Optimal control method for integrated photovoltaic storage and charging based on user side load response

Abstract

This invention relates to the field of optimization control technology for smart grids and electric vehicle charging infrastructure. It addresses the problem that existing technologies cannot extract elastic resources with dynamic response potential from the micro-load side and establish a real-time feedback mechanism between these resources and macro-scheduling tasks. Specifically, it is an integrated photovoltaic-storage-charging optimization control method based on user-side load response, comprising: acquiring real-time status data of each charging terminal within the charging station; dividing the charging terminals into a high-elasticity load set and a low-elasticity load set based on the departure probability coefficient; identifying charging terminals whose adjustment cost indicators meet the adjustment cost threshold as quasi-elastic loads and including them in the adjustment resource pool; and constructing a departure probability evaluation model based on sliding integral and normalization processing by integrating multi-dimensional real-time data such as vehicle locking status, rearview mirror folding status, and in-vehicle activity intensity, ensuring the authenticity and availability of load status in the adjustment resource pool.

Description

Technical Field

[0001] This invention belongs to the field of optimization control technology for smart grids and electric vehicle charging infrastructure, and mainly studies the method of integrated photovoltaic, energy storage and charging optimization control under user-side load response. Background Technology

[0002] The photovoltaic-storage-charging integrated system refers to a platform that organically integrates photovoltaics, energy storage, and charging piles. It is used to assist the power grid in balancing and regulating between peak and off-peak electricity demand periods. Whether the charging station can flexibly adjust its power directly affects the response speed of the power grid commands.

[0003] Most existing mainstream methods perform load forecasting from the grid side, setting fixed power limits for charging piles to complete task scheduling. The dispatch center issues instructions based on pre-set power curves, and each charging pile then allocates power according to actual charging demand. However, this model has an implicit assumption that each charging pile's charging and discharging order will be executed as agreed. User behavior is not entirely controllable, and the actual adjustable power of the charging terminal is limited by the vehicle's immediate intention to leave. If the user leaves early during charging due to unforeseen circumstances, the terminal's adjustment capability will instantly fail. Because the existing scheduling logic does not perceive the user's micro-behavioral intentions in real time, it cannot establish a dynamic connection between macro-scheduling tasks and micro-load responses. As a result, after the dispatch instructions are issued, some terminals cannot respond due to the user's departure, leading to the failure of the adjustment task.

[0004] Therefore, how to discover elastic resources with dynamic response capabilities from the micro-load side and establish an instant feedback connection between them and macro-scheduling tasks has become the main bottleneck in improving the reliability of system scheduling. Summary of the Invention

[0005] The purpose of this invention is to provide an integrated photovoltaic, energy storage, and charging optimization control method based on user-side load response, which solves the problem that existing technologies cannot extract elastic resources with dynamic response potential from the micro-load side and establish a real-time feedback mechanism between them and macro-scheduling tasks.

[0006] The technical problem to be solved by this invention is: how to provide an integrated optical storage and charging optimization control method based on user-side load response that can mine elastic resources with dynamic response potential from the micro-load side and establish a real-time feedback mechanism between them and macro-scheduling tasks.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] An integrated photovoltaic-storage-charging optimization control method based on user-side load response includes:

[0009] Acquire real-time status data of each charging terminal in the charging station. The real-time status data includes vehicle locking status, rearview mirror folding status, and in-vehicle activity intensity characteristics.

[0010] The real-time status data is processed using a preset departure probability assessment model to generate departure probability coefficients for each charging terminal; based on the departure probability coefficients, the charging terminals are divided into a high-elasticity load set and a low-elasticity load set.

[0011] Obtain the task urgency index issued by the macro scheduling center, and dynamically adjust the preset adjustment cost threshold based on the task urgency index;

[0012] Obtain the historical charging records of each charging terminal in the low-elasticity load set, calculate the adjustment cost index of each charging terminal, and the adjustment cost index characterizes the product of the response delay of power regulation and the historical fluctuation variance.

[0013] Charging terminals whose adjustment cost indicators meet the adjustment cost threshold are identified as quasi-elastic loads and included in the adjustment resource pool.

[0014] Calculate the total adjustment capacity of the highly elastic load set and the quasi-elastic load set, and calculate the elasticity gap based on the difference between the total adjustment capacity and the macro-scheduling task requirements;

[0015] If the elasticity gap is greater than zero, calculate the regional group stay tendency index, and feed the elasticity gap and the regional group stay tendency index back to the macro dispatch center so that the macro dispatch center can reconstruct the regional dispatch strategy.

