Building energy optimization method and system based on industrial internet of things

Abstract

The application discloses a building energy optimization method and system based on an industrial Internet of Things, and relates to the field of energy optimization management.The method comprises the following steps: based on operation data of a train braking process when entering a station, a braking energy prediction model is used to obtain a braking energy time sequence in a preset future time window; based on station environment data and train timetable data, a station load prediction model is used to obtain a station load time sequence; a collaborative control optimization model is used, and a multi-objective optimization algorithm is used to solve an optimal control strategy at a current time; and the optimal control strategy is used to collaboratively control an energy storage system, a station flexible load and a grid-connected interface, a rolling time domain optimization framework is used, and the optimization control step is repeatedly executed every preset period, so that dynamic updating control is realized.The technical problem of insufficient collaborative control capability of train regenerative braking energy and station building load in an existing subway station is solved.

Description

Technical Field

[0001] This invention relates to the field of energy optimization management, and more specifically to a method and system for optimizing building energy consumption based on the Industrial Internet of Things. Background Technology

[0002] Trains frequently experience acceleration and braking during operation, especially on urban lines with short station intervals. The braking process during train entry generates a large amount of regenerative braking energy, which can be converted into electrical energy by the traction motor and fed back to the DC traction grid.

[0003] However, existing technologies only focus on optimizing a single type of load. Station loads are easily affected by various factors, resulting in large fluctuations in station bus voltage and insufficient coordinated control capabilities between train regenerative braking energy and station building loads. Summary of the Invention

[0004] This application provides a method and system for optimizing building energy consumption based on the Industrial Internet of Things, which addresses the technical problem of insufficient coordinated regulation of train regenerative braking energy and building load in existing subway stations.

[0005] In view of the above problems, this application provides a method and system for optimizing building energy consumption based on the Industrial Internet of Things.

[0006] In a first aspect, this application provides a method for optimizing building energy consumption based on the Industrial Internet of Things, the method comprising: Based on the operational data during the train's braking process at the station, the braking energy time series within a preset future time window is obtained through braking energy prediction model analysis. Based on station environment data and train timetable data, the station load time series within the preset future time window is obtained through the station load prediction model. The current energy storage status data of the energy storage system is obtained, and the braking energy time series, the station load time series and the current energy storage status data are input into the collaborative regulation and optimization model. The optimal regulation strategy at the current moment is solved by a multi-objective optimization algorithm. The optimal regulation strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. The energy storage system, on-site flexible loads, and grid connection interface are coordinated and controlled according to the optimal control strategy. A rolling time-domain optimization framework is adopted to repeatedly execute the optimization control steps at preset intervals to achieve dynamic update control.

[0007] Secondly, the present invention provides a building energy optimization system based on the Industrial Internet of Things, the system comprising: The energy time series generation module is used to obtain the braking energy time series within a preset future time window based on the operating data of the train entering the station braking process and through the braking energy prediction model. The load time series generation module is used to obtain the station load time series within the preset future time window based on station environment data and train timetable data through the station load prediction model; The optimal control strategy acquisition module is used to acquire the current energy storage status data of the energy storage system, input the braking energy time series, the station load time series and the current energy storage status data into the collaborative control optimization model, and solve the optimal control strategy at the current moment through a multi-objective optimization algorithm. The optimal control strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. The control strategy control module is used to coordinate the control of the energy storage system, the flexible load in the station and the grid connection interface according to the optimal control strategy. It adopts a rolling time domain optimization framework and repeatedly executes the optimization control steps at preset intervals to achieve dynamic update control.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application first predicts the regenerative braking energy time series based on the operational data during train braking at the station using a braking energy prediction model, providing accurate boundaries for subsequent energy scheduling and laying a data foundation for subsequent analysis. Second, it predicts the station load time series based on station environmental data and train timetable data using a station load prediction model, achieving multi-factor analysis of station building energy consumption. Third, it obtains the optimal control strategy through a collaborative control optimization model and a multi-objective optimization algorithm, seeking the best trade-off among multiple objectives to achieve multi-objective optimization. Finally, it performs collaborative control according to the optimal control strategy and adopts a rolling time-domain optimization framework, repeatedly executing prediction and optimization at preset intervals, realizing real-time dynamic response to train braking events and station load changes, ensuring the safe and stable operation of the power supply system.

[0009] In summary, this invention significantly improves the overall utilization rate of train regenerative braking energy through an industrial IoT-based building energy optimization method, effectively reduces station bus voltage fluctuations, extends the cycle life of energy storage systems, reduces station operating electricity costs, and enhances the local absorption capacity of renewable energy, thereby achieving synergistic optimization of the operational economy, low carbon emissions, and power supply security of subway station energy systems. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating the building energy optimization method based on the Industrial Internet of Things (IIoT) proposed in this application. Figure 2This is a schematic diagram of the building energy optimization system based on the Industrial Internet of Things (IIoT) of this application.

[0011] In the attached diagram, the components represented by each number are as follows: Energy time series generation module 11, load time series generation module 12, optimal control strategy acquisition module 13, control strategy control module 14. Detailed Implementation

[0012] This application provides a building energy optimization method based on the Industrial Internet of Things, which specifically addresses the technical problem of insufficient coordinated regulation of train regenerative braking energy and building load in existing subway stations.

[0013] The present invention will now be described in detail with reference to the accompanying drawings.

[0014] Example 1, as Figure 1 As shown, this application provides a building energy optimization method based on the Industrial Internet of Things, the method comprising: S10: Based on the operating data during the train's braking process at the station, the braking energy time series within a preset future time window is obtained through braking energy prediction model analysis; Step S10 in the method provided in this application embodiment includes: The instantaneous speed, initial speed at the moment of entry, braking distance, number of train formations, real-time load status, and braking mode type of the train are collected in real time through an industrial Internet of Things platform, and the collected data are used as the operation data. The operational data is input into a pre-trained braking energy prediction model, wherein the braking energy prediction model is analyzed and calculated based on the principle of train kinetic energy change and the correction coefficients fitted from historical braking data. The braking energy prediction model outputs time series data of regenerative braking power changing over time from the current moment to a preset future duration, which is used as the braking energy time series.

[0015] Specifically, firstly, based on the Industrial Internet of Things (IIoT) deployed in and along subway stations, consisting of various sensors or data acquisition terminals, network communication equipment, and data aggregation nodes, the system collects real-time data on the train's instantaneous speed upon entering the station, initial speed at the moment of entry, braking distance, number of train carriages, real-time load status, and braking mode type. This collected data is used as operational data. Specifically, braking distance is the track length traversed from the start of braking to a complete stop at the predetermined position on the platform; the number of train carriages is the number of cars constituting the train; real-time load status is the ratio or absolute difference between the actual total mass of the train and its empty weight, calculated from the number of passengers in the carriages; and braking mode type is the classification of the braking strategy currently being implemented by the train's braking system.

