Roof photovoltaic energy storage distribution intelligent scheduling control method

Abstract

The invention relates to a roof photovoltaic energy storage distribution intelligent scheduling control method, which belongs to the technical field of distributed energy management, and comprises the following steps: step 1, based on the collected total solar irradiance and ambient temperature, resolving to obtain the roof temperature through a preset roof equivalent thermal network model; and 2, determining the working temperature of the photovoltaic cell in combination with the roof temperature and the total solar irradiance. According to the method, accurate prediction of a future instantaneous power gap is realized, energy scheduling is promoted to a brand new dimension of beforehand pre-control from post-event response, the instantaneous power demand on a power grid is reduced, the demand electric charge is reduced, and the power efficiency is improved. And meanwhile, the voltage stability in the plant area is ensured, flexible resources in the system can be dispatched prospectively and accurately, and the system stability and robustness higher than those of a traditional EMS are shown.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed energy management, in particular to a roof photovoltaic energy storage distribution intelligent scheduling control method. BACKGROUND

[0002] The distributed photovoltaic energy storage system of the industrial plant is a key technology to realize the optimization of enterprise energy use. In the existing energy management system, the thermal attenuation effect of photovoltaic power generation and the dynamic demand of industrial load are usually treated as two independent decoupled problems in the scheduling strategy. Specifically, the thermal inertia effect of the color steel tile roof under sunlight will cause the temperature of the photovoltaic cell to lag significantly and be higher than the ambient temperature, resulting in nonlinear attenuation of power generation efficiency; at the same time, the large equipment on the industrial production line will produce millisecond-level pulse power impact. The existing technology cannot predictively handle the instantaneous and huge energy gap caused by the superposition of the two effects, resulting in lagging energy compensation, large power grid impact and power quality decline. Therefore, there is an urgent need in the art for a predictive scheduling method that can couple the thermal-electric-load three-element dynamic characteristics.

[0003] The above information disclosed in the above BACKGROUND section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute the prior art known to those of ordinary skill in the art. SUMMARY

[0004] The purpose of the present application is to provide a roof photovoltaic energy storage distribution intelligent scheduling control method to solve the problems raised in the above BACKGROUND.

[0005] The technical solution of the present application is as follows:

[0006] Step one, based on the collected total solar irradiance and ambient temperature, the roof temperature is obtained by solving the pre-set roof equivalent thermal network model;

[0007] Step two, determine the working temperature of the photovoltaic cell in combination with the roof temperature and the total solar irradiance;

[0008] Step three, generate the predicted output power of the photovoltaic system based on the working temperature of the photovoltaic cell;

[0009] Step four, in response to the collected load instantaneous power meeting the pre-set load mutation event identification rule, identify the load mutation event;

[0010] Step five, based on the load mutation event, match and obtain the predicted peak power;

[0011] Step six, calculate and generate the predictive power gap in combination with the predicted peak power and the predicted output power of the photovoltaic system;

[0012] Step seven, according to the predicted power gap, combining preset energy storage system efficiency parameters and safety margin parameters, output the pre-scheduled discharge power instruction of the energy storage system.

[0013] Preferably, the specific process of calculating the roof temperature in step one includes:

[0014] Based on the roof temperature of the last scheduling period, plus the heat absorption increment determined by the total solar irradiance, roof equivalent heat capacity and solar radiation absorption rate, and minus the convective heat dissipation determined by the roof and the ambient temperature difference, and the radiation heat dissipation determined by the roof and the sky temperature difference, to iteratively calculate the roof temperature at the current scheduling time.

[0015] Preferably, the specific process of determining the photovoltaic cell operating temperature in step two includes:

[0016] Based on the total solar irradiance and the equivalent thermal resistance from the photovoltaic module to the roof, calculate the temperature increment;

[0017] And add the temperature increment to the roof temperature to determine the photovoltaic cell operating temperature.

[0018] Preferably, the specific process of generating the predicted output power of the photovoltaic system in step three includes:

[0019] Based on the difference between the photovoltaic cell operating temperature and the standard test condition temperature, combined with the power temperature coefficient, calculate the power correction factor;

[0020] And use the power correction factor to correct the photovoltaic output power under standard test conditions to generate the predicted output power of the photovoltaic system.

