantitative modeling of the physical degradation cost of power banks, charging only based on rental duration and failing to incorporate battery health degradation into dynamic pricing and equipment scheduling decisions, thus affecting long-term asset utilization. Finally, the generation of scheduling schemes relies heavily on rule engines or static optimization models, lacking the ability to collaboratively optimize multiple constraints such as time of day (TOU), transportation costs, and the authenticity of equipment locations, resulting in poor scheduling execution feasibility and difficulty in achieving the goal of maximizing revenue. Summary of the Invention

[0003] In view of the aforementioned existing problems, the present invention is proposed.

[0004] Therefore, this invention provides an intelligent rental management method and system based on shared power banks, which solves the problems of inaccurate demand decomposition, static site planning, and disconnect between scheduling and pricing in the prior art.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides an intelligent rental management method based on shared power banks, which includes: collecting rental data of shared power banks, using the VMD algorithm to obtain a time series of total rental demand, using DBSCAN clustering based on the time series to generate a list of station locations, and obtaining the total capacity demand based on the time series. The degradation cost of the power bank is calculated, the net rental income is obtained based on the degradation cost, a long-term revenue forecast is obtained based on the net rental income using a reinforcement learning algorithm, the real-time remaining capacity is obtained based on the revenue forecast and the total capacity demand, and a candidate scheduling scheme is generated based on the site location list. Calculate the charging time required for the power bank, generate the fitness of candidate scheduling schemes based on the charging time, optimize the candidate scheduling schemes based on the fitness, verify the optimized candidate scheduling schemes using RFID based on the remaining capacity, and generate the rental price based on the verified candidate scheduling schemes. The rental price is scheduled and fine-tuned in real time through a distributed algorithm to obtain real-time scheduling results. Rental decisions are made based on the real-time scheduling results, and rental records are generated. Settlement records and updated degradation costs are generated based on the rental records.

[0006] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the following steps are taken: rental data of shared power banks is collected; the total rental demand time series is obtained using the VMD algorithm; a site location list is generated using DBSCAN clustering based on the time series; and the total capacity demand index is obtained based on the time series. Historical rental data is extracted from a MySQL database; environmental data is obtained via API; RFID devices record real-time rental events; and the total rental demand time series is generated using the VMD algorithm. Including stable demand and fluctuating demand ; Based on RFID-recorded rental events, DBSCAN clustering is used to identify high-density rental areas as candidate sites, and VMD decomposition results are combined for optimization to generate a list of site locations. The total site capacity requirement is then obtained based on a capacity estimation method using historical peak statistics. .

[0007] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the following steps are included: calculating the degradation cost of the power bank, obtaining the net rental revenue based on the degradation cost, using a reinforcement learning algorithm to obtain a long-term revenue forecast based on the net rental revenue, obtaining the real-time remaining capacity based on the revenue forecast and total capacity demand, and generating candidate scheduling schemes based on the site location list. This involves scanning all power banks hourly using the Wi-Fi and communication modules integrated within the power bank to obtain the initial SoC and cycle count, and calculating the degradation cost of each power bank. and net income from actual leasing And based on net income The current state is calculated using a reinforcement learning algorithm. Next action Long-term earnings forecast Based on the output of the reinforcement learning algorithm, the number of devices is allocated according to the stable demand b(t), and the allocation for each site is accumulated to obtain the result. ,Will Divide by average turnover rate To obtain the number of devices already in use Calculate the real-time remaining capacity Based on the generated list of site locations, an initial candidate scheduling scheme is obtained.

[0008] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the steps include: calculating the charging time required for the power bank; generating a candidate scheduling scheme fitness based on the charging time; optimizing the candidate scheduling scheme based on the fitness; verifying the optimized candidate scheduling scheme using RFID based on the remaining capacity; generating a rental price based on the verified candidate scheduling scheme; dividing the day into off-peak, off-peak, and peak electricity price phases; and using an initial candidate scheduling scheme and degradation costs. The charging time of device i at time t is obtained. Based on TOU decision-making, the decision variables are obtained. Integrate degradation costs Prioritize low-degradation equipment, provide electricity price and time constraints based on TOU, and calculate the fitness of candidate scheduling schemes. ; Based on fitness A genetic algorithm is used to optimize the scheduling scheme, and RFID is used to verify the feasibility of the optimized candidate scheduling schemes. For the candidate scheduling schemes that pass the RFID verification, the rental price is adjusted based on real-time rental prices to obtain the real-time optimized rental price. .

[0009] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the real-time scheduling and price fine-tuning based on rental prices using a distributed algorithm, obtaining real-time scheduling results, refers to calculating the expected demand at each site, optimizing the allocation of equipment at sites through a distributed algorithm, and basing the optimized rental prices on... Based on the projected demand at each site, rental prices will be dynamically adjusted. If price fluctuations exceed a preset threshold... If the adjustment step size is reduced, the iterative average is taken to stabilize the price, and the optimized scheduling results and device status are recorded to the blockchain through smart contracts to ensure transparency and immutability.