[0016] The present invention has the following beneficial effects:

[0017] 1. By integrating a large amount of real-time data such as vehicle locking status, rearview mirror folding status, and activity intensity inside the vehicle, an evaluation model was created to determine the probability of leaving the site through sliding integral and normalization. This model effectively overcomes the shortcomings of using only electrical parameters for evaluation and focuses more on calculating the probability of the vehicle owner leaving early, ensuring that the load situation in the adjustment resource pool is real, reliable, and usable.

[0018] 2. After adding the macro-schedule task urgency index, a mapping relationship is obtained in which the adjustment cost threshold is monotonically increasing. The system will autonomously adjust the selection criteria of quasi-elastic loads based on the real-time pressure on the power grid side. When the task urgency is high, the entry threshold will be strengthened to maintain the response quality, and when the task urgency is low, the conditions will be relaxed to increase the adjustment range, thereby achieving the self-adjustment of the scheduling strategy.

[0019] 3. By calculating the regional group dwell tendency index, the overall dwell tendency of the load within the station is obtained. This is used as the main feedback parameter for reconstructing the regional dispatch strategy, enabling the macro dispatch center to carry out differentiated task decomposition and allocation based on the characteristics of group behavior at each station. This effectively solves the problem of insufficient regulation capacity of local charging stations and improves the overall peak shaving and valley filling effect of the regional power grid.

[0020] 4. After completing the macro-scheduling task, the deviation between the actual departure status and the predicted probability is compared. The least squares method is used to adaptively correct the preset weight coefficients in the evaluation model, so that the system can continuously optimize the departure probability calculation logic according to the different user behavior habits of charging stations, which greatly reduces the prediction deviation during long-term operation and improves the stability and intelligence level of the system. Attached Figure Description

[0021] 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 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.

[0022] Figure 1 This is the main flowchart of the method in Embodiment 1 of the present invention;

[0023] Figure 2 This is a sub-flowchart of step S6 in Embodiment 1 of the present invention. Detailed Implementation

[0024] The technical solution of the present invention will be further described in detail below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0025] In traditional integrated photovoltaic, energy storage, and charging scheduling, the load response characteristics of each charging terminal in the charging station are regarded as an important means of peak shaving and valley filling in the power grid. Whether its regulation capacity can be effectively released, in essence, is to predict the user's intention to leave the grid and dynamically match the predicted departure probability coefficient with the regulation resources in the power grid. That is, to convert specific physical states into continuous departure probability coefficients, and to achieve precise selection and allocation of regulation resource pool at the macro scheduling level.

[0026] Existing technologies lack a verification mechanism for the dynamic coupling between the real-time physical state of charging terminals and the urgency of macro-level scheduling tasks, making it impossible to accurately determine the actual adjustment potential of charging terminals in complex scheduling scenarios. Specifically, misjudgment of departure intentions occurs because, although the system detects the vehicle's locked state, it ignores the nonlinear impact of the intensity of occupant activity on the departure probability. Rigid assessment of adjustment costs arises because the system fails to promptly adjust the screening threshold for quasi-elastic loads when the urgency of macro-level tasks changes, resulting in a mismatch between the size of the adjustment resource pool and the actual needs of the power grid. Therefore, a strict response correspondence cannot be established between the physical state of charging terminals and the power grid's scheduling strategy, leading to misjudgments or delayed feedback regarding the availability of adjustment resources, thus affecting the overall optimization and control effect of the integrated photovoltaic-storage-charging system.

[0027] If the above problems are not resolved, the ability of the optimized control system to objectively judge the load response status will continue to be lost, and inaccurate dispatch feedback will systematically hinder the charging station from mastering the core skills of load-side response, affecting the achievement of the goal of stable operation of the photovoltaic-storage-charging integrated system in complex power grid environments.

[0028] Example 1: As Figure 1-2 As shown, the integrated photovoltaic-storage-charging optimization control method based on user-side load response includes:

[0029] Step S1: Obtain real-time status data of each charging terminal in the charging station. The real-time status data includes vehicle locking status, rearview mirror folding status, and in-vehicle activity intensity characteristics.

[0030] In the integrated photovoltaic-storage-charging optimization control method based on user-side load response, obtaining the real-time status data of each charging terminal in the charging station is a prerequisite for subsequent load classification and scheduling decisions. To ensure that this step has high feasibility and technical rigor, this paper provides a detailed analysis of the acquisition, parsing and storage process of real-time status data.