[0021] Preferably, the preset load mutation event recognition rule is:

[0022] By calculating the power change rate of the load instantaneous power in the millisecond level sampling time interval, judge whether the change rate is greater than the preset power impact threshold;

[0023] And, judge whether the absolute value of the load instantaneous power is greater than the preset power absolute value threshold;

[0024] Only when both the power change rate and the power absolute value exceed their corresponding thresholds, trigger the recognition of the load mutation event.

[0025] Preferably, the power impact threshold and the power absolute value threshold are calibrated by the following method:

[0026] The two-dimensional distribution of power change rate and power absolute value in historical high-frequency load data is analyzed, and a statistical analysis or clustering algorithm is used to determine the boundary for distinguishing normal operation condition data points from sudden event outlier data points, and the boundary is set as the threshold.

[0027] Preferably, the specific way of matching to obtain the predicted peak power in step five is:

[0028] Inquiring a preset feature database, wherein the feature database is established based on statistical analysis of historical data, and is associated with equipment model or production process and predicted peak power.

[0029] Preferably, the specific process of outputting the pre-scheduling discharge power instruction of the energy storage system in step seven includes:

[0030] Adding the predicted power gap to the preset safety margin power, and dividing the sum by the product of the energy storage inverter efficiency and the battery discharge efficiency to obtain the initial discharge power;

[0031] Comparing the initial discharge power, the maximum discharge power allowed by the energy storage system, and the power constraint function value related to the state of charge of the battery, and taking the minimum value of the three as the pre-scheduling discharge power instruction output.

[0032] Preferably, the roof equivalent heat capacity, the comprehensive convective heat transfer coefficient, and the equivalent thermal resistance are obtained by unified calibration as a set of thermodynamic model parameters in the following way:

[0033] Collecting continuous measured data at the initial deployment of the system, including total solar irradiance, ambient temperature, roof temperature and battery backboard temperature;

[0034] And using a system identification algorithm to fit the measured data, with the root mean square error of the model predicted temperature and the measured temperature sequence as the optimization objective, to inversely solve and determine the set of thermodynamic model parameters.

[0035] The present application improves a roof photovoltaic energy storage distribution intelligent scheduling control method, which has the following improvements and advantages compared with the prior art:

[0036] 1. Coupling modeling of the slowly changing building physical property of roof thermal inertia and the rapidly changing electrical property of millisecond-level load mutation, realizing accurate prediction of future instantaneous power gap, and improving energy scheduling from post-response to pre-control in a new dimension;

[0037] 2. Through the accurate pre-scheduling of energy storage, the power gap caused by the superposition of load mutation and photovoltaic thermal attenuation can be perfectly hedged, the instantaneous power demand to the power grid is reduced, the demand charge is reduced, and the voltage stability in the factory area is ensured;

[0038] 3. The model and rules proposed in the application have clear physical meanings, and clear and repeatable methods for calibrating all key adjustable parameters based on field measurement data are provided, so that the application can be easily implemented by those skilled in the art;

[0039] 4. In the harsh working conditions of superposition of high temperature and high irradiation and production peak, the method of the application can prospectively and accurately schedule the flexible resources in the system, and has higher system stability and robustness than the traditional EMS. BRIEF DESCRIPTION OF DRAWINGS

[0040] The application will be further explained below in combination with the drawings and examples:

[0041] Figure 1 is a flow chart of the system of the application. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below in combination with specific examples.

[0043] Example 1:

[0044] Please refer to Figure 1 The application provides a roof photovoltaic energy storage distribution intelligent scheduling control method technical scheme, which comprises the following steps: step one, based on the collected total solar irradiance and environmental temperature, the roof temperature is obtained by solving through a pre-set roof equivalent thermal network model;

[0045] Step two, determine the working temperature of the photovoltaic cell in combination with the roof temperature and the total solar irradiance;

[0046] Step three, generate the predicted output power of the photovoltaic system based on the working temperature of the photovoltaic cell;

[0047] Step four, in response to the collected load instantaneous power meeting the pre-set load mutation event identification rule, identify the load mutation event;

[0048] Step five, based on the load mutation event, match and obtain the predicted peak power;

[0049] Step six, calculate and generate the predicted power gap in combination with the predicted peak power and the predicted output power of the photovoltaic system;

[0050] Step seven, according to the predicted power gap, combined with the preset energy storage system efficiency parameters and safety margin parameters, output the pre-scheduled discharge power instruction of the energy storage system.

