Electric vehicle charging station behavior optimization and joint pricing optimization method

Abstract

The invention discloses a joint pricing method based on behavior optimization of an electric vehicle charging station, and provides a framework for optimizing an electric vehicle pricing policy and a charging plan, and the framework combines endogenous behaviors of a charging station electric vehicle user and simulates the behavior response of the electric vehicle user to the price. The purpose of the present invention is to minimize the total cost of providing a charging service without sacrificing the user experience. According to the method provided by the invention, the user behavior of the charging station is stimulated based on time pricing so as to achieve the goal of a charging station operator, and the framework combines endogenous human behaviors by clearly considering the will of charging delay and the plug-in duration of each time period as a function of hour price. And finally, how to improve the operation of the charging station by increasing the net income of a charging station operator and reducing the public utility cost is simulated and displayed, the user experience is not influenced or the average cost paid by an electric vehicle owner is not increased, and the problem of overtime stay of the electric vehicle charging station is solved.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle charging technology, and to a method for optimizing the behavior of electric vehicle charging stations and optimizing joint pricing. In particular, it relates to a method for optimizing the behavior of electric vehicle charging stations by using time-based pricing to incentivize user behavior at charging stations, a random pricing experiment, and joint pricing optimization. Background Technology

[0002] Plug-in electric vehicles (PEVs) are a type of new energy vehicle that can be charged by an external power source. They have large battery capacities and can travel a long distance in pure electric mode. There are two main charging methods for PEVs: home charging and public charging. Home charging typically uses standard household power outlets or dedicated home charging stations. Charging takes longer but is less expensive. Public charging takes place at dedicated charging stations, which may offer fast charging services, replenishing a vehicle's battery quickly, but the cost may be higher.

[0003] Existing PEV charging methods mainly include home charging and workplace charging, with diverse pricing approaches, but all have certain drawbacks. Home charging typically uses fixed rates or time-of-use pricing, which has lower costs but cannot flexibly reflect real-time electricity supply and demand, potentially increasing load pressure on the grid during peak hours. Workplace charging, due to higher operating costs and complex and variable user behavior, struggles to implement effective pricing strategies, often resulting in excessively high prices or difficulty in attracting users. These pricing methods fail to fully consider the charging preferences of different users, differences in demand, and the actual load conditions of the power grid, thus affecting the effectiveness of pricing strategies and user satisfaction, and hindering the further development of the PEV charging market. Summary of the Invention

[0004] In view of this, the present invention provides a method for optimizing the behavior of electric vehicle charging stations and for joint pricing optimization. It aims to address the technical problem of how to jointly optimize hourly electricity prices and the schedule for providing energy to plug-in electric vehicle (PEV) users given a time-of-use (TOU) price, in order to minimize total utility costs and maximize the net revenue of the charging station operator. Furthermore, this framework also allows for achieving other objectives, such as minimizing emissions or altering the electricity demand curve, as the TOU price can serve as a representation of these objectives. For example, if the objective is to shift charging electricity demand from the afternoon peak, setting the TOU to be significantly more expensive during peak hours will incentivize station operators to attempt to shift the demand curve away from peak periods. Similarly, if the objective is to minimize emissions, the TOU rate can be set negatively correlated with high-emission periods to suppress charging during these periods.

[0005] On the one hand, the present invention provides a joint pricing method based on electric vehicle charging station behavior optimization, including:

[0006] The system obtains users' historical charging behavior information and assumed time-of-use electricity prices. The users' historical charging behavior information includes: arrival time at the charging station, required energy, and maximum power capacity.

[0007] The probability of a user selecting a charging mode is predicted based on the user's historical charging behavior information; the charging modes include: regular charging mode, planned charging mode, and off-duty mode.

[0008] Based on the cost of each charging mode, the operator's total expected cost is obtained;

[0009] With the goal of minimizing total cost, the hourly price R for regular charging and planned charging modes is obtained based on the predicted probability of user charging patterns. nor and R pre And the charging power P over time t tc .

[0010] Furthermore, the operator's total expected cost is expressed as:

[0011]

[0012] Where, p nor p pre and p dep These represent the probabilities that a user selects the regular charging mode, the planned charging mode, or leaves the charging mode within charging plan i; C Dep This refers to the total cost of providing charging services, excluding any opportunity costs, therefore assuming the cost of taking a vacation is zero; R nori This is the hourly rate paid by the user when selecting the regular charging mode; P maxi It is the maximum charging power that the car can accept during time period i; t ec The shortest time required to fully charge a car at maximum power, t, is calculated if the user leaves before the car is fully charged. ec =t nor , t ec P is the electricity cost at time t, expressed in yuan per kilowatt-hour; tc R is the charging power at time t; prei It is the hourly rate that users pay when selecting a planned charging mode in charging plan i.