[0016] During the train's entry into the station, speed sensors deployed on the train bogies collect the train's instantaneous speed in real time; the train control system records the initial speed at the moment braking begins upon entering the station; the actual braking distance from the braking start point to the stopping point is calculated by combining the speed measuring motor and wheel diameter correction system with track transponder information; the current train formation quantity is read from the train formation information database; the real-time load status is calculated by converting air spring pressure sensors or axle load sensors; and multi-source data of the currently selected braking mode type is read from the braking control unit, summarized, and used as the operating data of the same train.

[0017] For example, at 9:00:00 AM, a 6-car Type A train collects the following operational data: instantaneous speed 60 km / h; initial speed 60 km / h; braking distance 200 meters; number of train cars 6; rated passenger capacity 80%; braking mode type conventional braking.

[0018] Secondly, the collected operational data, including the train's instantaneous speed, initial speed, braking distance, number of trains, load status, and braking mode type, are organized into a specific input vector and fed into the braking energy prediction model for forward calculation. Internally, the model first calculates the total kinetic energy that the train can release as it decelerates from its initial speed to its instantaneous speed, based on the principle of kinetic energy change.

[0019] Due to factors such as mechanical friction losses and the nonlinear characteristics of motor power generation efficiency changing with speed during actual braking, kinetic energy cannot be completely converted into usable regenerative electrical energy. Therefore, based on the principle of train kinetic energy change, the braking energy prediction model is dynamically corrected using correction coefficients pre-fitted from historical braking data to make the output closer to the actual situation.

[0020] Finally, the braking energy prediction model outputs time series data of regenerative braking power changing over time within a preset future duration from the current moment, which is used as the braking energy time series. Here, regenerative braking power refers to the instantaneous electrical power generated by the traction motor and fed back to the DC bus grid during the braking process.

[0021] The obtained braking energy time series has a time resolution of no less than 0.1 seconds to match the response accuracy of the power electronic control equipment. It also exhibits a trend of first increasing and then decreasing: in the initial stage of braking, the regenerative power rises rapidly as the motor operates at high speed; in the middle and later stages of braking, the regenerative power gradually decreases to zero as the speed decreases. The obtained braking energy time series is then stored in the real-time database of the industrial IoT platform for subsequent collaborative control and optimization.

[0022] For example, the running dataset is input into the braking energy prediction model for calculation, and the output is time series data of the regenerative braking power change over time for the next 15 seconds from the current moment, with a resolution of 0.5 seconds. The prediction results are as follows: from 0 to 2.5 seconds, the power rises rapidly from 0 kW to 600 kW; from 3.0 to 7.5 seconds, the power is maintained in the peak plateau region of 1000-1200 kW (with the peak of 1200 kW reaching 5.0 seconds); from 8.0 to 12.5 seconds, the power gradually decreases to 300 kW; from 13.0 to 15.0 seconds, the power slowly decays to 0 kW.

[0023] In this embodiment of the application, by collecting operational data in real time on an industrial Internet of Things platform and constructing a braking energy prediction model, the regenerative braking power within a preset time period is predicted, providing an accurate supply boundary for subsequent coordinated regulation and optimization, improving the adaptability and prediction accuracy of the prediction model to different operating conditions, and laying a data foundation for subsequent analysis.

[0024] S20: Based on station environment data and train timetable data, obtain the station load time series within the preset future time window through the station load prediction model; Step S20 in the method provided in this application embodiment includes: The station's outdoor temperature and humidity data, outdoor light intensity data, real-time personnel density data in various areas of the station, and current date data are collected in real time through an industrial IoT platform, and the collected data are used as the station's environmental data. Access the train timetable system to obtain the train arrival and departure timetables and passenger flow forecast data for the corresponding platforms within a preset time period in the future, and use the train arrival and departure timetables and passenger flow forecast data as the train timetable data; The station environment data and train timetable data are input into a pre-built station load prediction model, wherein the station load prediction model analyzes and calculates the time series of air conditioning system load, lighting system load, escalator system load, and other equipment load. The load time series of the air conditioning system, the lighting system, the escalator system, and the other equipment are superimposed to obtain the load time series within the main station, which is used as the load time series within the station.

[0025] In this embodiment, temperature sensors, light sensors, cameras and other devices are first deployed through an industrial IoT platform to collect real-time outdoor temperature and humidity data, outdoor light intensity data, real-time personnel density data in various areas inside the station, and current date data. The collected data are used as station environmental data to provide input for subsequent station load prediction models.

[0026] Secondly, since the train timetable data stores the routes and train times for each station along the entire line, the number of passengers arriving at and disembarking at platforms within future time periods is used as passenger flow forecast data for the corresponding platforms through the train timetable system. This yields the train arrival and departure timetables and passenger flow forecast data for the corresponding platforms within a preset time period, which are then used as the train timetable data. By integrating the train timetable system and passenger flow forecasts, the train timetable data is obtained, forming the input data for the station load forecasting model.

[0027] Next, the station environment data and train timetable data are input into the pre-built station load prediction model. The station load prediction model analyzes and calculates the time series of air conditioning system load, lighting system load, escalator system load, and other equipment load.

[0028] Specifically, the air conditioning system load time series is the sequence of instantaneous electrical power required by the station's air conditioning system to maintain environmental indicators over a predetermined period of time. The lighting system load time series is the sequence of instantaneous electrical power of lighting fixtures in various areas of the station over time. The escalator / elevator system load time series is the sequence of instantaneous electrical power generated by elevators and escalators within the station due to passenger boarding and alighting over time, and is highly correlated with train arrival and departure times and passenger flow forecasts. The other equipment load time series is the sequence of instantaneous electrical power of other electrical equipment besides the three major systems mentioned above over time. The terminal station load time series is the sum of the above four types of load series, reflecting the change in the station's total electrical power consumption over a predetermined period of time.

[0029] The station environmental data and train timetable data collected by the Industrial Internet of Things (IIoT) platform are input into a pre-built station load prediction model for forward calculation. Internally, the model decomposes and calculates the load based on the physical characteristics or data correlations of each load subsystem: for the air conditioning system load, real-time cooling / heating load demand is calculated based on outdoor temperature and humidity and station personnel density. Combined with the air conditioning system's coefficient of performance (COP) and equipment operating strategies, the air conditioning power at future times is predicted. Higher outdoor temperature and humidity, and greater personnel density, result in a correspondingly higher predicted air conditioning load.

[0030] For the lighting system load, based on outdoor light intensity, population density in each area, and date type, combined with the lighting system's zone control and dimming strategies, the lighting power at each future time is predicted. When outdoor light is sufficient, the lighting load near entrances and exits can be appropriately reduced; higher illuminance is maintained in densely populated areas.

[0031] For the load of the escalator and elevator system, based on train arrival and departure timetables and platform passenger flow forecast data, the instantaneous peak passenger flow caused by train arrivals and departures is determined in each future time period. Combined with the operating characteristics of the escalators and elevators, the power of the escalators and elevators at each future time is predicted. Specifically, when a train is predicted to arrive, the load of the escalators in the exit direction will increase in the short term; when a train is predicted to depart, the load of the escalators in the entrance direction will increase accordingly.

[0032] For other equipment loads, based on historical data from the same period and equipment operating patterns, a relatively stable base load is predicted, and the loads of other equipment typically fluctuate less.