[0051] The roof photovoltaic energy storage distribution intelligent scheduling control method provided by the embodiment establishes an accurate prediction ability for future energy supply and demand mismatch, aiming to solve the problem of instantaneous power gap caused by concurrent photovoltaic power generation fluctuation and production line pulse load impact due to the roof thermal characteristics in the industrial plant scene; this method through a ring-by-ring process, starting from the modeling of the physical thermal characteristics of the building, then quantifying its influence on the photovoltaic power generation efficiency, and capturing the characteristic impact of the load in real time, finally in the time scale of milliseconds, a dynamic predicted photovoltaic output and a predicted load peak are compared in advance, so as to calculate the accurate power gap; the final product of this series of operations is an active and pre-dispatching energy storage scheduling instruction, which makes the energy storage system no longer passively respond to grid fluctuations, but changes to the key role of actively supporting power quality, fundamentally improving energy management from lagging response control to forward-looking predictive control dimension.

[0052] Embodiment 2

[0053] The specific process of obtaining the roof temperature in step one includes:

[0054] Based on the roof temperature of the last scheduling period, plus the heat absorption increment determined by the total solar irradiance, the equivalent heat capacity of the roof and the solar radiation absorption rate, and minus the convective heat dissipation determined by the temperature difference between the roof and the environment, and the radiation heat dissipation determined by the temperature difference between the roof and the sky, to obtain the roof temperature at the current scheduling time by iterative calculation.

[0055] The specific process of determining the working temperature of the photovoltaic cell in step two includes:

[0056] Based on the total solar irradiance and the equivalent thermal resistance of the photovoltaic module to the roof, the temperature increment is calculated;

[0057] And add the temperature increment to the roof temperature to determine the working temperature of the photovoltaic cell.

[0058] In this embodiment, the solution of the roof equivalent thermal network model and the determination of the working temperature of the photovoltaic cell constitute the physical basis of the prediction chain; the first step of the process is to establish a dynamic evolution model of the roof temperature, and the core is to quantify the thermal inertia effect of the color steel tile roof; the physical basis is derived from the first law of thermodynamics, and the lumped parameter method is used to simplify the complex heat exchange process of the roof-photovoltaic system, and the fundamental motivation is to overcome the defects of the traditional NOCT and other steady-state models that cannot capture the temperature change lag phenomenon of the color steel tile roof due to its heat capacity characteristics; the dynamic model regards the roof temperature as a continuous evolution state variable to realize the dynamic and forward-looking description of the thermal environment of the photovoltaic system;

[0059] The solution of the roof temperature is carried out by iteration of the discrete form of the following dynamic equation:

[0060]

[0061] In the formula, T roof (t k ) represents the predicted roof absolute temperature at the current scheduling time t k , which is the direct output of the formula; t k and t k-1 represent the current and last scheduling period time points respectively; Δt represents the time step, that is, the iteration period of the model; T roof (t k-1 ) represents the roof absolute temperature at the last time, which is the initial state input of this iteration, derived from the calculation output value of the last time or the measured value at the initial time; C roof represents the equivalent heat capacity of the roof, which is the heat storage capacity of the roof system and is a key adjustable parameter closely related to the roof material and structure; α represents the equivalent solar radiation absorption rate of the roof-photovoltaic module system, which is a physical property of the system; G(t k ) represents the total solar irradiance at time t k , which is derived from the real-time measurement value of the on-site irradiance meter; A pv represents the total area of the roof covered by the photovoltaic module, which is derived from the design parameters of the photovoltaic system; h c represents the comprehensive convective heat transfer coefficient of the roof and the environment, which is affected by factors such as wind speed and is calibrated as an adjustable parameter in actual application; T amb (t k ) represents the ambient absolute temperature at time t k , which is derived from the real-time measurement value of the on-site temperature sensor; ∈ represents the equivalent emissivity of the roof, which is a physical property of the system; σ represents the Stefan-Boltzmann constant, which is a well-known constant in physics; T sky (t k ) represents the ambient absolute temperature at time t kthe equivalent sky absolute temperature, which can be estimated by Swinbank's empirical formula;