[0013] Furthermore, in the normal charging mode, from the moment the vehicle is connected to the charging device, the charging system charges the vehicle at maximum output power until the vehicle is fully charged or the user drives the vehicle away, at which point the charging system will automatically stop charging; the normal charging mode charges fees according to the hourly billing standard.

[0014] In planned charging mode, the vehicle will be charged within the time specified by the user for vehicle access to charging until the set charging amount is reached or the user drives the vehicle away, at which point the charging system will automatically stop charging; if the user picks up the vehicle early, the vehicle may not reach the expected charging level; and if the user picks up the vehicle late, an overtime penalty will be incurred.

[0015] If a user decides not to charge due to high charging prices, they can choose the "Away" mode. In this mode, the user's cost is assumed to be zero, and no fees will be incurred when the vehicle is not charging.

[0016] Furthermore, in the normal charging mode, from time t0 to time t nor Start at time t ec The end of the charging plan has a cost:

[0017]

[0018] Furthermore, in the planned charging mode, given the required energy E req And a fixed departure time t pre The cost to the operator when they intend to pick up the car is expressed as follows:

[0019]

[0020] Furthermore, it also includes:

[0021] Based on the hourly price R for regular charging mode and planned charging mode. nor and R pre And the charging power P over time t tc Determine the timing for peak pricing.

[0022] Furthermore, it also includes:

[0023] Estimate the relationship between charging duration and hourly price;

[0024] With the goal of extending the duration of plugging in, we optimize charging power delivery and offer discounted hourly rates;

[0025] Furthermore, it also includes:

[0026] Estimate the relationship between charging duration and hourly price;

[0027] With the goal of maximizing utilization, increasing throughput, and minimizing charging time at charging stations, the hourly rate is increased to discourage prolonged charging and incentivize users to pick up their vehicles quickly.

[0028] In another aspect, the present invention provides a joint pricing system based on electric vehicle charging station behavior optimization, comprising:

[0029] The historical information acquisition module is used to acquire users' historical charging behavior information and assumed time-of-use electricity prices. The users' historical charging behavior information includes: arrival time at the charging station, required energy, and maximum power capacity.

[0030] The charging mode prediction module is used to predict the probability of a user selecting a charging mode based on the user's historical charging behavior information; the charging modes include: regular charging mode, planned charging mode, and off-duty mode.

[0031] The cost construction module is used to obtain the operator's total expected cost based on the cost of each charging mode, expressed as:

[0032]

[0033] Where, p nor p pre and p dep These represent the probabilities that a user selects the regular charging mode, the planned charging mode, or leaves the charging mode within charging plan i; C Dep This refers to the total cost of providing charging services, excluding any opportunity costs, therefore assuming the cost of taking a vacation is zero; R nori This is the hourly rate paid by the user when selecting the regular charging mode; P maxi It is the maximum charging power that the car can accept during time period i; t ec The shortest time required to fully charge a car at maximum power, t, is calculated if the user leaves before the car is fully charged. ec =t nor , t ec P is the electricity cost at time t, expressed in yuan per kilowatt-hour; tc R is the charging power at time t; prei It is the hourly rate that the user pays when selecting a planned charging mode in charging plan i;

[0034] The pricing module, with the objective of minimizing total cost, derives the hourly price R for regular charging and planned charging modes based on the predicted probabilities of user charging patterns. nor and R pre And the charging power P over time t tc .

[0035] In another aspect, the present invention also provides a readable computer storage medium storing a computer program that, when executed by a processor, implements the method described above.

[0036] The present invention has the following advantages:

[0037] (1) In terms of grid stability, it helps balance grid load. The SlrpEV charging plan can adjust the charging time and power of electric vehicles according to the real-time demand of the grid, avoiding concentrated charging during peak electricity consumption periods, thereby reducing the pressure on the grid and improving grid stability. It also reduces the impact on grid infrastructure. By rationally allocating the charging load, the instantaneous impact on grid transformers, transmission lines and other equipment during the charging process can be reduced, extending the service life of grid equipment.

[0038] (2) Reduced charging costs due to improved user convenience. Electricity prices are typically lower during off-peak hours, and users can choose to charge during these lower-price periods through the SlrpEV charging plan, thus reducing charging costs. Improved charging efficiency. This plan optimizes the charging process, reduces charging time, and enhances the ease of use of electric vehicles.