[0031] Before acquiring data, the system first establishes a "charging terminal status perception and mapping architecture." This architecture is not a simple sensor data reading, but a distributed data processing model based on edge computing nodes. Each charging terminal is equipped with a CAN bus communication interface module and a high-precision infrared / ultrasonic sensor array. The reason for adopting such a dual-modal perception structure is that a single electrical parameter cannot fully reflect the actual parking status of the vehicle and user behavior. By combining the "vehicle locking status" at the electrical level with the "intensity of in-vehicle activity" at the physical level, the vehicle's intention to leave can be better characterized. The CAN bus interface uses a preset protocol parser to listen to message frames in the vehicle control area network and extract status bit identifiers. The infrared or ultrasonic sensors emit ultrasonic pulses at a frequency of 40kHz, and the received echo time difference and Doppler frequency shift are used to calculate the minute displacement changes in the in-vehicle space, thereby obtaining the intensity of in-vehicle activity.

[0032] When the system is running, it first obtains messages from the vehicle control local area network (CAN) bus through the communication interface of the charging terminal. The system has been configured with a parsing mapping table for mainstream electric vehicle protocols to accurately parse the vehicle lock status bit and the rearview mirror folding status bit according to the message ID. The vehicle lock status bit is represented by a Boolean value. When the message parsing result is "1", it means that the vehicle is in the locked state. The rearview mirror folding status bit is also represented by a Boolean value. When the parsing result is "1", it means that the rearview mirror is in the folding state. At the same time, the infrared or ultrasonic sensors deployed on the charging terminal also start working and begin to collect the intensity of activities inside the vehicle. This value is defined as the amplitude of the spatial reflected wave energy fluctuation detected by the sensor within a preset sampling period (e.g., 100 milliseconds). Its value range is normalized to [0, 1]. The larger the value, the more frequent the activities of the people inside the vehicle.

[0033] To ensure the stability of the data source, the system needs to perform "multi-level filtering and anomaly removal" before executing step S2. For example, median filtering is used to remove spike noise caused by sudden changes in ambient light from the infrared sensor. For CAN bus messages, if the message verification fails for three consecutive sampling cycles, the terminal's status is marked as communication abnormal, and its departure probability coefficient is forcibly set to 0.5 (neutral value) to avoid misjudgment of the scheduling strategy due to abnormal data. For newly connected terminals, a cold start strategy is adopted. In the absence of historical charging records, a preset baseline adjustment cost index is given to it. After collecting data for at least 5 charging cycles, the system switches to a method that calculates based on the actual fluctuation variance.

[0034] Step S2: Process the real-time status data according to the preset departure probability assessment model to obtain the departure probability coefficient corresponding to each charging terminal; divide the charging terminals into a high-elasticity load set and a low-elasticity load set according to the departure probability coefficient.

[0035] After completing the real-time status data acquisition and storage, the system proceeds to step S2, where the real-time status data is processed using a pre-set departure probability assessment model to obtain the departure probability coefficient of each charging terminal, thereby completing the division of the load set.

[0036] This embodiment first establishes a departure probability prediction model based on multi-dimensional feature fusion. The basic idea of ​​this model is to transform discrete vehicle states into a continuous departure probability space. The input layer of the model has three main features, namely the vehicle locking state bit. ∈{0,1}, rearview mirror folding status position ∈{0,1} and in-vehicle activity intensity values ∈[0,1]; The model uses the sliding integral algorithm to extract the temporal features of the activity intensity, and then uses the preset weight coefficients to weight and sum them; Compared with instantaneous values, the sliding integral is more suitable for the temporal continuity of in-vehicle activities. Instantaneous fluctuations are easily interfered with by sensor noise. The sliding integral can smooth the noise and capture the behavioral trend of the user preparing to leave.

[0037] The intensity of in-vehicle activities is measured within a preset time window. (For example, set to 10 minutes) Perform sliding integral processing to obtain the cumulative activity amount. ;set up express The integral formula is as follows: (The value of the in-vehicle activity intensity at any given time is within the range [0,1]). ;

[0038] in, Let t be the integral variable within the time window, and t be the current time. In the discretized implementation, this integral is transformed into a summation operation: Where N is the total number of sampling points within the window, Let i be the in-vehicle activity intensity value at the i-th sampling point. The sampling interval is denoted as .