[0062] This is because the roof's radiation heat dissipation to the sky is not based on the ambient temperature, but on an equivalent sky temperature, which is usually much lower than the ambient temperature, especially on clear and cloudless nights; Swinbank's formula provides a simple way to estimate this equivalent sky temperature using the ambient temperature, in the form of an empirical formula T sky (t k ) = 0.0552·T amb (t k ) 1.5 , where the temperature unit is Kelvin (K);

[0063] As a logical extension, the process of determining the operating temperature of the photovoltaic cell then couples the above roof temperature with the thermal characteristics of the photovoltaic module itself through the following formula:

[0064] T pv_c (t k ) = T roof (t k ) + G(t k )·R th

[0065] In the formula, T pv_c (t k ) represents the predicted absolute temperature of the photovoltaic cell at the current time, which is the bridge connecting the thermal model and the electrical model; T roof (t k ) is derived from the calculation output of the aforementioned roof temperature solving formula, reflecting the close connection between the steps; G(t k ) is the real-time measurement value of the irradiance meter; R th represents the equivalent thermal resistance of the photovoltaic module to the roof, with a unit of (K·m 2 / W), which is a key adjustable parameter representing the heat transfer capacity from the cell to the roof, and is strongly related to the installation method of the module;

[0066] At the application level, these two formulas are sequentially called in the energy management system with a time step Δt as the period, using the previous time state and current measurement input to calculate the accurate prediction value of the photovoltaic cell temperature at the current time; The core contribution of this process is to realize the dynamic tracking and prediction of the operating temperature of the photovoltaic cell, accurately capturing the temperature lag effect caused by the thermal inertia of the color steel tile roof; Compared with traditional methods, this embodiment provides a dynamic and more physically realistic temperature boundary condition for subsequent power prediction, significantly improving the accuracy of the source of the entire prediction chain;

[0067] From the physical point of view, the three terms in the bracket on the right side of the equation represent three core heat exchange processes of the roof in a unit time step Δt: the first term αG(t k )A pv is the energy absorbed by the roof from the solar radiation; the second term h c A pv (T roof (t k-1 )-T amb (t k )) is the energy dissipated by the roof to the surrounding environment through convection; and the third term is the energy dissipated by the roof to the sky through thermal radiation; the entire equation is based on the net inflow or outflow of the three kinds of energy to update the roof temperature; here, the term G(t k )·R th represents the temperature gradient formed in the process of transferring the heat generated by the photovoltaic module itself as a heat source to the roof through the equivalent thermal resistance R th due to the existence of solar irradiance G(t k ). Therefore, the actual working temperature of the photovoltaic cell is further raised on the basis of the roof temperature T roof (t k ) attached to it.

[0068] Embodiment 3

[0069] The specific process of generating the predicted output power of the photovoltaic system in step three includes:

[0070] Based on the difference between the working temperature of the photovoltaic cell and the temperature under standard test conditions, and combined with the power temperature coefficient, a power correction factor is calculated and obtained;

[0071] And the power correction factor is used to correct the photovoltaic output power under standard test conditions to generate the predicted output power of the photovoltaic system;

[0072] The generation of the predicted output power of the photovoltaic system in this embodiment is the result of the calculation of the previous thermodynamic model, which is used to convert the predicted cell temperature into a quantitative evaluation of the actual power generation capacity; the calculation of this step itself is based on the standard model in the photovoltaic engineering field, which is the basic common sense known to those skilled in the art; the unique part of the present application is to inject a dynamic and accurate temperature input T pv_c (t k ) into this known model; this input value is exactly the product of the previous step of deeply coupling the thermal inertia effect of the roof, which makes a static power calculation formula reflect complex dynamic physical processes, so as to quantitatively predict the power generation efficiency decay caused by high temperature;

[0073] The predicted output power is calculated by the following model:

[0074]