[0010] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the following steps are included: making rental decisions based on real-time scheduling results and generating rental records; obtaining real-time scheduling results and blockchain records; the user scanning the RFID tag on the device; using the shared power bank intelligent rental management platform to read the device ID and verify whether it matches the site allocation in the optimized candidate scheduling scheme; generating a decentralized identity using federated learning to confirm the user's identity legitimacy; the user selecting the rental type on the mobile terminal; long-term contracts based on a fixed price; and real-time rentals combined with the optimized price from the real-time scheduling results. The rental decision-making process utilizes zero-knowledge proof technology. Users prove the validity of their rental requests, and the user queries the device status via blockchain. The shared power bank smart rental management platform verifies the remaining capacity of the device. To determine whether the rental requirements are met, the smart contract records the rental start time, device ID, battery level, rental type, and price, generating a rental record.

[0011] As a preferred embodiment of the intelligent rental management method based on shared power banks described in this invention, the step of generating settlement records and updated degradation costs based on rental records refers to obtaining rental records and user interaction data, acquiring the device status at the time of return through a Wi-Fi module, storing it in a MySQL database, and when a user returns the power bank, the Wi-Fi module scans the device ID to verify whether it matches the rental record, compares the battery level at the time of return with the initial battery level in the rental record, calculates the actual depth of discharge (DoD), calculates the rental fee according to the rental type, deducts the fee through the user account associated with RFID, and the smart contract records the settlement details and stores them in the blockchain.

[0012] Secondly, the present invention provides an intelligent rental management system based on shared power banks, comprising: The data acquisition and demand forecasting module is used to integrate historical and real-time data, use VMD and DBSCAN to generate a site list and estimate capacity demand; The cost accounting and revenue forecasting module is used to calculate net income and long-term revenue forecasts based on degradation costs and Q-learning. The scheduling scheme generation and optimization module is used to combine TOU electricity price and genetic algorithm to optimize equipment allocation and charging strategies; The solution verification and dynamic pricing module is used to verify scheduling feasibility through RFID and dynamically adjust the rental price based on the supply and demand ratio. The real-time scheduling and distributed collaboration module is used to implement demand forecasting, price fine-tuning, and blockchain recording using distributed algorithms.

[0013] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent rental management method based on shared power banks as described in the first aspect of the present invention.

[0014] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent rental management method based on shared power banks as described in the first aspect of the present invention.

[0015] The beneficial effects of this invention are as follows: By performing multi-scale analysis of the rental demand time series using the VMD algorithm, stable demand and fluctuating demand components are effectively separated, achieving accurate capacity planning based on structured demand; by combining the DBSCAN clustering algorithm to identify high-density rental areas and integrating the VMD decomposition results to dynamically screen and split candidate sites, the problem of static site layout and lack of demand adaptability in the existing system is solved, ensuring that site deployment conforms to spatial clustering patterns and matches long-term service intensity; by constructing a degradation cost model based on the number of cycles and incorporating it into the Q-learning reinforcement learning framework for long-term revenue prediction, the collaborative optimization of equipment physical wear and tear, dynamic pricing, and resource allocation strategies is achieved, making up for the management blind spot in the existing system that ignores asset depreciation and leads to inflated revenue. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of an intelligent rental management method based on shared power banks in Example 1.

[0018] Figure 2 This is a schematic diagram of the structure of an intelligent rental management system based on shared power banks in Example 1.