[0039] (3) Environmental Protection: Promoting the Consumption of Renewable Energy. The SlrpEV charging program can be combined with renewable energy power generation, scheduling electric vehicle charging during periods of high renewable energy generation to improve the utilization rate of renewable energy, reduce dependence on traditional fossil fuels, and lower carbon emissions. It also reduces energy waste. By rationally planning charging time and power, unnecessary energy waste can be avoided, and energy efficiency can be improved. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the SlrpEV ecosystem in an embodiment of the present invention;

[0042] Figure 2 This is a system structure diagram in an embodiment of the present invention. Detailed Implementation

[0043] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0044] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0045] SlrpEV1 (Smart Learning Electric Vehicle Pilot) is a workplace charging program that provides plug-in charging services to PEV owners at UC Berkeley and UC San Diego. SlrpEV can be viewed as a network-physical-human system (Network-Physical-Human System). Figure 1 The physical components consist of two charging stations with a total of eight electric vehicle power supply units (EVSEs). The network components refer to the backend software and communication systems for charging operations, namely pricing optimization and energy dispatch algorithms. The user component refers to the behavior of PEV users and their charging decisions, including users' energy needs, willingness to delay charging, and plug-in time. Specifically, the program aims to estimate the price thresholds and incentives for users to accept delayed vehicle charging rather than immediate charging, thereby providing charging operators with the flexibility to optimally supply the required energy.

[0046] The relationship between the time users plug in their vehicles and the hourly price they pay was modeled. This knowledge was then used to design a “smart” charger that jointly optimizes pricing strategies and energy delivery schedules while considering the needs of both consumers (PEV owners) and suppliers (charging facility operators). More specifically, the goal is to jointly optimize pricing and electricity to increase net profit and minimize the cost of providing charging services to charging station operators. To achieve this, SlrpEV offers PEV users two charging options: regular charging and scheduled charging.

[0047] Under the "Regular" option, the charging process begins as soon as the vehicle is plugged in, similar to how other standard charging stations operate. On the other hand, scheduled charging requires users to input their required energy (by added mileage) and desired departure time before starting charging. Users are then guaranteed to receive their requested charge by the specified deadline, not earlier. This provides station operators with the flexibility to optimize energy delivery schedules to achieve various goals: minimizing emissions, maximizing revenue and profits, minimizing power fluctuations, etc.

[0048] Currently, another feature that distinguishes SlrpEV from many existing charging schemes is its hourly pricing, rather than kilowatt-hour pricing. This automatically empowers operators to influence charging duration by leveraging user sensitivity to hourly prices. Utilizing this relationship between charging duration and hourly price to formulate optimal pricing policies enables a significant reduction in energy costs and an increase in charging station operator profits without sacrificing user service levels. An embodiment of this invention provides a method for optimizing electric vehicle charging station behavior by incentivizing user behavior through time-based pricing, comprising the following steps:

[0049] Step one: Based on scheduled charging, when the user selects the regular charging mode, the charging system will charge the vehicle at maximum output power from the moment it is connected to the charging device until the vehicle is fully charged or the user drives the vehicle away. This mode is charged according to the hourly billing standard. During this process, the charging system will continue to operate to ensure that the vehicle receives as much electricity as possible.

[0050] Each charging option is linked to the station operator's own energy costs. Under the standard options, charging starts at time t and continues at time t. ec The cost of ending a charging program i (from the perspective of the station operator) can be expressed as:

[0051]

[0052] Where t ec The shortest time required to fully charge a car at maximum power (if the user leaves before the car is fully charged, then t) ec =t nor ). t ec P is the electricity cost at time t, expressed in yuan per kilowatt-hour. maxi R is the maximum charging power that the car can accept during time period i. nori This is the hourly rate (i.e., the operator's revenue) that users pay when they select the standard option.

[0053] Step two, charging based on the plan option, requires users to specify the vehicle's charging duration (i.e., the estimated pick-up time) and the required charging amount (converted to energy values). It's important to note that the vehicle will only receive the pre-set full charge if the user picks up the vehicle after the specified deadline. If the user picks up the vehicle early, it may not reach the expected charge level; conversely, if the user picks up the vehicle late, an overdue penalty will be incurred.

[0054] If the user selects the plan option, they will be asked to specify the required energy E. req And a fixed departure time t pre When they intend to pick up the car. Then, the cost they bring to the operator can be expressed as:

[0055]

[0056] Among them, P tc The charging power at time t is controlled by the site operator, R prei This is the hourly rate, or revenue, paid by the user when selecting the reservation option in charging plan i.

[0057] Step 3: The "Leave" option is available upon completion of charging. If the user decides not to charge due to high charging prices, they can choose the "Leave" mode. In this mode, the user's cost is assumed to be zero. Since the vehicle is not charging, no fees are incurred. If the user begins charging in other modes, the charging system will automatically stop charging upon completion to avoid overcharging, energy waste, and equipment damage.

[0058] Given the costs associated with each alternative, the operator's total expected cost can be expressed as:

[0059]

[0060] Where, p nor p pre and p dep These represent the probabilities that a user selects the regular option, the planned option, or the leave option within charging plan i. Please note that p... nor P pre and P dep The sum of these must be 1. Because of the cost C... Dep The definition refers to the total cost of providing charging services, excluding any opportunity cost, thus assuming that the cost of taking a vacation is zero.