[0075] P pv_pred (t k ) represents the predicted total output power of the photovoltaic system at time t k , which is the final output of this step; P STC represents the rated power of the photovoltaic system under standard test conditions, which is derived from the cumulative addition of photovoltaic module nameplate parameters; G(t k ) represents the total solar irradiance at the current time; G STC represents the irradiance under standard test conditions, which is defined by the industry standard as 1000 W / m 2 ; k T represents the power temperature coefficient, which is derived from the data manual provided by the photovoltaic module manufacturer and is an inherent property of the module; T pv_c (t k ) represents the current predicted absolute temperature of the photovoltaic cell, which is derived from the output of the aforementioned cell temperature calculation formula; T STC represents the absolute temperature of the cell under standard test conditions, which is defined by the industry standard as 298.15 K;

[0076] The power temperature coefficient k T is usually a negative value for crystalline silicon photovoltaic modules, for example -0.3% / ℃ to -0.4% / ℃, which means that when the actual working temperature T pv_c (t k ) of the cell is higher than the standard test temperature T STC , the value of the correction term in the brackets [1+k T ·(T pv_c (t k )-T STC )] will be less than 1, resulting in actual output power lower than ideal power; the present application precisely predicts T pv_c (t k ) to accurately quantify the power decay caused by high temperature;

[0077] In the energy management system, this formula is called immediately after the temperature calculation, which converts the predicted cell working temperature in real time into a prediction of the power generation capacity of the photovoltaic system; this method accurately quantifies the thermal decay effect of photovoltaic power generation; the output is no longer an estimated power based on ideal or static temperature, but a forward-looking available power prediction value that takes into account the influence of building roof thermal inertia; this provides a real and reliable quantitative basis for the subsequent dispatching decision to deal with load impact;

[0078] Example 4

[0079] The preset load mutation event identification rule is:

[0080] By calculating the power change rate of the load instantaneous power in the millisecond level sampling time interval, it is judged whether the change rate is greater than the preset power impact degree threshold value;

[0081] And it is judged whether the absolute value of the load instantaneous power is greater than the preset power absolute value threshold value;

[0082] Only when the power change rate and the power absolute value both exceed the corresponding threshold value, the identification of the load mutation event is triggered.

[0083] The power impact degree threshold value and the power absolute value threshold value are calibrated by the following method:

[0084] The two-dimensional distribution of the power change rate and the power absolute value in the historical high-frequency load data is analyzed, and a statistical analysis or a clustering algorithm is used to determine the boundary for distinguishing the normal operation condition data points and the mutation event outlier data points, and the boundary is set as the threshold value.

[0085] In the embodiment, the identification of the load mutation event relies on a composite logic judgment rule customized for the impact characteristics of a specific industrial load; the rule runs on a millisecond level sampling time scale, and the core goal is to accurately and robustly capture the impact events that pose a substantial threat to power quality from continuous power data streams; the power data in industrial sites is full of noise, and if only a single power change rate threshold value is relied on, false positives are extremely easy to occur; on the contrary, if only a single power absolute value threshold is relied on, it is impossible to distinguish between steady high load and dynamic impact; therefore, the present application conceives an AND logic, which is constrained by the double thresholds of power impact degree and power absolute value, to ensure that only those events with high energy jump speed and high energy level are determined as mutation events that need to be intervened, greatly improving the signal-to-noise ratio and robustness of event identification;

[0086] The identification rule is defined as a trigger:

[0087]

[0088] In the formula, SE trigger (t i ) represents the load mutation event trigger flag at the millisecond level sampling time t i , which is a Boolean value, and when its value is 1, the subsequent pre-scheduling calculation will be immediately activated; t i and t i-δt represent the current and the last millisecond level sampling time respectively; δt represents the millisecond level sampling time interval, which is derived from the hardware setting of the high-frequency power acquisition module; P load (t i ) represents the load instantaneous power at time t iThe collected instantaneous load power is derived from real-time measurements of a high-frequency power quality analyzer deployed on the main load circuit; γ jerk represents the power impact threshold; γ power represents the power absolute value threshold, both of which are key adjustable parameters;