[0019] Figure 3 This is a flowchart of the scheduling optimization and pricing process in Example 1. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, referring to Figures 1 to 3 This is the first embodiment of the present invention, which provides an intelligent rental management method based on shared power banks, including the following steps: S1. Collect rental data of shared power banks, use the VMD algorithm to obtain the total rental demand time series, use DBSCAN clustering based on the time series to generate a list of site locations, and obtain the total capacity demand based on the time series. The degradation cost of the power bank is calculated, the net rental income is obtained based on the degradation cost, a long-term revenue forecast is obtained based on the net rental income using a reinforcement learning algorithm, the real-time remaining capacity is obtained based on the revenue forecast and the total capacity demand, and a candidate scheduling scheme is generated based on the site location list. Specifically, the system collects rental data of shared power banks, uses the VMD algorithm to obtain the total rental demand time series, uses DBSCAN clustering based on the time series to generate a list of station locations, and obtains the total capacity demand index based on the time series. It also extracts historical rental data (such as the past month, including rental time, station latitude and longitude, and number of rentals) from the MySQL database, obtains environmental data (weather, holidays, and their impact on demand fluctuations) through API, and uses RFID devices to record real-time rental events (pickup and return time, location, and device ID). Generate total rental demand time series using VMD algorithm. Including stable (basic) demand and fluctuating demand To quantify the intensity of demand for a site, the calculation formula is as follows: , , in, Total rentals refer to the number of rentals at all stations (the actual deployment points of the power banks) (times / hour, based on historical data statistics, such as an average of 50 rentals per hour in shopping malls). To stabilize demand (stabilize the leasing model, such as the regular demand of shopping malls on weekdays). For fluctuating demand (random rentals, such as during holidays or peak events). and This represents the multiple sub-signals obtained by decomposing a time series using the VMD algorithm. Each sub-signal corresponds to a different frequency component, and the low-frequency mode function is also mentioned. ( This constitutes stable demand. This reflects long-term, stable rental demand patterns, such as the regular demand in shopping malls on weekdays, while high-frequency mode functions... )( ) constitutes fluctuating demand It captures short-term, random changes in rental demand, such as demand fluctuations during holidays or peak event periods, and more accurately quantifies the demand intensity of sites. Z represents the number of low-frequency modes (to capture low-frequency stability requirements) and the total number of modes (including low-frequency and high-frequency modes). Based on RFID-recorded rental events (latitude, longitude, and timestamp), DBSCAN clustering is used to identify high-density rental areas as candidate sites. This is then optimized using VMD decomposition results to generate a list of site locations. Specifically, DBSCAN (using the Python sklearn library) is run on the latitude and longitude of rental events to generate candidate sites (high-density rental points, such as shopping mall entrances), combined with stable requirements. Prioritize areas with high stable demand (such as...) For locations with a capacity of >15 times / hour, ensuring stable user service, high-demand rental areas with large clusters (defined as clusters with diameters exceeding a set distance, such as 300 meters) are divided into multiple stations based on road network (OSM data), ensuring each station has a capacity of <100 power banks. Adjust the site coverage area (e.g., expand the coverage radius by 10% if demand is high during holidays), and generate a list of site locations (approximately 100 sites, in the format of [longitude, latitude, expected demand]). The total capacity requirement of a site is obtained using a capacity estimation method based on historical peak statistics. The formula is: , in, The total capacity is T (total number of power banks), and the total time is T = 24 hours. Let A be the total stable demand over 24 hours (e.g., 100 times / day), and let A be the coverage rate, such as a target of 80%. To ensure coverage of key needs, To obtain historical peak statistics, the peak values ​​were identified across all days by performing VMD decomposition on the historical rental sequence. The maximum value is equivalent to "the abnormal demand intensity of the busiest hour in the past month". Assign coefficients to fluctuating demand, such as , Average turnover rate (i.e., the average number of times each power bank is rented per day, for example) =3 means that each power bank is rented an average of 3 times per day.

[0024] By integrating historical rental data from MySQL, environmental factors such as weather and holidays obtained through APIs, and spatiotemporal information on pick-up and return events collected in real time by RFID devices, a multi-source collaborative data sensing system was constructed, overcoming the modeling bias caused by the single data dimension in traditional methods. Furthermore, the variational mode decomposition (VMD) algorithm was used to adaptively decompose the total rental demand time series, accurately separating it into stable and fluctuating demand components. This effectively identified the intensity of basic services and sudden peak patterns, solving the problems of signal aliasing and difficulty in distinguishing long-term trends from short-term disturbances in traditional time series analysis, providing a structured basis for refined operations. Based on this, the DBSCAN clustering algorithm was used to analyze R... Spatial density analysis of rental events recorded by FID identifies high-frequency usage areas as candidate sites. This is combined with stable demand intensity obtained from VMD decomposition for priority selection, ensuring that site layout reflects both user behavior aggregation characteristics and long-term service needs. Simultaneously, for large-scale high-demand areas, logical decomposition is performed using road network data to avoid single-point capacity overruns, improving service accessibility and management controllability. Based on the sum of historical peak fluctuating demand and stable demand extracted by VMD, a capacity estimation model is constructed that balances daily coverage and peak redundancy capabilities. By introducing a fluctuating demand allocation coefficient, the total capacity configuration is dynamically adjusted, avoiding the resource over-allocation or service gap problems caused by relying solely on historical maximum values.