[0061] The goal of a gas station operator is to minimize total costs, as shown in equation (3). To achieve this goal, the operator can control two levers: the hourly price for regular and scheduled options, respectively R nor and R pre And the charging power P over time ttc If the user selects the plan option. On the other hand, these decision variables are subject to constraints that station operators must adhere to, namely:

[0062] Charging power P at time t tc It must be non-negative, and the maximum power that the vehicle in stage i can accept is the upper limit;

[0063] P maxi ≥P tc ≥0 (4)

[0064] For user selection:

[0065] If the user selects scheduling, the total energy delivered to the user in charging plan i must be equal to the total energy E requested at the user-specified deadline. nor ,in:

[0066]

[0067] Where Δt is the system's time step i. The regular and scheduled hourly prices (Ri and Ri respectively) nor and R pre ) is non-negative.

[0068] 0≤R pre ,0≤R nor (6)

[0069] Many parameters depend on user behavior, and the price controlled by the operator is endogenous. That is, the probability of choosing each option, P... pre ,P nor and P tc It all depends on the hourly price for regular and scheduled options.

[0070] In addition, the duration of regular and planned charging, t ec and t pre This will also depend on the hourly rate the user pays. Discounted hourly rates will result in longer plug-in durations. Users can leave their vehicles parked for longer periods while still maintaining a low total charging plan cost.

[0071] Random pricing optimization is a method for optimizing the operation of electric vehicle charging stations, offering two options: regular charging and scheduled charging. Users are given a randomly selected hourly rate, with the price range taking into account competitiveness with other nearby charging stations. Most random prices are below 2 yuan / hour, but occasionally higher prices appear to analyze the relationship between price and plug-in time.

[0072] Regular charging is plug-and-charge and charged by the hour; scheduled charging requires the user to specify the plug-in time and desired charging range, and late fees may be incurred if the vehicle is picked up after the deadline.

[0073] For the choice model, stochastic utility theory is employed, and a binary logit model is used to capture the decision-making process of participants choosing between routine and planned options. Like the binary logit model, stochastic utility models are widely used in the literature for modeling individual choice behavior in the context of electric vehicles. These models are typically developed based on a utility maximization framework—each available alternative has a corresponding utility for each individual, and it is assumed that individuals are rational agents, i.e., they choose the alternative that maximizes their utility. Models developed and estimated using a similar framework are also discussed.

[0074] The model primarily reflects the impact of changing hourly prices and hourly price differences on the probability of choosing each option. It also examines whether these probabilities change with the time of day. Indeed, it's reasonable to expect that people arriving later in the afternoon, compared to those arriving in the early morning, might be more inclined to choose a regular time slot to minimize charging time. Personal characteristics and demographic information are not included in the final model specification. This is because the model will later be used to optimize prices to achieve a certain probability split between regular and planned times. Including individual-level covariates in the model would lead to price discrimination based on personal factors such as income, gender, etc., which must be avoided. Therefore, the final model only includes price and time features, as these are the inputs that will be used in the optimization framework.

[0075] More specifically, utility equations for each alternative are specified for users in charging plan i, as shown in equations (7) and (8):

[0076] U prei =ε prei -k Δp ·(R nori -R prei )+k pm ·I(8pm>t0>2pm) (7)

[0077] U nori =ε nori +OPC nor (8)

[0078] Where R prei R nori These are the hourly rates for the "Planned" and "Regular" options, respectively. Δp This is the coefficient for the difference in hourly rates between the planned and conventional plans. It should be interpreted as the marginal utility of the planned plan increasing by one unit for every unit increase in the hourly rate difference between the planned and conventional plans, where I(t0>2pm) is 1 if the marginal utility is 1 and 0 otherwise.pm It is the coefficient for the aforementioned arrival time index.

[0079] OPC nor It is an optional specific constant (intercept) of the regular options, and all other things being equal, it should be interpreted as an inherent preference for the regular options.

[0080] Here, 8pm>t0>2pm is an indicator function, which states that if the arrival time t0 is between 2 pm and 8 pm.

[0081] Where ε prei and ε nori These are random error terms, assumed to be independent and identically distributed (i,i,d) under both alternative and choice scenarios, with a location parameter of 0 and a scale parameter of 1 as the extreme type I.

[0082] When making a choice, if the utility of n exceeds that of m, then it is assumed that the user chooses to replace n instead of m. The system utility of replacing n with charging plan i. (The probability that a user chooses replacement n with charging plan i) This can be expressed mathematically as follows:

[0083]

[0084] Due to random disturbances and If I is an extreme value of i, i, d, then their difference follows a logistic distribution.