[0039] After obtaining the cumulative activity amount Subsequently, in order to eliminate the differences in measurement ranges of different sensors and ensure that the input features are within the [0,1] interval, the system performs normalization processing: the system maintains a historical activity intensity statistics database within a sliding time window, recording the historical maximum values ​​within this window. and minimum value Normalized cumulative activity The calculation formula is: ;in, For a very small constant (e.g., 10) −6 This is used to prevent calculation overflow caused by a denominator of zero; if the calculated... If the value exceeds the [0,1] interval, the system performs truncation. This process ensures that the influence of activity intensity characteristics on the exit probability coefficient is controllable and stable during subsequent weighted summation.

[0040] The system will lock the vehicle. Rearview mirror folding state and cumulative activity amount Multiply by the corresponding preset weighting coefficients respectively Then, the product terms are summed to obtain the exit probability coefficient. The calculation formula is as follows: Weighting coefficients It is calculated using the Analytic Hierarchy Process (AHP) or regression analysis based on historical departure samples, taking into account the weights of the contributions of vehicle locking, rearview mirror folding, and in-vehicle activity intensity to departure behavior.

[0041] Weighting coefficient It satisfies the normalization condition (the sum of the three equals 1), and is set to an empirical value. For example, if a vehicle is in a locked state ( =1), Rearview mirror folding ( =1), and the normalized cumulative activity within the window is If the value is 0.2, then the departure probability coefficient of this terminal is... =0.4×1+0.3×1+0.3×0.2=0.76.

[0042] After obtaining the exit probability coefficient Then, the system performs load set partitioning; the system presets a baseline threshold for the probability of leaving the site. (For example, set to 0.5), the baseline threshold for the probability of leaving the market. =0.5 is determined by statistical analysis of historical vehicle departure behavior at charging stations, based on the quartiles or median of the departure probability distribution, to ensure that the coverage of the selected high-elasticity load set reaches the preset ratio; the system will use the departure probability coefficient of each charging terminal. and Compare: If If the load corresponding to the charging terminal has a high tendency to leave the site, it is determined that it is classified into a set of highly elastic loads; if If the load detachment intention of the charging terminal is low, it is determined that the terminal is classified as a low-elasticity load set. Through this dynamic classification mechanism based on probability coefficients, the system can identify in real time which charging terminals are more suitable to participate in subsequent power regulation, thereby providing a precise regulation resource pool for the macro-dispatch center.

[0043] Step S3: Based on the task urgency index issued by the scheduling center, dynamically adjust the pre-set adjustment cost threshold to keep it consistent with the task urgency index.

[0044] After the load is allocated in step S2, step S3 is carried out, which is the real-time coordination of local charging station adjustment strategies and macro-grid dispatching requirements.

[0045] This embodiment first establishes a mapping model between task urgency and adjustment cost; the main idea of ​​this model is to establish a monotonically increasing relationship between macro-scheduling pressure and micro-adjustment threshold; the input of the model is the task urgency index (U∈[0,1]) issued by the macro-scheduling center, that is, the higher the task urgency, the larger the U value, and the output of the model is the dynamically adjusted adjustment cost threshold (Cth), that is, according to different macro-scheduling tasks, an appropriate admission standard is selected (Cth). To select charging terminals with good adjustment performance and small historical fluctuations, a monotonically increasing mapping is used.

[0046] The system receives a task urgency index U from the macro-dispatch center via a communication interface. This index is calculated by the macro-dispatch center based on the regional power grid's load gap, frequency deviation, and peak shaving and valley filling requirements. U is then input into a preset monotonically increasing mapping function. In the process, the adjustment cost threshold is calculated. This embodiment uses a linear mapping function as the basic implementation, and its calculation formula is as follows: ;

[0047] in, The preset adjustment cost threshold represents the minimum admission standard when the task urgency is zero; k is the adjustment sensitivity coefficient, used to control the degree of influence of task urgency on the threshold; when a serious power grid fault occurs and the task urgency U approaches 1, the system adjusts the adjustment cost threshold using the Sigmoid function. Automatically increase to the limit This allows for the selection of only high-quality core loads with extremely low response delays and minimal fluctuations, ensuring grid frequency stability.

[0048] To avoid excessively high thresholds leading to resource pool depletion under extreme urgency, the system can also use the Sigmoid function for mapping; its calculation formula is as follows: ;

[0049] in, and These are the lower and upper limits of the adjustment cost threshold, respectively. α represents the median point of urgency, and α is the steepness parameter of the mapping curve.