[0089] These two key thresholds γ jerk and γ power are data-driven calibrated, ensuring the implementability of those skilled in the art; the process is achieved by collecting and analyzing high-frequency historical load data of at least one complete production cycle of the industrial plant, and drawing a two-dimensional scatter plot of the power change rate and the power absolute value (P load ); on this graph, data points of normal operating conditions will be densely distributed in a low-value area, while sudden event data points caused by the start of equipment such as welding machines and punching machines will appear as outliers; at this time, statistical methods such as three times the standard deviation, or clustering algorithms such as DBSCAN, can be used to automatically or semi-automatically demarcate a boundary that clearly distinguishes sudden events from regular fluctuations, the projections of this boundary on the two coordinate axes are the thresholds γ jerk and γ power ;

[0090] The variable name γ jerk here is a metaphorical reference to the concept of physics; in physics, jerk is the rate of change of acceleration, which represents the degree of change of force, similarly, the power impact degree here is essentially the rate of change of power, which measures the degree of impact of the load on the power grid, rather than the absolute value of power.

[0091] Embodiment 5

[0092] The specific way to match and obtain the predicted peak power in step five is:

[0093] Query a pre-set feature database, wherein the feature database is established based on statistical analysis of historical data and is associated with equipment models or production processes and predicted peak power.

[0094] In this embodiment, the matching and obtaining of the predicted peak power is a quick quantification of the maximum power that the load sudden event may reach in the future after the event is identified; this step enables the system to predict the degree of impact; the implementation is to query a pre-set feature database, which reflects the combination of efficiency and experience; the feature database is essentially a lookup table, and its construction process is based on deep statistical analysis of historical high-frequency load data of the plant, and a specific equipment model or a specific production process, such as the 5# punching machine pressing process, is associated with a statistically determined typical peak power Pload_peak The correlation is performed; when the load mutation event identification rule is triggered, the system can match and call the corresponding P load_peak value according to the current production line operation state or event waveform characteristics, and the effect of this mechanism is that the response to an unknown impact is converted into a quantitative processing of a known characteristic event, greatly shortening the decision delay and providing key and forward-looking input data for subsequent gap calculation.

[0095] Embodiment 6

[0096] The specific process of outputting the pre-scheduling discharge power instruction of the energy storage system in step seven includes:

[0097] The predicted power gap is added to the preset safety margin power, and the sum is divided by the product of the energy storage inverter efficiency and the battery discharge efficiency to obtain the initial discharge power;

[0098] The initial discharge power, the maximum discharge power allowed by the energy storage system, and the power constraint function value related to the battery state of charge are compared, and the minimum value among the three is taken as the pre-scheduling discharge power instruction output.

[0099] The equivalent thermal capacity of the roof, the comprehensive convective heat transfer coefficient, and the equivalent thermal resistance are obtained by unified calibration as a set of thermodynamic model parameters in the following way:

[0100] Continuous measured data at the initial deployment of the system is collected, including total solar irradiance, ambient temperature, roof temperature, and battery backboard temperature;

[0101] The measured data is fitted using a system identification algorithm, with the root mean square error of the model predicted temperature and the measured temperature sequence being minimized as the optimization objective, and the set of thermodynamic model parameters is determined by reverse solving.

[0102] In this embodiment, the output of the energy storage system pre-scheduling discharge power instruction is the terminal point and core execution link of the scheduling logic of the present application; it is triggered by the identified mutation event, and generates the final control instruction by integrating all the previous prediction information; when the mutation event is triggered, the system immediately quantifies the power gap that is about to occur; this calculation is based on the principle of energy conservation, and the novelty lies in that it abandons the traditional, lag-based calculation method based on grid power feedback, and instead actively compares a predicted load peak with a real-time predicted photovoltaic output; this strategy of using prediction against prediction achieves accurate quantification of the maximum instantaneous power gap that is about to occur and is generated by the superposition of thermal-electric-load three-element mismatch effects;

[0103] The predicted power gap is calculated by the following formula:

[0104] ΔP gap_pred =Pload_peak -P pv_pred (t i )

[0105] where ΔP gap_pred represents the predicted power gap, which is the direct basis of the energy storage dispatching quantity; P load_peak represents the predicted peak power of the load mutation event of this type, which is obtained by querying the feature database; P pv_pred (t i ) represents the predicted output power of the photovoltaic system at the current millisecond level time t i , which is obtained by calculating the aforementioned photovoltaic power prediction formula, and this direct call embodies the seamless connection between steps, and introduces the quantization result of the thermal inertia effect into the final millisecond-level decision-making;