[0085] Given an expression for the individual probability of choosing each option, the likelihood equation of the model can be derived, and the model parameters can be estimated using maximum likelihood estimation. The model's likelihood is the probability predicted by the model of all individuals making the choices observed in real life. This is due to the assumption of random disturbances. and Since the likelihood is independent of the individual and the choice context, it is the product of the individual probabilities across all charging plans.

[0086] L=P(y1)·P(y2)·P(y3)·……·P(y N (10)

[0087] Where y i Observation selection for charging plan i. Model parameters k are estimated by maximizing the log-likelihood of the model. Δp k pm OPC nor :

[0088] max[logP(y1)+logP(y2)+logP(y3)+……+logP(yN )](11)

[0089] This model is able to estimate the number of key factors of interest to users and draw several conclusions about user behavior at charging stations. First, the estimation shows that for a price difference of zero (i.e., when the prices of the two alternatives are the same), the regular option is significantly better than the planned option. In fact, the model allows for the calculation of the discount amount that makes the planned and regular options competitive, where "competitive" is defined as having the same probability. To do this, note that for the two alternatives to have equal probabilities, their system utility to the user must be equal:

[0090]

[0091] In step three, another important characteristic to estimate is the relationship between charging duration and hourly price, as this relationship is key to stochastic pricing optimization. This means that if charging station operators seek to extend plug-in durations to give them greater flexibility in optimizing charging power delivery, then they must offer discounted hourly rates. These empirical results confirm the effectiveness of deadline-based differentiated pricing schemes. Similarly, if charging station operators aim to maximize utilization, increase throughput, and minimize charging time at their stations, then hourly rates should be increased to discourage prolonged charging and incentivize users to pick up their vehicles quickly.

[0092] A kernel regression model is fitted to the data to predict the expected charging plan duration for a given price. Compared to linear regression, kernel regression helps capture non-linearity while maintaining monotonicity and better interpretability properties compared to tree-based and other machine learning methods. The fitted model captures the fact that the marginal sensitivity of plug-in duration to increases in hourly rates decreases as prices rise, eventually leveling off.

[0093] In another embodiment, the present invention also provides a joint pricing system based on electric vehicle charging station behavior optimization, such as... Figure 2 As shown, it includes:

[0094] The inputs, in the simulation, are personal information (arrival time, required energy, and maximum power capacity), and a hypothetical hourly TOU electricity cost. The optimized price is then calculated using the framework described above.

[0095] Optimization is then performed, given these prices, using the behavioral models estimated in steps one and two to simulate the expected behavior—specifically, the choice of alternatives and dwell time. Since these models are validated and demonstrate good predictive ability in similar environments, [further optimization is needed].

[0096] The model reasonably assumes that the behavioral responses it predicts will reasonably replicate what would happen if the framework were implemented in real life. To simulate choosing between conventional and planned alternatives, each alternative (including the departure option) is first computed, and their optimal price is given using a discrete choice model. Then, Monte Carlo simulations are used to model the choice by a multinomial distribution parameterized by these probabilities.

[0097] In another embodiment, the present invention also provides a readable computer storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0098] like Figure 1 As shown, the main objective of the SlrpEV project and the focus of this invention is to understand the charging habits and preferences of PEV drivers in the workplace to improve the performance of charging stations. Specifically, it aims to estimate the price threshold and incentives for users to accept delayed vehicle charging rather than immediate charging, thereby providing charging operators with the flexibility to provide the required energy in the optimal way. The relationship between price and choice probability, and the relationship between hourly price and charging duration, are evaluated through simulation: First, a real-world, uncontrolled baseline of data is collected where charging schedules and power are not managed. Then, using the same arrival patterns and energy demands as a baseline, the optimal price is calculated using the defined optimization framework, and a validated behavioral model is used to simulate user choice behavior.

[0099] The baseline data was passively collected over a period of time, during which the prices of two charging options were fixed, and user charging behavior was observed accordingly. Prices were fixed for all time periods: 1.5 yuan / hour for regular periods and 2.5 yuan / hour for scheduled periods. The regular charge was set to be comparable to two other public charging stations within a 0.5-mile radius, where the charging cost was also 1.5 yuan / hour. Unsurprisingly, the regular charge was chosen for the vast majority of time periods, except for two specific periods, due to the significant discount compared to the scheduled charge. These charging plans were discarded and omitted from the final analysis because the goal of the simulation exercise was to compare the hybrid model (with both regular and scheduled options) with the uncontrolled model (equivalent to the regular option).

[0100] Table 1 summarizes the key characteristics of the baseline data, with specific data shown in the table below.

[0101] Table 1

[0102] Total charging plan 01 Hourly transaction price 1.5 / h Average charging plan duration 3.62h Average delivered energy 13.63 KW / h

[0103] Then, given baseline data, the simulation exercises aim to model the behavior of those identical baseline chargers, if they have been affected by the optimization framework. The simulated baseline behavior is then compared using key performance indicators (kpi) to evaluate the performance of the optimization framework and quantify its potential gains.