[0050] The sensitivity coefficient k and the kurtosis parameter α of the mapping curve were obtained through offline simulation experiments. The system was stress tested under different power grid load gap scenarios. With the optimization objectives of adjusting the response success rate and resource utilization rate, k and α were optimized and calibrated using a genetic algorithm or gradient descent method.

[0051] For example, suppose the system sets a baseline threshold. =0.2, sensitivity coefficient k=0.5; when the task urgency index U=0.4 issued by the macro-dispatch center, the system calculates the adjustment cost threshold. =0.2 + 0.5 × 0.4 = 0.4; If the urgency of subsequent tasks increases to U = 0.8, the adjustment cost threshold will be adjusted accordingly. =0.2+0.5×0.8=0.6; Through this dynamic adjustment mechanism, the system can automatically tighten or relax the screening criteria for aligning with flexible loads according to the real-time changes in macro-dispatch tasks, thereby maximizing the utilization of regulation resources within the charging station while ensuring the grid dispatch needs; This process is completely executed automatically by the system without human intervention, ensuring the real-time performance and efficiency of the optimized control strategy.

[0052] Step S4: Obtain the historical charging records of each charging terminal in the low-elasticity load set, and calculate the adjustment cost index of each charging terminal. The adjustment cost index represents the product of the response delay of power regulation and the historical fluctuation variance.

[0053] To ensure the objectivity and traceability of the calculation of the adjustment cost index, this embodiment naturally and reasonably establishes a "charging terminal adjustment performance evaluation model" in advance. The core logic of this model is to comprehensively evaluate the fluctuation characteristics during power adjustment by analyzing the timing deviation between power commands and actual current responses during historical charging processes, thereby evaluating the adjustment cost of the terminal. The model input is the historical charging records of each charging terminal, i.e., the power command sequence. Actual current response sequence The timestamp sequence {t}; the output is the adjustment cost index (denoted as ). The reason for choosing the product of response delay and historical variance as the evaluation index is that response delay can reflect the real-time response capability of the terminal, and variance can reflect the output stability during the adjustment process of the terminal. The product of the two can better reflect the overall cost of the terminal participating in power regulation.

[0054] Extract the power command issuance time from historical charging records. With actual current response time The difference between the two is used to calculate the response delay. ,Right now: ;in, The unit is seconds.

[0055] Extracting power regulation data sequences from historical charging processes Calculate the volatility variance of the sequence. The calculation formula is as follows: ;

[0056] in, The arithmetic mean of the power regulation data, i.e. , where n is the total number of sample points in the historical records.

[0057] Response delay With volatility and variance After multiplication, the original adjustment cost index of the charging terminal is obtained. : To avoid the influence of different dimensions on the adjustment cost index, the system calculates the original adjustment cost index. Perform normalization processing; pre-calculate the maximum response latency of all terminals in the historical charging records. and maximum volatility variance Then the normalized adjustment cost index The calculation is as follows: ;in, and It can be updated periodically (e.g., daily) to ensure that the normalization range adapts to actual operating data; this normalization method will Compressing to the dimensionless interval [0, 1] reduces the adjustment cost threshold. The setting and adjustment of the urgency mapping function are consistent.

[0058] For example, suppose a charging terminal issues a power command at 10:00:00 during a historical adjustment task, and the actual current response time is 10:00:05. Then the response delay is... =5 seconds; if the power regulation data fluctuation variance of the terminal during the adjustment process =0.08kW², then the initial adjustment cost index of this terminal is... =5×0.08=0.4; Through this calculation method, the system can assign a quantified adjustment cost index to each charging terminal in the low-elasticity load set; the smaller the index value, the faster the terminal responds and the more stable the output during power adjustment, that is, the lower the adjustment cost, thus providing a clear basis for identifying it as a quasi-elastic load and including it in the adjustment resource pool in subsequent steps; this calculation process is entirely based on historical data, ensuring the objectivity and accuracy of the evaluation results.

[0059] Step S5: Divide the charging terminals into quasi-elastic loads according to the adjustment cost threshold and add them to the adjustment resource pool.

[0060] This embodiment first establishes a "quasi-elastic load determination logic engine". The main logic of this engine is to map and match the adjustment cost index and the dynamic threshold by numerical comparison. In real-time scheduling, the system needs high computational efficiency, and the adjustment cost index itself has a clear physical meaning. Direct comparison can ensure the real-time performance and interpretability of the screening results. Therefore, numerical comparison is used instead of complex classification algorithms.