[0033] Finally, after completing the forecasting of the four types of loads, the time series of the air conditioning system load, the time series of the lighting system load, the time series of the escalator and elevator system load, and the time series of the loads of other equipment are superimposed to obtain the total station load time series, which is used as the station load time series.

[0034] For example, at 9:00:00.5 AM, the station load prediction model receives input data and predicts that the air conditioning load will remain stable at 400kW for the next 15 seconds; the lighting load will be 50kW for the next 15 seconds; the escalator load will be 230kW, while the basic standby load of 30kW will be maintained for the first 10 seconds (when no trains arrive); the predicted load of other equipment is 100kW. By superimposing the values, the total station load is calculated as follows: (0-10s) 580kW, (10-15s) 780kW. With an interval of 0.5 seconds, the load prediction time series for the next 15 seconds is obtained.

[0035] In this embodiment, environmental data and passenger flow forecast data are collected in real time through an industrial IoT platform. The load forecast model within the station is used for model analysis, taking into account the influence of multiple factors to improve the accuracy and interpretability of load forecasting and provide high-quality input data for subsequent analysis.

[0036] S30: Obtain the current energy storage status data of the energy storage system, input the braking energy time series, the station load time series and the current energy storage status data into the collaborative regulation and optimization model, and solve the optimal regulation strategy at the current moment through a multi-objective optimization algorithm. The optimal regulation strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. In step S30 of the method provided in this application embodiment, the current state of charge percentage, current health percentage, current maximum allowable charging power, current maximum allowable discharging power, and current available capacity of the energy storage system are collected in real time through an industrial Internet of Things platform as current energy storage state data.

[0037] In this embodiment, the current state of charge percentage refers to the percentage of the energy storage battery's current remaining capacity relative to its rated total capacity, where 0% represents a fully discharged battery and 100% represents a fully charged battery. The current health status percentage refers to the ratio of the energy storage battery's current maximum usable capacity to its rated capacity at the time of manufacture, reflecting the battery's aging degree, which decreases with increasing charge-discharge cycle count. The current maximum permissible charging power is the maximum charging power value that the energy storage system can safely accept. The current maximum permissible discharging power is the maximum discharging power value that the current energy storage system can safely provide. The current usable capacity is the total amount of electrical energy that the energy storage system can actually store and release.

[0038] Specifically, the battery management unit deployed in the energy storage battery cluster continuously monitors parameters such as voltage and temperature of each battery cell. After summarizing and calculating by the battery cluster management unit, the current state of charge percentage and current state of health percentage of the entire energy storage system are output. The calculation of the current state of charge percentage is usually based on a combination of the ampere-hour integration method and the open-circuit voltage correction method, while the current state of health percentage is based on a comprehensive analysis of multiple dimensions such as cycle count and capacity decay.

[0039] The energy storage converter calculates the current maximum allowable charging power and the current maximum allowable discharging power in real time based on the current grid voltage, equipment temperature, and power module status. The current available capacity is calculated in real time by the battery management system based on the current health status percentage. These various data are integrated to form a current energy storage state dataset describing the current operating state and capacity boundaries of the energy storage system, which serves as input data for subsequent collaborative control and optimization models.

[0040] Step S30 in the method provided in this application embodiment includes: A mathematical model is constructed with the optimization objectives of maximizing energy utilization, maximizing operational efficiency, and maximizing system stability. Here, the energy utilization rate is defined as the proportion of regenerative braking energy effectively utilized within the preset future time window to the total energy of the braking energy time series. The operational efficiency is defined as the net benefit after deducting the aging loss cost and energy conversion loss cost of the energy storage system from the electricity purchase cost saved within the preset future time window. The system stability is defined as a stability index constructed based on the bus voltage deviation value and the energy storage state of charge deviation value. The constraints for optimization are set, including the allowable fluctuation range of bus voltage, the allowable range of energy storage state of charge, the limit of energy storage charging and discharging power, the limit of flexible load adjustment, and the power balance constraint. Based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives. The optimal solution is then selected from the set of non-dominated optimal solutions according to the current operating mode, and is used as the optimal control strategy at the current moment.

[0041] In this embodiment, firstly, the energy utilization rate is obtained by calculating the proportion of regenerative braking energy effectively utilized within a preset future time window to the total braking energy over the time series. Secondly, the operational efficiency is obtained by calculating the net benefit after deducting the aging loss cost and energy conversion loss cost of the energy storage system from the electricity purchase cost saved within the preset future time window. Finally, the system stability is obtained by constructing a stability index based on the bus voltage deviation and the energy storage state of charge deviation.

[0042] Based on the above calculations, the corresponding objective function is obtained. Subsequently, a mathematical model is constructed with the optimization objectives of maximizing energy utilization, maximizing operational efficiency, and maximizing system stability. For example, a mathematical model composed of the three objective functions can be constructed.

[0043] Secondly, constraints are set for the optimization solution, including constraints on the allowable fluctuation range of bus voltage, the allowable range of energy storage state of charge, the limit constraints on energy storage charging and discharging power, the limit constraints on flexible load adjustment, and power balance constraints. Specifically, based on the requirements for safe operation of equipment in the current environment, constraints for optimization are first set, and these constraints are expressed mathematically. In the subsequent solution process, each candidate solution must satisfy all constraints. Among these, the allowable fluctuation range constraint for the bus voltage means that the DC bus voltage must be maintained within the range required for safe operation of the equipment. Exceeding this range may trigger protection actions or damage the equipment; for example, Umin ≤ U ≤ Umax. The allowable range constraint for the energy storage state of charge means that the current state of charge percentage of the energy storage battery must be maintained between a set lower and upper limit, for example, 20%-90%, which can be expressed as X1 ≤ X ≤ X2, to avoid over-discharge or overcharge that would accelerate battery life degradation.

[0044] Energy storage charging and discharging power limits restrict the charging and discharging power of an energy storage system from exceeding its rated power limit, for example, -Pmax ≤ P ≤ +Pmax. Flexible load regulation limits require that operations such as air conditioning temperature adjustment, lighting illuminance adjustment, and escalator speed adjustment must be performed within the range that ensures basic passenger comfort and safe equipment operation, for example, ΔPmin ≤ ΔP ≤ ΔPmax. Power balance constraints are based on Kirchhoff's current law, requiring that the DC bus always satisfy the condition that the injected power equals the outflow power, for example, P_inflow = P_outflow.

[0045] For example, for an energy storage system in a train station, the bus voltage range is set to 750V±5%; the energy storage SOC range is 20%~90%; the energy storage power limit is -500kW~+500kW; the flexible load adjustment limit is ±40kW for air conditioning, 0~15kW for lighting, and ±20kW for escalators, with a combined range of -50kW~+50kW; the power balance is P_inflow = P_outflow, which must be strictly met at every moment within the next 15 seconds.