[0104] We used Key Performance Indicators (KPIs) to simulate baseline behavior to evaluate the performance of our optimization framework and quantify its potential benefits. The simulation inputs were personal information about the baseline charging plan (arrival time, required energy, and maximum power capacity), along with a hypothetical hourly TOU cost. The optimized price was calculated using the described framework. Given these prices, specifically, alternatives and dwell times were chosen. Since these models are validated and demonstrate good predictive ability in similar environments, it is reasonable to assume that the behavioral responses predicted by the models will reasonably replicate what would happen if the framework were implemented in real-world settings.

[0105] Regarding the charging duration, using the estimated kernel regression, the expected duration for a given hourly price of an alternative is simply calculated. Note that this results in an artificial variance of the expected duration that is smaller than the observed du-ratio because a deterministic conditional expectation function is used to calculate the expected duration. An alternative is to model the duration as a stochastic outcome of a given price, which would make the problem a stochastic optimization problem and significantly complicate the solution algorithm, as the charging duration directly impacts the power optimization of scheduling options and energy delivery planning. This is beyond the scope of this paper, but it is one approach to improving the simulation and a welcome future extension of this work.

[0106] The optimization objective is to minimize total net costs, or, in other words, to maximize the station operator's net revenue. The second and third metrics aim to provide a deeper understanding of the mechanisms by which net revenue is maximized. In fact, net revenue can be increased by increasing total revenue or decreasing total electricity utility costs, or both. Ideally, we want to see an increase in net revenue resulting from shifting energy delivery to off-peak hours, without necessarily increasing the hourly prices paid by users.

[0107] The analysis focuses on the TOU (Time to Operate) program, under which peak pricing occurs between 4 PM and 9 PM, as evidenced by the high costs during this period. From the power plant operator's perspective, this TOU program would encourage the shift of energy load away from the peak energy price period of 4 PM to 9 PM.

[0108] The performance of the framework, as the first performance metric related to total revenue, cost, and generated profit, was assessed by the site operator. The behavior-aware optimization framework resulted in statistically significant improvements: a 16.2% reduction in total costs, a 16.4% increase in total revenue, and a significant 50.2% increase in total profit. The cost reduction and profit increase were the result of a combination of factors: shifting energy delivery from afternoon peak hours to cheaper off-peak hours, achieved through longer time intervals and greater flexibility for the site operator.

[0109] In the optimized framework work, compared with the baseline, energy delivery during super off-peak hours was significantly higher (from 46% to 57%), while energy delivery during peak hours was significantly lower (from 17% to 9%).

[0110] Regarding user behavior, a key finding is that the increase in profits is not a result of increased hourly rates borne by users. In fact, the average hourly rate for charging plans in the simulation was 1.41 yuan / hour, 6% lower than the benchmark hourly rate. Instead, the increase in revenue is a result of a significant increase in average plug-in duration, from the benchmark of 3.62 hours to 5.3 hours under optimized pricing, as well as a decrease in the rate of vacancy due to more discounted prices.

[0111] Overall, these results demonstrate that by using behavioral models to optimize price and power, it is possible to simultaneously increase operator net revenue and reduce user costs.

[0112] This paper evaluates a framework for optimizing the operation of PEV charging stations by leveraging the inefficiencies present in most existing charging stations. The framework collectively optimizes the pricing strategy and power of charging stations and explicitly incorporates the key role of human behavior and decision-making in the successful operation of these stations by explicitly modeling choice behavior and integrating it into the optimization plan. Experiments used to estimate and validate these behavioral models are described: a discrete choice model of choosing between two charging options and a kernel regression model of plug-in duration as a function of hourly price. Simulations then demonstrate how the framework can improve charging station operations by increasing the net revenue of charging station operators and reducing utility costs without impacting user experience or increasing the average cost paid by EV owners.

[0113] A natural extension is to implement this framework in real time and compare the results with simulations. This is not possible in this work due to technological limitations associated with charging infrastructure, which prevents price optimization based on real-time inputs such as required energy and the number of hours in a day. This limitation can be overcome with more flexible control mechanisms. Another methodological contribution that could extend this work is to compare experimental results on the price sensitivity of plug-in duration and willingness to delay charging with results calculated using observational data. As electric vehicle adoption increases, charging station and behavioral data become increasingly readily available; such a comparison, if the results are comparable to experimental benchmarks, would either validate the use of observational data as a method for modeling charging behavior.

[0114] The framework's performance was evaluated through simulation: First, a real-world, uncontrolled baseline of data was collected, where charging plans and power were not managed. Then, using the same arrival patterns and energy demands as a baseline, the optimized price was calculated using the defined optimization framework, and a validated behavioral model was used to simulate user choice behavior.