[0061] In the specific implementation steps, the system first obtains the adjustment cost index of each charging terminal calculated in step S4. And the adjustment cost threshold dynamically adjusted according to the urgency of the macro-scheduled tasks in step S3. The system iterates through each charging terminal in the set of low-elasticity loads and executes the following decision algorithm:

[0062] If the conditions are met If the charging terminal meets the adjustment conditions, the system will identify it as a quasi-elastic load and store its corresponding terminal ID and current adjustable power parameters in the adjustment resource pool.

[0063] If the conditions are met If the adjustment cost of the charging terminal is too high and does not meet the adjustment conditions, the system will exclude it from the adjustment resource pool and maintain its original charging state.

[0064] The establishment of the adjustment resource pool is not a static process. After completing the above judgment, the system will immediately update the index list of the adjustment resource pool and record the current adjustable power limit of each quasi-elastic load for use in the subsequent step S6 to calculate the total adjustment capacity. A dynamic threshold screening method is adopted to ensure that the terminals in the adjustment resource pool are always within a controllable adjustment cost range. While meeting the macro scheduling requirements, the adjustment cost and risk of local charging stations are minimized. The judgment logic is carried out entirely according to the pre-set numerical comparison rules to ensure that the screening process is objective and the system operates stably.

[0065] Step S6: Calculate the total regulation capacity of the high-elasticity load set and the quasi-elasticity load set, and use the difference between the total regulation capacity and the macro-scheduling task demand to calculate the elasticity gap;

[0066] If the elasticity gap is greater than 0, the regional group stay tendency index is calculated, and the elasticity gap and the regional group stay tendency index are fed back to the macro-dispatch center, which then reconstructs the regional dispatch strategy.

[0067] After screening quasi-elastic loads and creating the adjustment resource pool, the system proceeds to step S6, which calculates the total adjustment capacity of the high-elasticity load set and the quasi-elasticity load set. The elasticity gap is calculated based on the difference between the total adjustment capacity and the macro-scheduling task requirements, and then the regional scheduling strategy is fed back based on the elasticity gap.

[0068] To ensure the accuracy of the total regulation capacity calculation and the scientific nature of the elasticity gap evaluation, this embodiment first creates a global load regulation capacity evaluation model. The main idea of ​​this model is to sum the different load attribute sets according to the power dimension and then perform a difference mapping with the macro-scheduling task requirements. The reason for adopting this hierarchical aggregation method is that highly elastic loads and quasi-elastic loads have different response characteristics. Therefore, we obtain their current maximum adjustable power separately and then add them together to better reflect the power regulation potential of the entire charging station.

[0069] In the specific implementation steps, the total regulation capacity (denoted as...) is calculated. The system obtains the current adjustable power limit of each charging terminal in the highly elastic load set. And the current adjustable power limit of each charging terminal in the quasi-elastic load set. The formula for calculating total regulating capacity is: ;

[0070] Where m is the number of terminals in the highly elastic load set, and n is the number of terminals in the quasi-elastic load set.

[0071] Obtain the power requirement of the task issued by the macro scheduling center (denoted as ). ), and calculate the elastic gap (denoted as ). The calculation formula is as follows: ;

[0072] This indicator represents the power difference between the current charging station's adjustment capacity and the macro-scheduling task requirements.

[0073] In terms of decision-making logic, the system performs the following processing: If A value greater than 0 indicates that the current station's adjustment capacity is insufficient to meet the demands of the macro-scheduling task, and the system then triggers the calculation of the regional group dwell tendency index; this index (denoted as...) This value is used to characterize the overall dwell tendency of the load within the station. A larger value indicates that users are more likely to stay (lower probability of leaving), while a smaller value indicates that users are more likely to leave (higher probability of leaving). The calculation process is as follows: obtain the departure probability coefficient of all charging terminals within the station. Calculate the complement of each exit probability coefficient. And perform an arithmetic mean on all the complements: ;

[0074] Where C represents the total number of charging terminals within the station.