[0046] Finally, based on the mathematical model constructed in the above steps, a multi-objective optimization algorithm is used to solve the multi-objective optimization problem under constraints. The search generates a set of non-dominated optimal solutions that satisfy the optimization objectives. Subsequently, according to the current operating mode, the optimal solution is selected from the set of non-dominated optimal solutions. The selected solution is then encapsulated into a complete instruction set containing energy storage charging and discharging power, flexible load regulation, and grid-connected switching power, serving as the optimal control strategy output for the current moment.

[0047] In step S30 of the method provided in this application embodiment, based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives, including: The energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are used as decision variables. Combined sampling is performed within the value range of the decision variables to generate an initial candidate solution population. Based on the three optimization objectives of energy utilization, operational efficiency, and system stability, the three objective function values ​​corresponding to each candidate solution in the initial candidate solution population are calculated respectively. The candidate solutions in the initial candidate solution population are sorted in a non-dominated order, and the candidate solutions are divided into multiple non-dominated levels. In this order, a candidate solution in any non-dominated level is not dominated by other candidate solutions in the same level, and is only dominated by candidate solutions in higher levels. Starting from the first non-dominated level, candidate solutions from each non-dominated level are selected sequentially to enter the new generation population until the number of selected candidate solutions reaches the preset population size. The new generation population is evolved through crossover and mutation operations to generate an evolved candidate solution population, and the three objective function values ​​corresponding to each candidate solution are recalculated. Repeat the non-dominated sorting, selection, crossover, and mutation operations until the preset maximum number of iterations is reached. All candidate solutions in the first non-dominated level are then used as the non-dominated optimal solution set.

[0048] In this embodiment, the energy storage charging and discharging power, flexible load regulation, and grid-connected power exchange are first used as decision variables to determine the feasible regions of the three decision variables: the energy storage charging and discharging power ranges from [-Pmax, +Pmax]; the flexible load regulation ranges from [ΔPmin, ΔPmax]; and the grid-connected power exchange ranges from [-P...Pmax ... 交换 max,+P 交换 [max]. Subsequently, random sampling or Latin hypercube sampling is used to perform combined sampling within the range of values ​​of the decision variable, uniformly or randomly generating N initial points, each point corresponding to a specific set of values ​​(P, ΔP, P). 采样 The N points together form the initial candidate solution population. The population size N is usually set between 50 and 200 to ensure sufficient population diversity and provide a good starting foundation for subsequent evolutionary search.

[0049] Secondly, based on the three optimization objectives of energy utilization, operational efficiency, and system stability, an optimization algorithm is used to calculate the efficiency for each individual i in the initial population.

[0050] Based on the braking energy time series and the station load time series, combined with the energy storage charging and discharging strategy and flexible load adjustment strategy determined by the candidate solution, the proportion of regenerative braking energy effectively utilized to the total braking energy is calculated, yielding the energy utilization rate objective function value f1i. Subsequently, based on the time-of-use electricity price curve, energy storage charging and discharging efficiency, and energy storage aging cost model, the electricity purchase cost saved in the future time window is calculated. Subtracting the aging loss cost and energy conversion loss cost of the energy storage system, the operational efficiency objective function value f2i is obtained. Then, based on the bus voltage deviation value and the energy storage state of charge deviation value, the system stability index value f3i under the control of this candidate solution is calculated. Through the above calculations, each candidate solution i obtains a three-dimensional objective vector (f1i, f2i, f3i), reflecting the comprehensive performance of the candidate solution on the three optimization objectives.

[0051] Furthermore, a fast non-dominated sorting method is adopted to perform non-dominated sorting on the candidate solutions in the initial candidate solution population, and to divide the candidate solutions in the current population into multiple non-dominated levels. In this case, the candidate solutions in any non-dominated level are not dominated by other candidate solutions in the same level, and are only dominated by candidate solutions in higher levels.

[0052] Candidate solutions in the first non-dominated level are not dominated by any other candidate solutions in the initial candidate solution population. Candidate solutions in the second non-dominated level are dominated only by candidate solutions in the first non-dominated level. Candidate solutions in the third non-dominated level are dominated only by candidate solutions in the first and second non-dominated levels, and not by candidate solutions in other non-dominated levels, and so on, until all candidate solutions in the initial candidate solution population are assigned to the corresponding non-dominated level. This ensures that individuals within the same level do not dominate each other, while individuals in higher levels dominate individuals in lower levels.

[0053] Furthermore, following the order of the non-dominated levels, starting from the first non-dominated level, candidate solutions from each non-dominated level are sequentially selected to enter the new generation population. Assuming the preset population size is N, after adding layer F1, the cumulative number of individuals is |F1|. If |F1| is less than N, then all individuals from layer F2 are added, at which point the cumulative number is |F1| + |F2|. This process continues until a certain layer Fk is added, at which point the cumulative number first exceeds or equals N, meaning the number of selected candidate solutions reaches the preset population size. At this point, for this last layer Fk, individuals are sorted according to their crowding distance, with priority given to individuals with larger distances, and these are added to the new generation population until the preset size N is reached. This sequential selection by non-dominated levels ensures the priority retention of superior individuals while maintaining population diversity.

[0054] For example, a random sampling method is used to generate 100 initial points in the three-dimensional decision space as an initial candidate solution population. Then, for each initial candidate solution, the three objective function values ​​for the next 15 seconds are calculated, and the 100 initial candidate solutions are sorted in a non-dominated order. By comparing the dominance relationships of each solution, the first non-dominated level F1 contains 15 solutions, which are mutually non-dominated; the second non-dominated level F2 contains 25 solutions; the third level F3 contains 30 solutions; and the remaining 30 solutions are distributed in lower levels. First, all 15 solutions in the F1 level are selected into the new generation population, accumulating to 15; then all 25 solutions in the F2 level are selected, accumulating to 40; then all 30 solutions in the F3 level are selected, accumulating to 70; at this point, after adding the F4 level, the cumulative total reaches 90, still less than 100; continuing to add the F5 level, the cumulative total reaches 105, exceeding 100. Therefore, the crowding distance is calculated for the 15 solutions of layer F5, and the 10 solutions with the largest crowding distance are selected to be added to the new generation population, reaching 100 solutions to form the new generation population.

[0055] Furthermore, parent individuals are selected from the new generation population according to a certain selection strategy. Subsequently, the new generation population is evolved through crossover and mutation operations. For each selected pair of parent individuals, the algorithm performs a crossover operation with a preset crossover probability, usually 0.7 to 0.9, generating new solution combinations while preserving the superior characteristics of the parents.

[0056] After the crossover operation, for each newly generated individual, a mutation operation is performed with a preset mutation probability, which is usually small, such as 0.01 to 0.1. Subsequently, some decision variable components of the individual are randomly adjusted. After crossover and mutation, an evolved candidate solution population is obtained. Then, for each individual in the new population, the corresponding three objective function values ​​(f1, f2, f3) are recalculated to prepare for the next round of sorting and selection.

[0057] For example, parent individuals are then selected from the new generation population, and simulated binary crossover is performed with a crossover probability of 0.8 to generate offspring individuals. For instance, parent A (+200, -10, +100) is crossed with parent B (-150, +20, -50) to produce offspring C (+50, +5, +30) and offspring D (0, +5, +20). The offspring individuals are then mutated with a mutation probability of 0.05. After the crossover and mutation are completed, 100 new individuals are obtained, and the three objective function values ​​for each individual are recalculated.