[0025] Furthermore, the degradation cost of the power bank is calculated, and the net rental revenue is obtained based on the degradation cost. A reinforcement learning algorithm is then used to obtain a long-term revenue forecast based on the net rental revenue. Real-time remaining capacity is obtained based on the revenue forecast and total capacity demand. Candidate scheduling schemes are generated based on the site location list. The system uses the integrated Wi-Fi and communication modules within the power bank to scan all power banks every hour, obtaining the initial System-on-Chips (SoC) (0%–100% battery level) and cycle count, recording the depth of discharge (DoD), and calculating the degradation cost of each power bank. (This refers to the degradation of battery health in power banks due to prolonged charging and discharging (such as reduced capacity and increased internal resistance), which is considered physical aging. Priority will be given to allocating power banks to devices with low degradation levels, using the following formula: , in, Cost of degradation for a single power bank The initial cost of purchasing the power bank is represented by B, which is the number of cycles, indicating the cumulative number of charge and discharge cycles (obtained from Wi-Fi data, such as 500 times). For battery life cycles (1000 cycles); Calculate the net revenue from actual rentals (referring to the specific act of a user "borrowing a power bank once"). This represents the total net revenue (in yuan) of all leases at a certain time step. Taking into account both revenue and cost, the formula is: , in, The rental duration (in hours, e.g., 2 hours / time). The rental price is real-time (initially 5 yuan / hour). Costs are amortized for each rental transaction; The current state is calculated using a reinforcement learning algorithm (Q-learning). Next action Long-term earnings forecast The status refers to site demand and battery status, while the actions refer to device allocation (device ID to site) and price adjustment. The Q value reflects the cumulative expected revenue after dynamic price adjustment and optimized allocation. The formula is: , Where α is the learning rate, such as α=0.1 (controlling the update speed), and γ is the discount factor, such as γ=0.9 (weight of future returns). Based on historical expected returns, To predict subsequent returns for (the maximum Q-value of future states), For future actions, it indicates the possible actions at the next time step t+1 (such as equipment allocation or price adjustment). Based on the output of Q-learning, the number of devices is allocated according to the stable demand b(t) (e.g., 20 power banks are allocated to site A, prioritizing b(t) > 15 times / hour). The allocation for each site is summed to obtain the total number of devices. ,Will Divide by average turnover rate To obtain the number of devices already in use This reflects the total stable demand for long-term equipment allocation; The formula for calculating real-time remaining capacity is: , in, Remaining capacity (number of power banks); Based on the generated list of site locations, the sites are sorted according to their projected demand. Sites with stable demand exceeding a threshold (e.g., >15 times / hour) are prioritized for allocation of power bank quantity, and the base capacity is calculated. Based on the geographical distribution of sites and the total number of devices, device IDs are evenly distributed to each site, and an initial rental price is set as the base price (e.g., 5 yuan / hour). Finally, the device IDs, allocated sites, SoC thresholds (>50% priority) and charging strategies (off-peak charging) are integrated to form an initial candidate scheduling scheme that includes the mapping relationship between "devices → sites".

[0026] By integrating Wi-Fi and communication modules to collect key state parameters of the power bank in real time, such as SoC, cycle count, and depth of discharge, a degradation cost quantification model based on cycle count was constructed. This explicitly incorporates battery physical aging factors into the calculation of net revenue per rental, solving the problem of inflated revenue caused by traditional systems that only focus on rental income while ignoring equipment deterioration. Furthermore, a Q-learning reinforcement learning algorithm was introduced, using site demand intensity and battery health status as system states and equipment allocation and price adjustment as actions. Through iterative learning of long-term revenue estimates using Q-values, the decision-making objective was upgraded from immediate revenue to cumulative expected revenue, effectively overcoming the shortcomings of traditional rule-based scheduling that lack foresight and adaptability. On this basis, combined with the stable demand sequence and total capacity demand obtained from VMD decomposition, the real-time remaining capacity was dynamically calculated. Initial candidate scheduling schemes were generated based on demand priority, equipment degradation degree (prioritizing low-degradation equipment), and geographical distribution balance, ensuring that resource allocation both meets service demand and extends asset lifespan. This achieved an organic unity of accurate equipment cost accounting, dynamic evolution of the revenue model, and intelligent generation of scheduling strategies, significantly enhancing the system's economy and sustainability in complex operating environments.