[0075] For example, assuming the total regulating capacity of a highly flexible load set is 50kW and the total regulating capacity of a quasi-flexible load set is 30kW, then... =80kW; if the macro-scheduling task requires... =100kW, then the elastic gap =100-80=20kW; due to >0, the system calculates the regional population stay tendency index. If calculated =0.25, the system will =20kW and =0.25 is fed back to the macro-dispatch center; based on this, the macro-dispatch center reconstructs the regional dispatch strategy, breaks down the peak shaving and valley filling task requirements of the station, and allocates them to other charging stations in the region; through this closed-loop feedback mechanism, the system can ensure the stable operation of the power grid through macro-level resource coordination when local adjustment capabilities are limited; the process has clear logic and well-defined calculation parameters, ensuring the effective execution of the optimized control strategy in complex dispatch scenarios.

[0076] After completing the macro-scheduling task, the system adaptively corrects the departure probability assessment model. The system compares the actual departure status data (denoted as Y∈{0,1}, where 1 represents departure and 0 represents not leaving) within a preset time period (in this embodiment, the preset time period is set to 30 seconds; however, those skilled in the art will understand that this time period can be adjusted within the range of 1 to 60 seconds based on factors such as system communication delay and energy storage response speed) with the predicted departure probability coefficient PoutPout to calculate the prediction deviation value. The system uses the least squares method to adjust the weighting coefficients. Perform iterative updates, with the update formula as follows: ,in The learning rate (value 0.01). The corresponding input feature values ​​are used; through this iterative process, the model can automatically adjust the weight allocation according to the actual user behavior characteristics of different charging stations, ensuring that the accuracy of the departure probability assessment is continuously optimized over time.

[0077] After completing the feedback of the elastic gap, the system will execute an adaptive adjustment strategy based on the changing trend of the regional group stay tendency index, that is, dynamically change the adjustment cost threshold to achieve fine control over the number of elastic loads to be screened, thereby improving the system's response flexibility while ensuring the scheduling task.

[0078] The system obtains the trend of regional population dwell tendency index over a set time series (e.g., the past 30 minutes), and then dynamically adjusts the adjustment cost threshold based on this trend.

[0079] when When the trend is downward, it indicates that the overall willingness of users to leave the site is increasing, and the available adjustment resources may decrease. The system then performs processing to compress the adjustment cost threshold, that is, it sets... (in (Using a compression factor, for example, 0.9) to reduce the number of quasi-elastic loads to screen, thus avoiding including terminals that are about to leave the site in the resource pool.

[0080] When the trend is upward, it indicates that the overall willingness of users to stay on the site is increasing, and more terminals have long-term adjustment potential; the system executes the process of expanding the adjustment cost threshold, that is, it makes... (where β > 1 is the expansion coefficient, for example, 1.1); to increase the number of quasi-elastic loads to be screened and expand the adjustment resource pool.

[0081] After performing the above dynamic adjustments, the system adjusts the cost threshold accordingly. The system re-screens the set of low-elasticity loads; it iterates through the set of low-elasticity loads and adjusts the cost index. The charging terminals are identified as the updated quasi-elastic load set; subsequently, the system recalculates the updated total regulation capacity. : ;

[0082] in, This is the sum of the adjustable power limits of each charging terminal in the updated quasi-elastic load set.

[0083] Based on the updated overall adjustment capacity and macro-scheduling task requirements The difference is used to recalculate the elastic gap. The updated elasticity gap is fed back to the macro-dispatch center. Through this adaptive closed loop, the system can optimize and adjust the size and quality of the resource pool in real time according to the dynamic behavior characteristics of the load in the region, ensuring that the optimal load response performance is always maintained in the complex and ever-changing power grid dispatch environment. This process realizes a complete closed loop from indicator monitoring, trend judgment, threshold adjustment to resource reconfiguration, which significantly improves the stability of the photovoltaic-storage-charging integrated system.

[0084] This invention establishes a multi-dimensional state perception and dynamic threshold adaptive feedback closed loop to mathematically map the physical departure intention of the charging terminal with the scheduling urgency of the macro power grid, thus solving the problems of load response lag and coarse resource selection in traditional scheduling strategies. The system uses sliding integral and normalization processing to obtain in-vehicle activity characteristics and uses CAN bus status bits to quantify the departure probability, thereby achieving precise stratification of highly elastic and quasi-elastic loads.