[0058] Finally, the non-dominated sorting, selection, crossover, and mutation operations are repeated. A maximum number of iterations can be set, or the number of iterations can be set to 0. In each generation, the current population is first non-dominated sorted. Then, a new generation of population is selected based on hierarchy and crowding distance. Next, the selected individuals are crossovered and mutated to generate offspring. Finally, the objective function value of the offspring population is calculated, completing one iteration. If the number of iterations is less than the maximum number of iterations, the offspring population is used as the new current population, and iteration continues. If the number of iterations reaches the maximum number of iterations, evolution stops, and the final generation is subjected to a final non-dominated sort. All candidate solutions in the first non-dominated hierarchy are extracted as the non-dominated optimal solution set.

[0059] For example, the maximum number of iterations is set to 200 generations. Starting from generation 1, the non-dominated sorting, selection, crossover mutation, and target value calculation are repeated until generation 200 is reached. After the 200th generation of evolution is completed, the final population is subjected to non-dominated sorting, and the 23 individuals in the first non-dominated level F1 are extracted to form the non-dominated optimal solution set.

[0060] In step S30 of the method provided in this application embodiment, selecting the optimal solution from the set of non-dominated optimal solutions according to the current operating mode as the optimal control strategy at the current moment includes: When the current operating mode is the economic priority mode, the solution with the highest operating efficiency index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment. The current operating mode can be automatically switched according to the real-time operating status. When the current operating mode is the low-carbon priority mode, the solution with the highest energy utilization rate in the non-dominated optimal solution set is selected as the optimal control strategy at the current moment. When the current operating mode is the safety-first mode, the solution with the highest system stability index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment.

[0061] In this embodiment, firstly, when the current operating mode is set or automatically switched to the economic priority mode, the optimization algorithm sorts all candidate solutions from the non-dominated optimal solution set according to the operating efficiency index, and then selects the solution with the largest operating efficiency index value as the current optimal control strategy. Simultaneously, the current operating mode automatically switches according to the real-time operating status. For example, when peak electricity price periods are detected, the mode switches from low-carbon priority mode to economic priority mode to maximize electricity-saving benefits.

[0062] Secondly, when the current operating mode is low-carbon priority mode, the optimization algorithm sorts all candidate solutions in the non-dominated optimal solution set according to the energy utilization rate index. The solution with the highest energy utilization rate index value is selected as the current optimal control strategy. This can minimize the waste caused by renewable energy being fed back into the grid, thereby reducing the carbon emissions of the station purchasing electricity from the external grid.

[0063] Ultimately, when the current operating mode is the safety-first mode, the optimization algorithm sorts all candidate solutions in the non-dominated optimal solution set according to the system stability index, selects the solution with the highest system stability index as the optimal control strategy at the current moment, ensures the safe and reliable operation of the power supply system, ensures that the bus voltage fluctuation is minimized and the energy storage state of charge percentage operates in the healthy range, and avoids equipment failures or protection actions caused by overcharging, over-discharging or drastic voltage fluctuations.

[0064] In this embodiment, data from the energy storage system is collected in real time through an industrial IoT platform. Then, a mathematical model with three optimization objectives is constructed, and a multi-objective optimization algorithm is used to find the multi-objective optimization. At the same time, based on the current operating mode, the optimal solution is selected as the control strategy at the current moment, which significantly improves the flexibility of decision-making and the optimization effect, and provides scientific and executable decision support for coordinated control under different operating scenarios.

[0065] S40: The energy storage system, the flexible load in the station and the grid connection interface are coordinated and controlled according to the optimal control strategy. The rolling time domain optimization framework is adopted to repeatedly execute the optimization control steps at each preset period to achieve dynamic update control.

[0066] Step S40 of the method provided in this application embodiment, wherein the optimal control strategy includes energy storage charging and discharging power, flexible load regulation amount, and grid-connected switching power, includes: Setting the energy storage charging and discharging power to a positive value indicates that the energy storage system is in a charging state, setting it to a negative value indicates that the energy storage system is in a discharging state, and setting it to zero indicates that the energy storage system is in a standby state. The flexible load adjustment amount is set to include the temperature setpoint adjustment amount of the air conditioning system, the illuminance adjustment amount of the lighting system, and the speed adjustment amount of the escalator and elevator, and each flexible load adjustment amount is taken within a preset adjustable range. Setting the grid-connected power exchange to a positive value indicates purchasing electricity from the grid, setting it to a negative value indicates feeding back energy to the grid, and setting it to zero indicates no power exchange with the grid. The sum of the energy storage charging and discharging power, the flexible load adjustment amount, and the grid-connected switching power satisfies the power balance relationship with the current value of the braking energy time series and the current value of the station load time series.

[0067] In this embodiment, firstly, setting the energy storage charging and discharging power to a positive value indicates that the energy storage system is in a charging state. When it is positive, the converter controls the power switching device to draw electrical energy from the DC bus to charge the battery pack. When it is negative, the converter controls the battery pack to release electrical energy to the bus. Setting it to a negative value indicates that the energy storage system is in a discharging state. Setting it to zero indicates that the energy storage system is in a standby state. When it is zero, the converter locks the power device, so that the energy storage system is electrically isolated or in a high-resistance state from the bus, enters the standby mode, and reduces its own power consumption.

[0068] Specifically, based on the optimal control strategy for the energy storage system, the industrial IoT platform sends energy storage charging and discharging power commands to the energy storage system controller, controlling the energy storage system to charge or discharge according to the energy storage charging and discharging power in the optimal control strategy; the industrial IoT platform also sends flexible load adjustment commands to the station environmental control system, adjusting the temperature setpoint of the air conditioning system, the illuminance value of the lighting system, and the operating speed of escalators and elevators according to the flexible load adjustment amount in the optimal control strategy.

[0069] The grid-connected power exchange command is issued to the grid-connected converter through the industrial IoT platform. According to the grid-connected power exchange in the optimal control strategy, the power exchange between the station and the power grid is controlled. While executing the control command, each actuator feeds back the real-time operation data to the collaborative control optimization model through the industrial IoT platform for optimization calculation in the next preset cycle.

[0070] Secondly, the flexible load adjustment parameters are set to include adjustments for the air conditioning system temperature setpoint, lighting system illuminance, and escalator / elevator speed. The air conditioning system temperature setpoint adjustment changes the instantaneous power. The lighting system illuminance adjustment changes the luminous flux output of the luminaires, adjusting the power of the lighting circuit. The escalator / elevator speed adjustment changes the escalator speed. All adjustments must be within a preset adjustable range, for example, the air conditioning temperature adjustment range is -2℃ to +2℃. Adjustments exceeding this range will be truncated or automatically excluded by the optimization algorithm to ensure that the adjustment process does not reduce passenger comfort.