[0106] Based on the predicted gap, the final energy storage pre-dispatch discharging power instruction is formulated; the formulation process of the instruction considers the physical constraints and safety boundaries of the system, ensuring the real feasibility of the dispatching instruction; the calculation process is as follows:

[0107]

[0108] where P ess_pre represents the energy storage system pre-dispatch discharging power instruction, which is the final output of the method of the present application; η inv and η bat respectively represent the energy storage inverter efficiency and the battery discharging efficiency, which are obtained from the data manual provided by the equipment manufacturer; P margin represents the safety margin power, which is an adjustable parameter, used to compensate for the prediction model error and unmodeled dynamics; P ess_max represents the maximum discharging power allowed by the energy storage system, which is obtained from the hardware specifications of the energy storage system; f(SOC) represents a power constraint function related to the battery state of charge SOC, used to implement protection strategies, for example, when the SOC is lower than 20%, the function value linearly decreases to avoid over-discharging of the battery;

[0109] The function f(SOC) is the key protection logic to ensure the long-term healthy operation and safety of the energy storage system; in addition to preventing over-discharging, it usually also contains logic to prevent high-power charging at high SOC, and strategies to limit the charging and discharging power under extreme temperature conditions, which is an embodiment of the core functions of the battery management system;

[0110] To ensure the basic accuracy of the entire prediction chain, the key thermodynamic model parameters C roof ,h c ,R thMust be accurately calibrated; this calibration process is performed at the beginning of system deployment, by collecting continuous measured data of at least one full sunny day, including total solar irradiance G, ambient temperature T amb , and actual temperature of roof and battery backsheet T roof_measured , T pv_c_measured ; with these data, a system identification algorithm is employed, such as a gradient descent based parameter optimization method, to fit the historical data, with the objective of minimizing the root mean square error (RMSE) between the model predicted temperature sequence and the measured temperature sequence, to inversely solve and determine a set of optimal C roof , h c , R th values;

[0111] This optimization objective, minimizing the root mean square error (RMSE) between the model predicted temperature and the measured temperature sequence, is mathematically expressed as: seeking a set of parameters (C roof , h c , R th ) that makes reach a minimum, where T predicted is the temperature calculated by the model proposed in this invention, and T measured is the actual measured value;

[0112] This calibration process ensures that the physical parameters of the model can truly reflect the current specific site's building and installation characteristics, greatly improving the practicality and reliability of the model;

[0113] In application, the calculation of the above power gap and the generation of the dispatching instruction are both completed within milliseconds when the load mutation event is triggered, P ess_pre instruction is immediately sent to the energy storage converter for execution; this complete decision and execution process realizes a fundamental change from passive compensation to active support; the energy storage system has outputted the appropriate support power in advance based on the triple accurate prediction of the roof thermal inertia, photovoltaic thermal attenuation and load impact characteristics before the actual impact of the load impact on the power grid; as a direct result, under the most adverse working conditions such as summer high temperature and high irradiance production peak, this method can greatly smooth the power curve on the grid side, avoid voltage sag caused by the start of large equipment, so as to ensure the stable operation of the production line while significantly reducing the demand charge, enhancing the resilience and self-consistency of the entire plant energy system;

[0114] The essence of this calculation method is to combine a feature-based deterministic prediction value P load_peak of future load impact with a physical model-based dynamic prediction value P pv_pred of current available power generation capacity P i) Prospective matching is performed, thereby realizing accurate prediction of the maximum supply-demand mismatch risk in future millisecond to second time scale.

[0115] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the present application, and all modifications and equivalent replacements should be covered in the scope of the claims of the present application.