[0027] S2. Calculate the charging time required for the power bank, generate the fitness of candidate scheduling schemes based on the charging time, optimize the candidate scheduling schemes based on the fitness, verify the optimized candidate scheduling schemes using RFID based on the remaining capacity, and generate the rental price based on the verified candidate scheduling schemes. Rental prices are scheduled and fine-tuned in real time using a distributed algorithm to obtain real-time scheduling results; Specifically, the process involves calculating the charging time required for the power bank, generating a candidate scheduling scheme based on the charging time's fitness, optimizing the candidate scheduling scheme based on the fitness, verifying the optimized candidate scheduling scheme using RFID based on the remaining capacity, and generating a rental price based on the verified candidate scheduling scheme. This involves dividing the day into off-peak (0-6 AM, 0.3 yuan / kWh), off-peak (6-10 PM, 0.6 yuan / kWh), and peak (10-12 AM, 1.0 yuan / kWh) electricity price periods. Specifically, this involves obtaining local grid electricity price data (such as the publicly available TOU electricity price list) via API, combining this with the off-peak charging demand of shared power banks to determine that off-peak periods are low-consumption periods at night, off-peak periods are typical daytime periods, and peak periods are high-consumption periods at night, and then using candidate scheduling schemes and degradation costs. The charging time of device i at time t is obtained. (Refers to the current state of charge) To reach the target state of charge The required time), the formula is: , in, The target battery level (e.g., 80%). The current battery level (obtained from Wi-Fi, 0%–100%, combined with the device ID from the initial scheduling plan), where 'r' represents the device ID used to distinguish different power bank devices. This refers to the charging power (5W, standard for power banks). Based on TOU decision-making, the decision variables are obtained. Integrate degradation costs Prioritize devices with low degradation, using the following formula: , in, When the value is 1, it means that charging is allowed; otherwise, it is 0, which means that charging is not allowed. D is a preset threshold. Based on TOU decision-making, electricity price and time constraints are provided to calculate the fitness of candidate scheduling schemes. The formula is: , in, Fitness (yuan) represents the net gain at time t. The rental price is in real time (RMB / hour). Charging costs (via TOU electricity price ×) ×Power acquisition); Based on fitness The scheduling scheme is optimized using a genetic algorithm (GA). The specific operations include: Selection operation: Calculate the fitness of each scheduling scheme in the candidate scheduling scheme set. (With the goal of maximizing profits, considering factors such as rental income and transportation costs), select a highly fit option (such as through a tournament selection method). Cross operation: Cross the selected schemes (such as single-point cross), merge the equipment allocation and pricing strategies of different schemes, and generate new candidate scheduling schemes; Mutation operations: Make minor adjustments to some schemes with a low probability (such as randomly changing the equipment allocation site or adjusting the price) to increase population diversity; Repeated selection, crossover, and mutation are performed until the fitness converges (e.g., the change in the optimal solution's profit is less than 0.01%), at which point the process stops and generates an optimized set of scheduling schemes. Each scheme includes equipment allocation (e.g., E001→S1, E002→S2) and an initial price. Verify the feasibility of the optimized candidate scheduling scheme using RFID (Radio Frequency Identification) technology. Specific operations: Check the actual location of the equipment: Confirm whether the equipment is at the allocation station (e.g., whether E001 is at S1) through RFID tags. Verify remaining capacity: RFID, combined with the battery management chip (BMS) built into the power bank, transmits the remaining power capacity data back via the Wi-Fi communication module, and compares the current remaining power capacity of the device with the real-time rental demand (e.g., if the user needs 1000mAh, is the device's remaining capacity sufficient). Eliminate infeasible solutions: If the equipment is in the wrong location or the capacity is insufficient, remove it from the candidate solutions and output a set of feasible solutions, including the verified equipment allocation and the initial price; For candidate scheduling schemes verified by RFID, prices are adjusted based on real-time leasing (such as demand fluctuations and market conditions at time t) to optimize prices and maximize revenue, resulting in a real-time optimized leasing price. The formula is: , , , in, The supply-demand ratio (unitless) at time t reflects the relative proportion of demand and supply. Let i be the rental demand (times / hour) for site i at time t. Let i be the theoretical number of power banks available at time t. This is the upper limit for the retail price (in yuan / hour, e.g., set to 6 yuan). The base rental price is (yuan / hour, e.g., 5 yuan). This is a correction factor.

[0028] By acquiring grid TOU electricity price data through API, and combining the current power consumption of the equipment, the target power consumption, and the charging power, the charging time is accurately calculated. Time-based charging decision variables are constructed, and a priority charging strategy for low-degradation equipment is explicitly introduced during the optimization process, effectively reducing long-term operation and maintenance costs. Furthermore, a fitness function based on net revenue minus charging cost is designed, and a genetic algorithm is used to select, crossover, and mutate candidate scheduling schemes. This maximizes global revenue under multiple constraints, including equipment allocation, pricing strategies, and time-of-use pricing, overcoming the local optima problem caused by traditional scheduling methods relying on empirical rules or static models. Crucially, RFID technology is introduced to monitor the optimized... The scheduling scheme undergoes physical layer verification. By verifying the actual location and remaining capacity of equipment in real time, infeasible schemes caused by equipment misalignment or insufficient capacity are eliminated, ensuring that scheduling instructions are consistent with the actual equipment status and solving the industry problem of "disconnect between planning and execution". Finally, based on the verified feasible schemes, the rental price is dynamically adjusted in conjunction with the real-time supply and demand ratio. Prices are moderately increased during peak demand to curb overuse, and prices are reasonably reduced when supply is sufficient to improve utilization. This achieves refined and real-time control of the price mechanism. This scheme realizes a leap from "extensive scheduling" to intelligent optimization based on "cost perception, physical reliability, and revenue drive", significantly improving the system's operational efficiency and economic benefits.

[0029] Furthermore, based on the rental price, a distributed algorithm is used for real-time scheduling and price fine-tuning. The real-time scheduling result is obtained by the shared power bank smart rental management platform collecting the pick-up and return records, current quantity, and power status of power banks at each site every 15 minutes through the Wi-Fi module. This data is then aggregated to form a real-time rental event stream. Based on the real-time rental event stream, the rental frequency of each site is statistically calculated according to time windows (e.g., hourly). This data is then compared with historical demand data (from MySQL) to identify the fluctuation trend of current demand relative to the average (e.g., sudden increase or decrease). Based on the moving average time series forecasting method, combined with external factors such as holidays and weather, a short-term forecast of demand changes for the next 2-4 hours is made, and the expected demand for each site is output. The allocation of equipment across sites is optimized using a distributed algorithm. Specifically, multiple computing nodes (servers executing computing tasks) are deployed across different sites. Each node is responsible for processing a portion of the site's data to achieve parallel computing and load balancing. During optimization, it is ensured that the power variation at each site does not exceed 10% of its maximum power (e.g., if a site's maximum power is 100W, the adjustment range should be between 90W and 110W) to maintain stable site operation, based on the optimized rental price. Based on the projected demand at each site, rental prices will be dynamically adjusted. If price fluctuations exceed a preset threshold... (For example, if the price fluctuates from 2.5 yuan / hour to 2.625 yuan / hour), then reduce the adjustment step size (from 0.1 yuan to 0.08 yuan), take the iterative average to stabilize the price, and then apply the optimized scheduling result (equipment allocation, such as E001→S1, price) to the system. The price (RMB 0.52 / hour) and device status (battery level, site) are recorded on the blockchain via smart contracts to ensure transparency and immutability.