[0071] Furthermore, setting the grid-connected exchange power to a positive value indicates that electricity is purchased from the grid, and the converter will transmit power from the grid side to the DC bus side to meet the load demand within the station; setting it to a negative value indicates that power is fed back to the grid, and the converter will control the transmission of energy from the bus side to the grid side, transmitting excess regenerative energy or energy storage power back to the grid; setting it to zero indicates that there is no power exchange with the grid, and the load within the station is supported by regenerative braking energy and the energy storage system.

[0072] Ultimately, the sum of the energy storage charging and discharging power, the flexible load regulation, and the grid-connected switching power, along with the current values ​​of the braking energy time series and the station load time series, satisfies a power balance relationship. This relationship is enforced as an equality constraint, ensuring that any optimal solution is physically achievable. The power balance relationship is: P + P 交换 +ΔP=P 制动能量时间 +P 负荷时间 .

[0073] Step S40 in the method provided in this application embodiment further includes: The prediction time domain length is set to 10 seconds, the control time domain length is set to 2 seconds, and the optimization period is set to 0.5 seconds, which are used as the preset period. At the end of each optimization cycle, the latest train operation data, station environment data, train timetable data, and current energy storage status data are reacquired. Based on the updated data, the braking energy time series analysis, station load time series analysis, and multi-objective optimization solution are re-executed to obtain the updated optimal control strategy. Among them, only the first control command corresponding to the current time in the updated optimal control strategy is executed, and the other control commands are not executed for the time being.

[0074] In this embodiment, firstly, a preset period is set, wherein the prediction time domain length is set to 10 seconds to consider the typical duration of train braking upon entering the station, the response time of load changes within the station, and the capacity of computing resources; the control time domain length is set to 2 seconds, indicating that the control time is multiple moments within the next 2 seconds; the optimization period is set to 0.5 seconds, and then an optimization calculation is triggered every 0.5 seconds to generate new control commands.

[0075] Secondly, based on the optimization cycle, an optimization task is triggered every 0.5 seconds. When the optimization cycle is reached, a new round of data acquisition is initiated. Using an industrial IoT platform, train operation data, temperature, humidity, and light intensity data, personnel density data, timetable information, and current energy storage status data are reacquired. This newly acquired data serves as input for the next round of prediction and optimization, ensuring that control decisions align with the current actual operating conditions.

[0076] Finally, after completing a new round of optimization, the updated optimal control strategy is obtained by re-executing braking energy time series analysis, station load time series analysis, and multi-objective optimization. The first control command at the current moment is then executed, while the remaining control commands are not executed. When the next optimization cycle arrives, the latest data is acquired again, predictions are re-made, and optimization is re-optimized. This process is repeated cyclically to achieve rolling updates to the control strategy.

[0077] In this embodiment, the symbols for energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are defined, and the flexible load regulation is specified to ensure the accurate execution of control commands. Power balance constraints ensure the feasibility of optimization results. Subsequently, a rolling time-domain optimization framework is adopted to correct the dynamic changes of the system, effectively compensate for the deviation between the prediction model and the actual operating conditions, improve the real-time performance and control accuracy of the control system, and dynamically maintain the supply and demand balance.

[0078] The embodiments of this application, through the above specific implementation methods, achieve the following technical effects: In this embodiment, operational data is first collected in real time on an industrial IoT platform, and a braking energy prediction model is constructed to predict the regenerative braking power within a preset time period in the future. This provides an accurate supply boundary for subsequent coordinated regulation and optimization, improves the adaptability and prediction accuracy of the prediction model to different operating conditions, and lays a data foundation for subsequent analysis.

[0079] Secondly, by collecting environmental data and passenger flow forecast data in real time through the industrial IoT platform, and conducting model analysis through the station load forecasting model, the accuracy and interpretability of load forecasting are improved by comprehensively considering the influence of multiple factors, providing high-quality input data for subsequent analysis.

[0080] Furthermore, data from the energy storage system is collected in real time through an industrial IoT platform. Subsequently, a mathematical model with three optimization objectives is constructed, and a multi-objective optimization algorithm is used to find the optimal solution for each objective. At the same time, based on the current operating mode, the optimal solution is selected as the control strategy at the current moment, which significantly improves the flexibility of decision-making and the optimization effect. This provides scientific and executable decision support for coordinated control under different operating scenarios.

[0081] Finally, the symbolic meanings of energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are defined, and the flexible load regulation is specified to ensure the accurate execution of control commands. The feasibility of optimization results is guaranteed by power balance constraints. Subsequently, a rolling time-domain optimization framework is adopted to correct the dynamic changes of the system, effectively compensate for the deviation between the prediction model and the actual operating conditions, improve the real-time performance and control accuracy of the control system, and dynamically maintain the supply and demand balance.

[0082] Example 2, as Figure 2 As shown, based on the same inventive concept as the building energy optimization method based on the Industrial Internet of Things provided in Embodiment 1, this embodiment of the invention also provides a building energy optimization system based on the Industrial Internet of Things, including: The energy time series generation module 11 is used to obtain the braking energy time series within a preset future time window by analyzing the operating data during the train's braking process at the station through a braking energy prediction model. The load time series generation module 12 is used to obtain the station load time series within the preset future time window based on station environment data and train timetable data, through the station load prediction model. The optimal control strategy acquisition module 13 is used to acquire the current energy storage status data of the energy storage system, input the braking energy time series, the station load time series and the current energy storage status data into the collaborative control optimization model, and solve the optimal control strategy at the current moment through a multi-objective optimization algorithm. The optimal control strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. The regulation strategy regulation module 14 is used to coordinate the regulation of the energy storage system, the flexible load in the station and the grid connection interface according to the optimal regulation strategy, and adopts a rolling time domain optimization framework to repeatedly execute the optimization regulation steps at a preset period to achieve dynamic update regulation.

[0083] In one embodiment, the energy time series generation module 11 is used for: The instantaneous speed, initial speed at the moment of entry, braking distance, number of train formations, real-time load status, and braking mode type of the train are collected in real time through an industrial Internet of Things platform, and the collected data are used as the operation data. The operational data is input into a pre-trained braking energy prediction model, wherein the braking energy prediction model is analyzed and calculated based on the principle of train kinetic energy change and the correction coefficients fitted from historical braking data. The braking energy prediction model outputs time series data of regenerative braking power changing over time from the current moment to a preset future duration, which is used as the braking energy time series.

[0084] In one embodiment, the load time series generation module 12 is used for: The station's outdoor temperature and humidity data, outdoor light intensity data, real-time personnel density data in various areas of the station, and current date data are collected in real time through an industrial IoT platform, and the collected data are used as the station's environmental data. Access the train timetable system to obtain the train arrival and departure timetables and passenger flow forecast data for the corresponding platforms within a preset time period in the future, and use the train arrival and departure timetables and passenger flow forecast data as the train timetable data; The station environment data and train timetable data are input into a pre-built station load prediction model, wherein the station load prediction model analyzes and calculates the time series of air conditioning system load, lighting system load, escalator system load, and other equipment load. The load time series of the air conditioning system, the lighting system, the escalator system, and the other equipment are superimposed to obtain the load time series within the main station, which is used as the load time series within the station.