Claims

1. A roof photovoltaic energy storage distribution intelligent scheduling control method, characterized in that, The method comprises the following steps: Step 1: based on the collected total solar radiation and ambient temperature, the roof temperature is calculated through the preset equivalent thermal network model of the roof; Step 2: the working temperature of the photovoltaic cell is determined by combining the roof temperature and the total solar radiation; Step 3: the predicted output power of the photovoltaic system is generated based on the working temperature of the photovoltaic cell; Step 4: in response to the collected load instantaneous power meeting the preset load mutation event identification rule, the load mutation event is identified; Step 5: based on the load mutation event, the predicted peak power is matched and obtained; Step 6: the predictive power gap is calculated by combining the predicted peak power and the predicted output power of the photovoltaic system; Step 7: according to the predictive power gap, the preset energy storage system efficiency parameter and safety margin parameter are combined to output the pre-scheduled discharge power instruction of the energy storage system.

2. The roof photovoltaic energy storage distribution intelligent scheduling control method according to claim 1, characterized in that, The specific process of calculating the roof temperature in step 1 includes: Based on the roof temperature of the last scheduling period, add the heat absorption increment determined by the total solar radiation, roof equivalent heat capacity and solar radiation absorption rate, and subtract the convective heat dissipation determined by the temperature difference between the roof and the environment, and the radiation heat dissipation determined by the temperature difference between the roof and the sky, to iteratively calculate the roof temperature at the current scheduling time.

3. The method according to claim 2, wherein, The specific process of determining the working temperature of the photovoltaic cell in step 2 includes: Based on the total solar radiation and the equivalent thermal resistance from the photovoltaic module to the roof, the temperature increment is calculated; And add the temperature increment to the roof temperature to determine the working temperature of the photovoltaic cell.

4. The roof photovoltaic energy storage distribution intelligent scheduling control method according to claim 3, characterized in that, The specific process of generating the predicted output power of the photovoltaic system in step 3 includes: Based on the difference between the working temperature of the photovoltaic cell and the standard test condition temperature, the power temperature coefficient is combined to calculate the power correction factor; And the power correction factor is used to correct the photovoltaic output power under standard test conditions to generate the predicted output power of the photovoltaic system.

5. The roof photovoltaic energy storage and distribution intelligent scheduling control method according to claim 1, characterized in that, The preset load mutation event identification rule is: By calculating the power change rate of the load instantaneous power within the millisecond level sampling time interval, it is judged whether the change rate is greater than the preset power impact threshold; And, it is judged whether the absolute value of the load instantaneous power is greater than the preset power absolute value threshold; Only when the power change rate and the power absolute value exceed their corresponding thresholds, the identification of the load mutation event is triggered.

6. The roof photovoltaic energy storage and distribution intelligent scheduling control method according to claim 5, characterized in that, The power impact threshold and the power absolute value threshold are calibrated by the following method: Analyze the two-dimensional distribution of power change rate and power absolute value in historical high-frequency load data, and use statistical analysis or clustering algorithm to determine the boundary for distinguishing between regular operating condition data points and mutation event outlier data points, and set the boundary as the threshold.

7. The roof photovoltaic energy storage and distribution intelligent scheduling control method according to claim 1, characterized in that, The specific way of matching and obtaining the predicted peak power in step 5 is: Query the preset feature database, wherein the feature database is established based on statistical analysis of historical data and is associated with equipment model or production process and predicted peak power.

8. The roof photovoltaic energy storage and distribution intelligent scheduling control method according to claim 1, characterized in that, The specific process of outputting the pre-scheduled discharge power instruction of the energy storage system in step 7 includes: The predicted power gap is added to a preset safety margin power, and the sum is divided by the product of the energy storage inverter efficiency and the battery discharge efficiency to obtain an initial discharge power; The initial discharge power, the maximum discharge power allowed by the energy storage system, and the power constraint function value related to the battery state of charge are compared, and the minimum value among the three is taken as the pre-scheduling discharge power instruction output.

9. The method of claim 3, wherein the method further comprises: The roof equivalent heat capacity, the comprehensive convective heat transfer coefficient, and the equivalent thermal resistance are obtained by unified calibration as a set of thermodynamic model parameters in the following way: Continuous measured data at the initial deployment of the system are collected, including total solar irradiance, ambient temperature, roof temperature, and battery backsheet temperature; A system identification algorithm is used to fit the measured data, with the root mean square error between the model predicted temperature and the measured temperature sequence being minimized as the optimization objective, and the set of thermodynamic model parameters being determined by reverse solving.

Images