A method and system for adaptive control of heating based on microclimate prediction

Abstract

The application discloses a kind of heating adaptive control method and system based on microclimate prediction, it is related to heating control technical field, the method includes: obtaining microclimate prediction data in future period, based on predicted solar radiation intensity sequence and predicted outdoor temperature sequence, obtain predicted heat storage sequence, based on predicted outdoor temperature sequence and predicted relative humidity sequence, obtain predicted energy efficiency coefficient sequence, according to predicted heat storage sequence, the target heat that needs to be supplemented by second heating unit is calculated, and target heat is converted into the operation plan of second heating unit in future period, real-time correction is carried out to operation plan, generates execution instruction and controls second heating unit to carry out heating.The technical problem that the prior art cannot adjust heating strategy in advance, resulting in high heating energy consumption and model control performance decline is solved.

Description

Technical Field

[0001] This application relates to the field of heating control technology, specifically to a heating adaptive control method and system based on microclimate prediction. Background Technology

[0002] With the advancement of urbanization and the development of prefabricated building technology, the application rate of prefabricated buildings in cold regions is increasing year by year. Compared with traditional cast-in-place concrete buildings, prefabricated buildings mostly use prefabrication and assembly processes for their wall components. Their thermal inertia parameters and heat transfer characteristics are more clearly defined, but this also places higher demands on the dynamic adjustment capabilities of the heating control system. Especially in severely cold regions, where outdoor temperature fluctuations are large and solar radiation intensity varies significantly, the heat storage and release process of the building envelope itself has a significant impact on the indoor thermal environment. If changes in the building's heat storage state cannot be accurately predicted, it can easily lead to excessive indoor temperature fluctuations and unnecessary waste of heating energy.

[0003] However, most existing heating control methods for buildings in cold regions employ on / off control or traditional PID control based on fixed indoor temperature setpoints. These methods adjust only based on current indoor temperature feedback, failing to anticipate future microclimate changes and building thermal storage characteristics to adjust heating strategies, and thus failing to fully utilize free heat sources such as solar energy to reduce heating energy consumption. Some heating control methods that incorporate predictive control also fail to adaptively adjust the operating plan based on the building's real-time thermal inertia, resulting in discrepancies between the optimized operating plan and actual heating demand, making it difficult to achieve the expected control accuracy and energy-saving effects.

[0004] In addition, most existing control methods lack a closed-loop parameter update mechanism, and the model parameters cannot be automatically corrected as the thermal properties of components change slowly during building use. As a result, the control performance will gradually decline after long-term operation. Summary of the Invention

[0005] This application provides a heating adaptive control method and system based on microclimate prediction, which solves the technical problem that the existing technology cannot adjust the heating strategy in advance, resulting in high heating energy consumption and decreased model control performance.

[0006] The technical solution to the above-mentioned technical problems in this application is as follows: In a first aspect, this application provides a heating adaptive control method based on microclimate prediction, the method comprising: Obtain microclimate prediction data for future time periods, wherein the microclimate prediction data includes predicted solar radiation intensity sequences, predicted outdoor temperature sequences, and predicted relative humidity sequences; Based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence, the predicted stored heat sequence of the first heating unit in the future time period is obtained. Based on the predicted outdoor temperature sequence and the predicted relative humidity sequence, the predicted energy efficiency coefficient sequence of the second heating unit in the future time period is obtained. Constrained by the preset heating conditions of prefabricated buildings in cold regions, the target heat that needs to be supplemented by the second heating unit is calculated based on the predicted stored heat sequence, and the target heat is converted into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. Based on real-time temperature feedback and building thermal inertia, an adaptive control algorithm is used to correct the operation plan in real time and generate execution instructions to control the second heating unit to provide heating.