[0030] By constructing a real-time rental event stream and combining it with a moving average time-series forecasting model that incorporates external factors such as holidays and weather, accurate short-term forecasts of demand for the next 2-4 hours are achieved, overcoming the problems of delayed response and single-prediction in traditional systems. Furthermore, a distributed algorithm is adopted to offload computational tasks to multiple nodes, with each node processing local site data in parallel, achieving load balancing and efficient collaborative optimization. This ensures overall scheduling efficiency while avoiding the performance bottleneck of the central server. During the optimization process, a physical constraint is introduced that site power changes do not exceed 10%, effectively suppressing sudden power load changes caused by concentrated charging and discharging of equipment, thus improving the safety and stability of site operation. For the pricing mechanism, a dynamic fine-tuning strategy based on supply and demand forecasts is designed, and a price oscillation suppression mechanism is introduced—when price fluctuations exceed 5%, the adjustment step size is automatically reduced and the iterative average is taken, avoiding the user experience degradation and market chaos caused by frequent price jumps in traditional dynamic pricing. This achieves a shift from "centralized delayed response" to "distributed real-time collaboration," constructing an efficient, stable, and transparent intelligent rental management system.

[0031] S3. Make leasing decisions based on real-time scheduling results and generate leasing records. Generate settlement records and updated degradation costs based on the leasing records. Specifically, rental decisions are made based on real-time scheduling results, and rental records are generated. This involves obtaining real-time scheduling results and blockchain records. Users scan the RFID tag on the device, use the shared power bank smart rental management platform to read the device ID and verify whether it matches the optimized candidate scheduling scheme's assigned site (e.g., whether E001 is in S1). Decentralized identity (DID) is generated using federated learning to confirm the user's legitimacy, and the association between the user ID and device ID is recorded. Users select the rental type (long-term contract or real-time rental) on their mobile devices. Long-term contracts are based on a fixed price (e.g., 10 yuan / day), while real-time rentals incorporate the optimized price from the real-time scheduling results. The rental decision-making process utilizes zero-knowledge proof (ZKP) technology. Users prove the validity of their rental requests (equipment availability and identity legitimacy) without disclosing time or location information (e.g., verifying "battery capacity ≥ demand" via ZKP). Users query the device status via blockchain (e.g., "Device ID: E001, battery capacity 85%), ensuring transparency. The shared power bank smart rental management platform verifies the remaining capacity of the equipment. Whether the rental requirements are met (e.g., user needs 1000mAh, device battery capacity 2000mAh), the smart contract records the rental start time, device ID, battery capacity, rental type and price, generates a rental record, and stores user interaction data (e.g. rental duration, user ID) together with the rental record in the database.

[0032] By reading device IDs using RFID tags and verifying their consistency with the assigned stations in the scheduling scheme in real time, invalid rentals due to device misplacement or illegal movement are effectively prevented. Combined with federated learning to generate decentralized identities (DIDs), identity legitimacy authentication is completed without centralized storage of sensitive user information, solving the data leakage risks and privacy compliance issues associated with traditional centralized authentication. Furthermore, zero-knowledge proof (ZKP) technology is introduced, enabling users to prove to the system that key conditions such as "device available," "legitimate identity," and "sufficient battery power" are met without exposing specific time, location, or other private information, achieving the dual goals of privacy protection and functional verification. Users can publicly query device status (such as battery power and location) through the blockchain, ensuring transparent and verifiable service information and avoiding information asymmetry on the platform's side. The rental process is automatically recorded by smart contracts, including start time, device ID, battery power, and price, generating an immutable rental record. This record, combined with user interaction data, is stored in a database, ensuring the traceability and legal validity of the entire transaction process.

[0033] Furthermore, generating settlement records and updated degradation costs based on rental records involves obtaining rental records (e.g., "Device ID: E001, Rental Start Time: 2025-09-22 10:00, Battery Level: 85%, Price: ht=2.58 yuan / hour") and user interaction data. The device status at the time of return is obtained via the Wi-Fi module and stored in a MySQL database. When a user returns the power bank, the Wi-Fi module scans the device ID to verify if it matches the rental record (e.g., E001 returned to S1), compares the battery level at the time of return with the initial battery level in the rental record, calculates the actual depth of discharge (DoD) (e.g., initial 85%, returned 50%, DoD=35%), and calculates the rental fee based on the rental type. Specifically, the real-time rental fee is calculated as follows: Fee = × Rental period (e.g., 2.58 × 2 = 5.16 yuan) to calculate long-term fees. Fee = deduction from a fixed price (e.g., 10 yuan / day). Fees are deducted from user accounts linked to RFID, and smart contracts record settlement details (device ID, fee, return time), which are stored in the blockchain. Degradation costs are updated based on Discharge Depth of Discharge (DoD), and return data and degradation costs are analyzed to generate scheduling optimization feedback. Settlement records (fee, return time, device ID), updated degradation costs, and optimization feedback are output and stored in a MySQL database to provide data support for optimization, forming a closed loop.