[0085] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0086] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0087] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A photovoltaic-storage-charging integrated optimization control method based on user-side load response, characterized in that, include: Acquire real-time status data of each charging terminal in the charging station. The real-time status data includes vehicle locking status, rearview mirror folding status, and in-vehicle activity intensity characteristics. The real-time status data is processed using a preset departure probability assessment model to generate departure probability coefficients for each charging terminal. Based on the departure probability coefficient, charging terminals are divided into a set of highly elastic loads and a set of low-elastic loads. Obtain the task urgency index issued by the macro scheduling center, and dynamically adjust the preset adjustment cost threshold based on the task urgency index; Obtain the historical charging records of each charging terminal in the low-elasticity load set, calculate the adjustment cost index of each charging terminal, and the adjustment cost index characterizes the product of the response delay of power regulation and the historical fluctuation variance. Charging terminals whose adjustment cost indicators meet the adjustment cost threshold are identified as quasi-elastic loads and included in the adjustment resource pool. Calculate the total adjustment capacity of the highly elastic load set and the quasi-elastic load set, and calculate the elasticity gap based on the difference between the total adjustment capacity and the macro-scheduling task requirements; If the elasticity gap is greater than zero, calculate the regional group stay tendency index, and feed the elasticity gap and the regional group stay tendency index back to the macro-dispatch center so that the macro-dispatch center can reconstruct the regional dispatch strategy.

2. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 1, characterized in that, The specific process of obtaining real-time status data includes: reading the vehicle control local area network CAN bus message through the communication interface of the charging terminal, parsing out the vehicle lock status position and the rearview mirror folding status position; at the same time, collecting the activity intensity value inside the vehicle through infrared or ultrasonic sensors deployed on the charging terminal, and storing the status position and activity intensity value after associating them with a timestamp.

3. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 1, characterized in that, The steps for generating the departure probability coefficient include: performing sliding integral processing on the activity intensity value within a preset time window to obtain the activity accumulation; multiplying the vehicle locking state, rearview mirror folding state, and activity accumulation by the corresponding preset weight coefficients, and summing the product terms to obtain the departure probability coefficient.

4. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 1, characterized in that, The steps for dynamically adjusting the preset adjustment cost threshold include: obtaining the task urgency index issued by the macro scheduling center, inputting the task urgency index into a preset monotonically increasing mapping function, and outputting the corresponding adjustment cost threshold. The value of the task urgency index is positively correlated with the value of the output adjustment cost threshold.

5. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 1, characterized in that, The steps for calculating the regulation cost index include: extracting the timestamps of the power command issuance time and the actual current response time from historical charging records, and calculating the difference between the two to obtain the response delay; extracting power regulation data from historical charging processes and calculating the fluctuation variance of the power regulation data; and multiplying the response delay by the fluctuation variance to obtain the regulation cost index.

6. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 5, characterized in that, The steps for identifying a charging terminal as a quasi-elastic load include: comparing the calculated adjustment cost index with the dynamically adjusted adjustment cost threshold; if the value of the adjustment cost index is less than the value of the adjustment cost threshold, the charging terminal is determined to meet the adjustment conditions, is identified as a quasi-elastic load, and added to the adjustment resource pool.

7. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 6, characterized in that, The specific calculation process of the total regulation capability includes: obtaining the current adjustable power limit of each charging terminal in the high-elasticity load set and the quasi-elasticity load set; and accumulating the sum of the adjustable power limits of each charging terminal in the high-elasticity load set with the sum of the adjustable power limits of each charging terminal in the quasi-elasticity load set to obtain the total regulation capability.

8. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 7, characterized in that, The steps for calculating the regional group stay tendency index include: obtaining the departure probability coefficient of all charging terminals in the station, calculating the complement value of each departure probability coefficient; performing an arithmetic average on all complement values ​​to obtain the regional group stay tendency index, which is used to characterize the overall stay tendency of the load in the station.

9. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 8, characterized in that, The steps of reconstructing the regional scheduling strategy include: breaking down the peak shaving and valley filling task requirements of the current station based on the feedback elasticity gap and the regional group dwell tendency index; and allocating the decomposed task requirements to other charging stations in the region, wherein the priority of task allocation is positively correlated with the regional group dwell tendency index of the receiving station.

10. The integrated photovoltaic-storage-charging optimization control method based on user-side load response according to claim 9, characterized in that, After completing the macro scheduling task, the actual departure status data of each charging terminal within a preset time period is obtained; the actual departure status data is compared with the corresponding departure probability coefficient, and the prediction deviation value is calculated; the preset weight coefficient in the departure probability assessment model is adaptively corrected according to the prediction deviation value to update the calculation logic of the subsequent departure probability coefficient; the preset time period is from 1 second to 60 seconds.

Images