[0007] Secondly, this application provides a heating adaptive control system based on microclimate prediction, comprising: The prediction data acquisition module is used to acquire microclimate prediction data for future periods, wherein the microclimate prediction data includes a predicted solar radiation intensity sequence, a predicted outdoor temperature sequence, and a predicted relative humidity sequence. A stored heat sequence prediction module is used to predict and obtain the predicted stored heat sequence of the first heating unit in the future time period based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence. An energy efficiency coefficient sequence prediction module is used to predict and obtain the predicted energy efficiency coefficient sequence of the second heating unit in the future time period based on the predicted outdoor temperature sequence and the predicted relative humidity sequence. The operation plan generation module is used to calculate the target heat that needs to be supplemented by the second heating unit based on the predicted stored heat sequence, constrained by the preset heating conditions of the cold-region prefabricated building, and to convert the target heat into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. The operation plan correction module is used to correct the operation plan in real time based on real-time temperature feedback and building thermal inertia state using an adaptive control algorithm, and generate execution instructions to control the second heating unit to provide heating.

[0008] This application provides one or more technical solutions, which have at least the following technical effects or advantages: This application provides a heating adaptive control method and system based on microclimate prediction. First, it acquires microclimate prediction data for future periods to predict the impact of the outdoor environment on the indoor thermal environment. Combined with the predicted stored heat sequence of the first heating unit, it calculates the heat that free heat sources such as solar energy can provide. Based on this, it plans the operation of the second heating unit to maximize the utilization of free heat sources and reduce unnecessary energy consumption. Second, after generating the initial operation plan, this application uses real-time temperature feedback and building thermal inertia to correct the plan in real time, offsetting the deviation between the prediction results and actual needs, improving the control accuracy of indoor temperature, and avoiding excessive temperature fluctuations. Finally, a closed-loop feedback update mechanism is set up to continuously correct model parameters based on actual heating response results, adapting to the slow changes in the thermal properties of building components over time. This ensures stable control performance even after long-term operation, solving the problem of performance degradation after long-term operation in traditional methods.

[0009] Through the above technical solution, this application can reduce the energy consumption of the second heating unit by making full use of natural heat sources such as solar energy while ensuring the indoor thermal comfort of prefabricated buildings in cold regions. At the same time, by combining adaptive correction of the building's real-time thermal inertia and closed-loop rolling model parameter updates, it can maintain high control accuracy and energy-saving effect in the long term. Compared with traditional control methods, it can effectively reduce heating energy consumption, improve the stability of the indoor thermal environment, and meet the heating control needs of prefabricated buildings in severely cold regions. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating a heating adaptive control method based on microclimate prediction provided in an embodiment of this application. Figure 2 This is a schematic diagram of a heating adaptive control system based on microclimate prediction provided in an embodiment of this application.

[0012] The components represented by each number in the attached diagram are explained below: The system includes a prediction data acquisition module 11, a heat sequence prediction storage module 12, an energy efficiency coefficient sequence prediction module 13, an operation plan generation module 14, and an operation plan correction module 15. Detailed Implementation

[0013] This application provides a heating adaptive control method and system based on microclimate prediction, which addresses the technical problem that existing technologies cannot adjust heating strategies in advance, resulting in high heating energy consumption and decreased model control performance.

[0014] Example 1, as Figure 1 As shown in the figure, this application provides a heating adaptive control method based on microclimate prediction, including: S10: Obtain microclimate prediction data for future periods, wherein the microclimate prediction data includes predicted solar radiation intensity sequence, predicted outdoor temperature sequence, and predicted relative humidity sequence; In this embodiment of the application, microclimate prediction data for future periods is obtained by first obtaining forecast data released by a publicly available meteorological service platform, including forecast solar intensity sequence, forecast outdoor temperature sequence and forecast relative humidity sequence for future periods; and then obtaining measured data collected by micro-meteorological stations pre-deployed in the roof area of ​​prefabricated buildings in cold regions, including measured solar intensity sequence, measured outdoor temperature sequence and measured relative humidity sequence. Secondly, the acquired forecast data and measured data are aligned with the same timestamp, and missing values ​​in the data sequence are filled using interpolation. Then, the Kalman filter algorithm is used to fuse the forecast data and measured data to generate microclimate prediction data. The Kalman filter algorithm is a recursive algorithm for optimal estimation of noisy observation data. By continuously correcting the prediction values, the error inherent in the forecast data itself is reduced, and finally, more accurate microclimate prediction data is obtained. In the Kalman filter algorithm, the fusion weight coefficient of the forecast data decreases as the prediction time window extends.