[0034] By automatically acquiring rental records and combining them with real-time data collection of device status upon return via Wi-Fi, the system automatically matches and verifies device ID, power level changes, and rental information, effectively preventing incorrect or fraudulent returns. Based on the power level difference before and after return, the system accurately calculates the actual depth of discharge (DoD) and dynamically updates the degradation cost model for each power bank, overcoming the cost distortion caused by traditional systems that only roughly estimate losses per use, making asset depreciation accounting more scientific and reasonable. Rental fees are automatically calculated based on real-time prices or long-term contract rules and deducted from the user account linked to RFID. Settlement details are written to the blockchain via smart contracts, ensuring tamper-proof and traceable transactions, enhancing financial transparency and user trust. More importantly, the system aggregates and analyzes each return data with the updated degradation cost, generating scheduling optimization feedback information to provide data support for subsequent device allocation, maintenance plans, and pricing strategies. This breaks the limitations of the traditional one-way "use-settlement" process, constructing a complete closed-loop management system of "use-settlement-cost update-feedback optimization."

[0035] This embodiment also provides an intelligent rental management system based on shared power banks, including: The data acquisition and demand forecasting module is used to integrate historical and real-time data, use VMD and DBSCAN to generate a site list and estimate capacity demand; The cost accounting and revenue forecasting module is used to calculate net income and long-term revenue forecasts based on degradation costs and Q-learning. The scheduling scheme generation and optimization module is used to combine TOU electricity price and genetic algorithm to optimize equipment allocation and charging strategies; The solution verification and dynamic pricing module is used to verify scheduling feasibility through RFID and dynamically adjust the rental price based on the supply and demand ratio. The real-time scheduling and distributed collaboration module is used to implement demand forecasting, price fine-tuning, and blockchain recording using distributed algorithms.

[0036] This embodiment also provides a computer device applicable to a smart rental management method based on shared power banks, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the smart rental management method based on shared power banks as proposed in the above embodiment.

[0037] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0038] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the intelligent rental management method and system based on shared power banks as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0039] In summary, this invention utilizes the VMD algorithm to perform multi-scale analysis of rental demand time series, effectively separating stable and fluctuating demand components, and achieving precise capacity planning based on structured demand. By combining the DBSCAN clustering algorithm to identify high-density rental areas and integrating VMD decomposition results for dynamic screening and splitting of candidate sites, it solves the problems of static site layout and lack of demand adaptability in existing systems, ensuring that site deployment conforms to spatial clustering patterns and matches long-term service intensity. Furthermore, by constructing a degradation cost model based on the number of iterations and incorporating it into the Q-learning reinforcement learning framework for long-term revenue prediction, it achieves synergistic optimization of equipment physical wear and tear, dynamic pricing, and resource allocation strategies, compensating for the management blind spot in existing systems that neglects asset depreciation, leading to inflated revenue.