[0115] For choice models, these are typically developed based on a utility maximization framework—each available alternative has a corresponding utility for each individual, and it is assumed that individuals are rational agents, i.e., they choose the alternative that maximizes their utility. A similar framework was used to develop and estimate the models, considering the impact of varying hourly prices and hourly price differences on the probability of choosing each option. Do these choice probabilities change with the time of day? Indeed, users can reasonably expect that people arriving later in the afternoon might be more inclined to choose a normal time slot to minimize charging time compared to the early morning hours. However, the final model specification did not include individual characteristics and demographic information.

[0116] This model is able to estimate key quantities and draw several conclusions about user behavior at charging stations. First, the estimation shows that for a price difference of zero (i.e., when the prices of the two alternatives are the same), the regular option is significantly better than the planned option. In fact, the model allows us to calculate the discount amount that makes the planned and regular options competitive, where "competitive" is defined as having the same probability.

[0117] Another important characteristic of the estimation is the relationship between charging duration and hourly price, as this relationship is key to optimization objectives. If charging station operators seek to extend plug-in durations to give them greater flexibility in optimizing charging power delivery, then they must offer discounted hourly rates. Similarly, if the goal of charging station operators is to maximize utilization, increase throughput, and minimize charging time at charging stations, then hourly rates should be increased to discourage prolonged charging and incentivize users to pick up their vehicles quickly.

[0118] The framework's effectiveness was evaluated through simulation: We first collected a real-world, uncontrolled baseline of data where charging schedules and power were not managed. Then, we used the same arrival patterns and energy demand as a baseline, with key performance indicators (KPIs) including: 1. Total revenue, costs, and profits generated by station operators; 2. Total energy supply distribution during peak and off-peak hours; and 3. Average hourly price paid by consumers.

[0119] Given baseline data, simulate the behavior of those identical baseline charging plans, if they have been affected by the optimization framework. Then, compare the simulated baseline behavior using key performance indicators (kpi) to evaluate the performance of the optimization framework and quantify its potential benefits.

[0120] We used Key Performance Indicators (KPIs) to simulate baseline behavior to evaluate the performance of our optimization framework and quantify its potential benefits. The simulation inputs were personal information about the baseline session (time of arrival, required energy, and maximum power capacity), along with a hypothetical hourly TOU electricity cost. The optimization price was calculated using the described framework. Given these prices, specifically, alternatives and dwell times were chosen. Since these models are validated and demonstrate good predictive ability in similar environments, it is reasonable to assume that the behavioral responses predicted by the models will reasonably replicate what would happen if the framework were implemented in real-world settings.

[0121] Regarding charging duration, using estimated kernel regression, the expected duration for a given hourly price of an alternative is simply calculated. This results in a smaller artificial variance in the expected duration than the observed du-ratio because a deterministic conditional expectation function is used to calculate the expected duration.

[0122] The optimization objective is to minimize total net costs, or, in other words, to maximize the station operator's net revenue. The second and third metrics aim to provide a deeper understanding of the mechanisms by which net revenue is maximized. In fact, net revenue can be increased by increasing total revenue or decreasing total electricity utility costs, or both. Ideally, we want to see an increase in net revenue resulting from shifting energy delivery to off-peak hours, without necessarily increasing the hourly prices paid by users.

[0123] The performance of the framework, as the first performance metric related to total revenue, cost, and generated profit, was determined by the station operator. The behavior-aware optimization framework resulted in statistically significant improvements; the cost reduction and profit increase were the result of a combination of factors: shifting energy delivery from afternoon peak hours to cheaper off-peak hours, achieved through longer time intervals and greater flexibility for station operators.

[0124] The analysis showed that peak pricing occurred between 4 p.m. and 9 p.m.

[0125] From the perspective of power station operators, this TOU (Time to Use) program will encourage the shift of energy load away from the peak energy price period of 4 PM to 9 PM. In the optimized implementation, compared to the baseline, off-peak energy delivery increased significantly by 11%, while peak-peak energy delivery decreased significantly by 8%. Therefore, the operator's pricing should peak between 7 PM and 9 PM, with an hourly charging fee of 1.41 yuan.

[0126] Regarding user behavior, a key finding is that the average hourly rate for charging plans in the simulation was 1.41 yuan / hour, significantly lower than the baseline hourly rate. Conversely, the increase in revenue resulted from a significant increase in average plug-in duration, from the baseline of 3.62 hours to 5.3 hours under optimized pricing, which greatly increases the probability that individual users interested in using the plan options will opt for the charging service.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A joint pricing method based on electric vehicle charging station behavior optimization, characterized in that, include: The system obtains users' historical charging behavior information and assumed time-of-use electricity prices. The users' historical charging behavior information includes: arrival time at the charging station, required energy, and maximum power capacity. The probability of a user selecting a charging mode is predicted based on the user's historical charging behavior information; the charging modes include: regular charging mode, planned charging mode, and off-duty mode. Based on the cost of each charging mode and the probability of a user choosing that charging mode, the operator's total expected cost is obtained; With the goal of minimizing total cost, the hourly price and charging power for regular charging mode and planned charging mode are obtained based on the predicted probability of user charging modes.

2. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, The operator's total expected cost is expressed as: Where, p nor p pre and p dep These represent the probabilities that a user selects the regular charging mode, the planned charging mode, or leaves the charging mode within charging plan i; C Dep This refers to the total cost of providing charging services, excluding any opportunity costs, therefore assuming the cost of taking a vacation is zero; R nori This is the hourly rate paid by the user when selecting the regular charging mode; P maxi It is the maximum charging power that the car can accept during time period i; t ec The shortest time required to fully charge a car at maximum power, t, is calculated if the user leaves before the car is fully charged. ec =t nor , t ec P is the electricity cost at time t, expressed in yuan per kilowatt-hour; tc R is the charging power at time t; prei It is the hourly rate that users pay when selecting a planned charging mode in charging plan i.

3. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, In normal charging mode, the charging system charges the vehicle at maximum output power from the moment the vehicle is connected to the charging device until the vehicle is fully charged or the user drives the vehicle away, at which point the charging system will automatically stop charging; the normal charging mode is charged according to the hourly billing standard. In planned charging mode, the vehicle will be charged within the time specified by the user for vehicle access to charging until the set charging amount is reached or the user drives the vehicle away, at which point the charging system will automatically stop charging; if the user picks up the vehicle early, the vehicle may not reach the expected charging level; and if the user picks up the vehicle late, an overtime penalty will be incurred. If a user decides not to charge due to high charging prices, they can choose the "Away" mode. In this mode, the user's cost is assumed to be zero, and no fees will be incurred when the vehicle is not charging.

4. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, In normal charging mode, from time t0 to time t nor Start at time t ec The end of the charging plan has a cost: The cost includes the maximum acceptable charging power for different time periods, which means that the cost calculation will vary according to the actual charging power, rather than simply assuming a constant power. Furthermore, the hourly rate paid by users when choosing the regular option is directly related to the operator's revenue and costs, which helps operators to formulate charging plans and pricing strategies.

5. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, In planned charging mode, given the required energy E req And a fixed departure time t pre The cost to the operator when they intend to pick up the car is expressed as follows: The clear distinction between charging costs and user-paid rates helps operators conduct more accurate charging plan analysis; it helps operators predict and control risks arising from uncertain user behavior, such as users picking up their vehicles earlier or later; and it allows operators to adjust costs based on users' actual charging needs and behaviors, thereby improving user satisfaction.

6. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, Also includes: Based on the hourly price R for regular charging mode and planned charging mode. nor and R pre And the charging power P over time t tc Determine the timing for peak pricing.

7. A joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, Also includes: Estimate the relationship between charging duration and hourly price; With the goal of extending plug-in duration, we optimize charging power delivery and offer discounted hourly rates.

8. The joint pricing method based on electric vehicle charging station behavior optimization according to claim 1, characterized in that, Also includes: Estimate the relationship between charging duration and hourly price; With the goal of maximizing utilization, increasing throughput, and minimizing charging time at charging stations, the hourly rate is increased to discourage prolonged charging and incentivize users to pick up their vehicles quickly.

9. A joint pricing system based on electric vehicle charging station behavior optimization, characterized in that, include: The historical information acquisition module is used to acquire users' historical charging behavior information and assumed time-of-use electricity prices. The users' historical charging behavior information includes: arrival time at the charging station, required energy, and maximum power capacity. The charging mode prediction module is used to predict the probability of a user selecting a charging mode based on the user's historical charging behavior information; the charging modes include: regular charging mode, planned charging mode, and away mode. The cost construction module is used to obtain the operator's total expected cost based on the cost of each charging mode, expressed as: Where, p nor p pre and p dep These represent the probabilities that a user selects the regular charging mode, the planned charging mode, or leaves the charging mode within charging plan i; C Dep This refers to the total cost of providing charging services, excluding any opportunity costs, therefore assuming the cost of taking a vacation is zero; R nori This is the hourly rate paid by the user when selecting the regular charging mode; P maxi It is the maximum charging power that the car can accept during time period i; t ec The shortest time required to fully charge a car at maximum power, t, is calculated if the user leaves before the car is fully charged. ec =t nor , t ec P is the electricity cost at time t, expressed in yuan per kilowatt-hour; tc R is the charging power at time t; prei It is the hourly rate that the user pays when selecting a planned charging mode in charging plan i; The pricing module, with the objective of minimizing total cost, derives the hourly price R for regular charging and planned charging modes based on the predicted probabilities of user charging patterns. nor and R pre And the charging power P over time t tc .

10. A readable computer storage medium storing a computer program, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-8.

Images