[0085] In one embodiment, the optimal control strategy acquisition module 13 is used for: The current state of charge percentage, current health percentage, current maximum allowable charging power, current maximum allowable discharging power, and current available capacity of the energy storage system are collected in real time through the industrial IoT platform as the current energy storage status data.

[0086] In one embodiment, the optimal control strategy acquisition module 13 is used for: A mathematical model is constructed with the optimization objectives of maximizing energy utilization, maximizing operational efficiency, and maximizing system stability. Here, the energy utilization rate is defined as the proportion of regenerative braking energy effectively utilized within the preset future time window to the total energy of the braking energy time series. The operational efficiency is defined as the net benefit after deducting the aging loss cost and energy conversion loss cost of the energy storage system from the electricity purchase cost saved within the preset future time window. The system stability is defined as a stability index constructed based on the bus voltage deviation value and the energy storage state of charge deviation value. The constraints for optimization are set, including the allowable fluctuation range of bus voltage, the allowable range of energy storage state of charge, the limit of energy storage charging and discharging power, the limit of flexible load adjustment, and the power balance constraint. Based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives. The optimal solution is then selected from the set of non-dominated optimal solutions according to the current operating mode, and is used as the optimal control strategy at the current moment.

[0087] Based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives, including: The energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are used as decision variables. Combined sampling is performed within the value range of the decision variables to generate an initial candidate solution population. Based on the three optimization objectives of energy utilization, operational efficiency, and system stability, the three objective function values ​​corresponding to each candidate solution in the initial candidate solution population are calculated respectively. The candidate solutions in the initial candidate solution population are sorted in a non-dominated order, and the candidate solutions are divided into multiple non-dominated levels. In this order, a candidate solution in any non-dominated level is not dominated by other candidate solutions in the same level, and is only dominated by candidate solutions in higher levels. Starting from the first non-dominated level, candidate solutions from each non-dominated level are selected sequentially to enter the new generation population until the number of selected candidate solutions reaches the preset population size. The new generation population is evolved through crossover and mutation operations to generate an evolved candidate solution population, and the three objective function values ​​corresponding to each candidate solution are recalculated. Repeat the non-dominated sorting, selection, crossover, and mutation operations until the preset maximum number of iterations is reached. All candidate solutions in the first non-dominated level are then used as the non-dominated optimal solution set.

[0088] Among them, selecting the optimal solution from the set of non-dominated optimal solutions according to the current operating mode as the optimal control strategy at the current moment includes: When the current operating mode is the economic priority mode, the solution with the highest operating efficiency index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment. The current operating mode can be automatically switched according to the real-time operating status. When the current operating mode is the low-carbon priority mode, the solution with the highest energy utilization rate in the non-dominated optimal solution set is selected as the optimal control strategy at the current moment. When the current operating mode is the safety-first mode, the solution with the highest system stability index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment.

[0089] In one embodiment, the control strategy control module 14 is used for: Setting the energy storage charging and discharging power to a positive value indicates that the energy storage system is in a charging state, setting it to a negative value indicates that the energy storage system is in a discharging state, and setting it to zero indicates that the energy storage system is in a standby state. The flexible load adjustment amount is set to include the temperature setpoint adjustment amount of the air conditioning system, the illuminance adjustment amount of the lighting system, and the speed adjustment amount of the escalator and elevator, and each flexible load adjustment amount is taken within a preset adjustable range. Setting the grid-connected power exchange to a positive value indicates purchasing electricity from the grid, setting it to a negative value indicates feeding back energy to the grid, and setting it to zero indicates no power exchange with the grid. The sum of the energy storage charging and discharging power, the flexible load adjustment amount, and the grid-connected switching power satisfies the power balance relationship with the current value of the braking energy time series and the current value of the station load time series.

[0090] In one embodiment, the control strategy control module 14 is further configured to: The prediction time domain length is set to 10 seconds, the control time domain length is set to 2 seconds, and the optimization period is set to 0.5 seconds, which are used as the preset period. At the end of each optimization cycle, the latest train operation data, station environment data, train timetable data, and current energy storage status data are reacquired. Based on the updated data, the braking energy time series analysis, station load time series analysis, and multi-objective optimization solution are re-executed to obtain the updated optimal control strategy. Among them, only the first control command corresponding to the current time in the updated optimal control strategy is executed, and the other control commands are not executed for the time being.

[0091] Compared with existing technologies, this application first collects operational data in real time on an industrial Internet of Things platform and constructs a braking energy prediction model to predict the regenerative braking power within a preset time period in the future. This provides an accurate supply boundary for subsequent coordinated control and optimization, improves the adaptability and prediction accuracy of the prediction model to different working conditions, and lays a data foundation for subsequent analysis.

[0092] Secondly, by collecting environmental data and passenger flow forecast data in real time through the industrial IoT platform, and conducting model analysis through the station load forecasting model, the accuracy and interpretability of load forecasting are improved by comprehensively considering the influence of multiple factors, providing high-quality input data for subsequent analysis.

[0093] Furthermore, data from the energy storage system is collected in real time through an industrial IoT platform. Subsequently, a mathematical model with three optimization objectives is constructed, and a multi-objective optimization algorithm is used to find the optimal solution for each objective. At the same time, based on the current operating mode, the optimal solution is selected as the control strategy at the current moment, which significantly improves the flexibility of decision-making and the optimization effect. This provides scientific and executable decision support for coordinated control under different operating scenarios.

[0094] Finally, the symbolic meanings of energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are defined, and the flexible load regulation is specified to ensure the accurate execution of control commands. The feasibility of optimization results is guaranteed by power balance constraints. Subsequently, a rolling time-domain optimization framework is adopted to correct the dynamic changes of the system, effectively making up for the deviation between the prediction model and the actual operating conditions, improving the real-time performance and control accuracy of the control system, and dynamically maintaining the supply and demand balance.

Claims

1. A building energy consumption optimization method based on the Industrial Internet of Things, characterized in that, The methods include: Based on the operational data during the train's braking process at the station, the braking energy time series within a preset future time window is obtained through braking energy prediction model analysis. Based on station environment data and train timetable data, the station load time series within the preset future time window is obtained through the station load prediction model. The current energy storage status data of the energy storage system is obtained, and the braking energy time series, the station load time series and the current energy storage status data are input into the collaborative regulation and optimization model. The optimal regulation strategy at the current moment is solved by a multi-objective optimization algorithm. The optimal regulation strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. The energy storage system, on-site flexible loads, and grid connection interface are coordinated and controlled according to the optimal control strategy. A rolling time-domain optimization framework is adopted to repeatedly execute the optimization control steps at preset intervals to achieve dynamic update control.

2. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, Based on the operational data during the train's braking process at the station, the braking energy time series within a preset future time window is obtained through braking energy prediction model analysis, including: The instantaneous speed, initial speed at the moment of entry, braking distance, number of train formations, real-time load status, and braking mode type of the train are collected in real time through an industrial Internet of Things platform, and the collected data are used as the operation data. The operational data is input into a pre-trained braking energy prediction model, wherein the braking energy prediction model is analyzed and calculated based on the principle of train kinetic energy change and the correction coefficients fitted from historical braking data. The braking energy prediction model outputs time series data of regenerative braking power changing over time from the current moment to a preset future duration, which is used as the braking energy time series.

3. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, Based on station environment data and train timetable data, the station load time series within the preset future time window is obtained through a station load prediction model, including: The station's outdoor temperature and humidity data, outdoor light intensity data, real-time personnel density data in various areas of the station, and current date data are collected in real time through an industrial IoT platform, and the collected data are used as the station's environmental data. Access the train timetable system to obtain the train arrival and departure timetables and passenger flow forecast data for the corresponding platforms within a preset time period in the future, and use the train arrival and departure timetables and passenger flow forecast data as the train timetable data; The station environment data and train timetable data are input into a pre-built station load prediction model, wherein the station load prediction model analyzes and calculates the time series of air conditioning system load, lighting system load, escalator system load, and other equipment load. The load time series of the air conditioning system, the lighting system, the escalator system, and the other equipment are superimposed to obtain the load time series within the main station, which is used as the load time series within the station.

4. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, The current state of charge percentage, current health percentage, current maximum allowable charging power, current maximum allowable discharging power, and current available capacity of the energy storage system are collected in real time through the industrial IoT platform as the current energy storage status data.

5. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, The braking energy time series, station load time series, and current energy storage state data are input into the collaborative regulation and optimization model. A multi-objective optimization algorithm is used to solve for the optimal regulation strategy at the current moment, including: A mathematical model is constructed with the optimization objectives of maximizing energy utilization, maximizing operational efficiency, and maximizing system stability. Here, the energy utilization rate is defined as the proportion of regenerative braking energy effectively utilized within the preset future time window to the total energy of the braking energy time series. The operational efficiency is defined as the net benefit after deducting the aging loss cost and energy conversion loss cost of the energy storage system from the electricity purchase cost saved within the preset future time window. The system stability is defined as a stability index constructed based on the bus voltage deviation value and the energy storage state of charge deviation value. The constraints for optimization are set, including the allowable fluctuation range of bus voltage, the allowable range of energy storage state of charge, the limit of energy storage charging and discharging power, the limit of flexible load adjustment, and the power balance constraint. Based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives. The optimal solution is then selected from the set of non-dominated optimal solutions according to the current operating mode, and is used as the optimal control strategy at the current moment.

6. The building energy optimization method based on the Industrial Internet of Things according to claim 5, characterized in that, Based on the mathematical model, a multi-objective optimization problem is solved under the constraints to obtain a set of non-dominated optimal solutions that satisfy the optimization objectives, including: The energy storage charging and discharging power, flexible load regulation, and grid-connected switching power are used as decision variables. Combined sampling is performed within the value range of the decision variables to generate an initial candidate solution population. Based on the three optimization objectives of energy utilization, operational efficiency, and system stability, the three objective function values ​​corresponding to each candidate solution in the initial candidate solution population are calculated respectively. The candidate solutions in the initial candidate solution population are sorted in a non-dominated order, and the candidate solutions are divided into multiple non-dominated levels. In any non-dominated level, the candidate solutions are not dominated by other candidate solutions in the same level, and are only dominated by candidate solutions in higher levels. Starting from the first non-dominated level, candidate solutions from each non-dominated level are selected sequentially to enter the new generation population until the number of selected candidate solutions reaches the preset population size. The new generation population is evolved through crossover and mutation operations to generate an evolved candidate solution population, and the three objective function values ​​corresponding to each candidate solution are recalculated. Repeat the non-dominated sorting, selection, crossover, and mutation operations until the preset maximum number of iterations is reached. All candidate solutions in the first non-dominated level are then used as the non-dominated optimal solution set.

7. The building energy optimization method based on the Industrial Internet of Things according to claim 5, characterized in that, Based on the current operating mode, the optimal solution is selected from the set of non-dominated optimal solutions as the optimal control strategy for the current moment, including: When the current operating mode is the economic priority mode, the solution with the highest operating efficiency index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment. The current operating mode can be automatically switched according to the real-time operating status. When the current operating mode is the low-carbon priority mode, the solution with the highest energy utilization rate in the non-dominated optimal solution set is selected as the optimal control strategy at the current moment. When the current operating mode is the safety-first mode, the solution with the highest system stability index is selected from the set of non-dominated optimal solutions as the optimal control strategy at the current moment.

8. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, The optimal control strategy includes energy storage charging and discharging power, flexible load regulation, and grid-connected switching power, including: Setting the energy storage charging and discharging power to a positive value indicates that the energy storage system is in a charging state, setting it to a negative value indicates that the energy storage system is in a discharging state, and setting it to zero indicates that the energy storage system is in a standby state. The flexible load adjustment amount is set to include the temperature setpoint adjustment amount of the air conditioning system, the illuminance adjustment amount of the lighting system, and the speed adjustment amount of the escalator and elevator, and each flexible load adjustment amount is taken within a preset adjustable range. Setting the grid-connected power exchange to a positive value indicates purchasing electricity from the grid, setting it to a negative value indicates feeding back energy to the grid, and setting it to zero indicates no power exchange with the grid. The sum of the energy storage charging and discharging power, the flexible load adjustment amount, and the grid-connected switching power satisfies the power balance relationship with the current value of the braking energy time series and the current value of the station load time series.

9. The building energy optimization method based on the Industrial Internet of Things according to claim 1, characterized in that, A rolling time-domain optimization framework is adopted, and optimization and control steps are repeatedly executed at preset intervals, including: The prediction time domain length is set to 10 seconds, the control time domain length is set to 2 seconds, and the optimization period is set to 0.5 seconds, which are used as the preset period. At the end of each optimization cycle, the latest train operation data, station environment data, train timetable data, and current energy storage status data are reacquired. Based on the updated data, the braking energy time series analysis, station load time series analysis, and multi-objective optimization solution are re-executed to obtain the updated optimal control strategy. Among them, only the first control command corresponding to the current time in the updated optimal control strategy is executed, and the other control commands are not executed for the time being.

10. A building energy optimization system based on the Industrial Internet of Things, characterized in that, The system is used to implement the building energy optimization method based on the Industrial Internet of Things as described in any one of claims 1-9, the system comprising: The energy time series generation module is used to obtain the braking energy time series within a preset future time window based on the operating data of the train entering the station braking process and through the braking energy prediction model. The load time series generation module is used to obtain the station load time series within the preset future time window based on station environment data and train timetable data through the station load prediction model; The optimal control strategy acquisition module is used to acquire the current energy storage status data of the energy storage system, input the braking energy time series, the station load time series and the current energy storage status data into the collaborative control optimization model, and solve the optimal control strategy at the current moment through a multi-objective optimization algorithm. The optimal control strategy includes energy storage charging and discharging power, flexible load adjustment amount and grid-connected switching power. The control strategy control module is used to coordinate the control of the energy storage system, the flexible load in the station and the grid connection interface according to the optimal control strategy. It adopts a rolling time domain optimization framework and repeatedly executes the optimization control steps at preset intervals to achieve dynamic update control.

Images