[0040] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A smart leasing management method based on a shared power bank, characterized in that: comprising, collecting the leasing data of the shared power banks, obtaining the total leasing demand time series using the VMD algorithm, generating the site location list based on the time series using DBSCAN clustering, and obtaining the total capacity demand based on the time series; calculating the degradation cost of the power banks, obtaining the net leasing revenue based on the degradation cost, obtaining the long-term revenue estimate using the reinforcement learning algorithm based on the net leasing revenue, obtaining the real-time residual capacity based on the revenue estimate and the total capacity demand, and generating the candidate scheduling scheme based on the site location list; calculating the charging time required by the power banks, generating the candidate scheduling scheme fitness based on the charging time, optimizing the candidate scheduling scheme based on the fitness, verifying the optimized candidate scheduling scheme using RFID based on the residual capacity, and generating the leasing price based on the verified candidate scheduling scheme; The leasing price is scheduled and the price is fine-tuned in real time through a distributed algorithm to obtain a real-time scheduling result, leasing decisions are made based on the real-time scheduling result, and leasing records are generated. Settlement records and updated degradation costs are generated based on the leasing records. 2.The smart management method of sharing power bank-based rental according to claim 1, characterized in that: The collection of the shared power bank rental data, the use of VMD algorithm obtains the total rental demand time sequence, the use of DBSCAN clustering based on the time sequence generates the site position list, and the total capacity demand is obtained based on the time sequence , including stable demand and fluctuating demand ; Based on RFID-recorded rental events, DBSCAN clustering is used to identify high-density rental areas as candidate sites, and VMD decomposition results are combined for optimization to generate a list of site locations. The total site capacity requirement is then obtained based on a capacity estimation method using historical peak statistics. . 3.The smart management method based on the shared power bank, according to claim 2, wherein: The degradation cost of the power bank is calculated, the net income of the lease is obtained based on the degradation cost, the long-term income estimate value is obtained based to the net income of the lease using a reinforcement learning algorithm, the real-time residual capacity is obtained based on the income estimate value and the total capacity demand, and the candidate scheduling scheme is generated based on the generated site location list and the net income of the actual lease , and the net income is used to calculate the long-term income estimate value of the current state by using a reinforcement learning algorithm , the long-term income estimate value of the action executed in the next state is calculated based on the output of the reinforcement learning algorithm, the number of devices is allocated according to the stable demand b(t), the allocation amount of each site is accumulated to obtain , the used device number is obtained by dividing by the average turnover rate , the real-time residual capacity is calculated, and the initial candidate scheduling scheme is obtained based on the generated site location list. 4.The smart management method based on the shared power bank, according to claim 3, wherein: The computing charging time of the power bank is based on the charging time to generate a candidate scheduling scheme fitness, based on the fitness to optimize the candidate scheduling scheme, based on the remaining capacity to verify the optimized candidate scheduling scheme using RFID, based on the verified candidate scheduling scheme, generating a rental price by dividing a day into low, flat, and peak price stages, using the initial candidate scheduling scheme and degradation cost , obtaining the charging time of device i at time t , based on the TOU decision, obtaining the decision variable , integrating the degradation cost constraint low degradation device priority, based on the TOU providing electricity price and time period constraints, calculating the fitness of the candidate scheduling scheme ; Based on fitness The scheduling scheme is optimized using a genetic algorithm, the feasibility of the optimized candidate scheduling scheme is verified using RFID, and the real-time optimized rental price is obtained based on real-time rental adjustment of the price for the candidate scheduling scheme that passes the RFID verification .

5. The intelligent leasing management method based on shared power banks according to claim 4, wherein: The lease price-based real-time scheduling and price fine-tuning are performed by a distributed algorithm to obtain a real-time scheduling result , which calculates the expected demand of each site, optimizes the allocation of equipment at the site by a distributed algorithm, and dynamically adjusts the lease price based on the optimized lease price and the expected demand of each site. If the price oscillation exceeds a preset threshold , the adjustment step is reduced, the iterative average is taken to stabilize the price, the optimized scheduling result and the equipment state are recorded to the blockchain through the smart contract to ensure transparency and non-tamperability.

6. The intelligent leasing management method based on shared power banks according to claim 5, wherein: The lease decision is made based on the real-time scheduling result, and the lease record is generated. The real-time scheduling result and the blockchain record are obtained, the RFID tag on the user's device is scanned, the device ID is read using the shared power bank intelligent lease management platform, and it is verified whether it is consistent with the site allocated by the optimized candidate scheduling scheme. Combined with federated learning, a decentralized identity is generated to confirm the legitimacy of the user's identity. The user selects the lease type on the mobile terminal. The long-term contract is based on a fixed price, and the real-time lease is combined with the optimized price in the real-time scheduling result The lease decision is made, the zero-knowledge proof technology is applied, the user proves the validity of the lease request, the user queries the device status through the blockchain, and the shared power bank intelligent lease management platform checks the remaining capacity of the device Whether the lease demand is met, the smart contract records the lease start time, device ID, power, lease type and price, and generates a lease record.

7. The intelligent leasing management method based on shared power banks according to claim 6, wherein: The settlement records and the updated degradation costs based on the leasing records refer to obtaining the leasing records and the user interaction data, obtaining the device state at the time of return through the Wi-Fi module, and storing them in the MySQL database. When the user returns the power bank, the Wi-Fi module scans the device ID, verifies whether it matches the leasing record, compares the power at the time of return with the initial power in the leasing record, calculates the actual depth of discharge DoD, calculates the leasing fee according to the leasing type, deducts the fee through the RFID associated user account, records the settlement details through the smart contract, and stores them in the blockchain.

8. A smart leasing management system based on shared power banks, based on the smart leasing management method of shared power banks according to any one of claims 1-7, characterized in that: comprising, The data acquisition and demand prediction module is used for fusing historical and real-time data, generating a site list and estimating capacity demand using VMD and DBSCAN; The cost accounting and revenue estimation module is used for calculating net revenue and long-term revenue estimates based on degradation costs and Q-learning; The scheduling scheme generation and optimization module is used for optimizing device allocation and charging strategies in combination with TOU electricity prices and genetic algorithms; The scheme verification and dynamic pricing module is used for verifying the feasibility of scheduling through RFID and dynamically adjusting the leasing price based on the supply-demand ratio; The real-time scheduling and distributed collaboration module is used for demand prediction, price fine-tuning, and blockchain recording using a distributed algorithm. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to implement the steps of the intelligent leasing management method based on shared power banks according to any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to implement the steps of the intelligent leasing management method based on shared power banks according to any one of claims 1-7.

Images