[0015] For example, the prediction time window is the next 12 hours, and the microclimate prediction data is updated every 120 minutes to ensure that the prediction data can match the actual changing trend of the outdoor microclimate in a timely manner.

[0016] S20: Based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence, predict and obtain the predicted stored heat sequence of the first heating unit in the future time period; The first heating unit is a solar energy storage and heat collection unit for prefabricated buildings in cold regions. In this embodiment, based on the equivalent thermal parameter model of the prefabricated building envelope, the model's initial state is assigned values ​​by combining the measured indoor temperature and measured wall surface temperature collected at the current moment. Then, the predicted solar intensity sequence and predicted outdoor temperature sequence are substituted into the model, and the hourly temperature of the building envelope's heat storage body in the future period is obtained through recursive calculation. The hourly stored heat is then calculated by combining the heat storage body's heat capacity parameters. After sorting, the predicted stored heat sequence is obtained. This sequence can intuitively reflect the changes in the heat absorbed and stored by the building envelope through solar radiation in the future period. This part of the heat can be naturally released to maintain the indoor temperature and is a usable free heat source.

[0017] Specifically, step S20 in the method includes: Establish an equivalent thermal parameter model for the first heating unit; The predicted solar radiation intensity sequence is used as the solar radiation heat flow input of the equivalent thermal parameter model, the predicted outdoor temperature sequence is used as the boundary temperature input of the equivalent thermal parameter model, and the heat storage body heat capacity parameter and thermal resistance parameter in the equivalent thermal parameter model are set as fixed parameters determined in advance through system identification experiments. Run the equivalent thermal parameter model to perform time-by-time simulation calculations and output the average temperature of the heat storage body of the first heating unit at each time in the future time period; The average temperature of the heat storage body is subtracted from the preset reference temperature value to obtain the heat storage body temperature difference sequence. Each temperature difference value in the heat storage body temperature difference sequence is multiplied by the heat capacity parameter of the heat storage body to obtain the instantaneous stored heat value at each time. The reference temperature value is the equilibrium temperature of the first heating unit in the state of completely releasing heat. Arrange the instantaneous stored heat values ​​at each moment in chronological order to generate the predicted stored heat sequence of the first heating unit in the future time period; The equivalent thermal parameter model adopts a three-thermal-resistance, two-heat-capacity model structure, and its construction method includes: The first heating unit is divided into two heat capacity nodes: a surface air layer and an internal heat storage wall layer. The surface air layer corresponds to the first heat capacity parameter, and the internal heat storage wall layer corresponds to the second heat capacity parameter. A first thermal resistance parameter is set between the outdoor environment and the surface air layer to characterize the first convective heat transfer resistance between the outdoor air and the surface air layer. A second thermal resistance parameter is provided between the surface air layer and the internal heat storage wall layer to characterize the thermal resistance between the surface air layer and the internal heat storage wall layer; A third thermal resistance parameter is set between the internal heat storage wall layer and the indoor environment to characterize the second convective heat transfer resistance between the internal heat storage wall layer and the indoor air. The predicted solar radiation intensity sequence is converted into a heat flow source term and applied to the surface air layer node, and the predicted outdoor temperature sequence is applied as a boundary condition to the outdoor environment node.

[0018] In this embodiment, firstly, a three-thermal-resistance, two-heat-capacity model structure is adopted to establish an equivalent thermal parameter model for the first heating unit. This equivalent thermal parameter model simplifies the complex unsteady-state heat conduction process of the building envelope into a lumped parameter model, which is used to describe the dynamic process of the first heating unit absorbing solar energy and storing heat. This reduces the amount of calculation while accurately describing the dynamic process of heat storage and release of the heat storage body.

[0019] Secondly, the predicted solar radiation intensity sequence is converted into a heat flow source term and input into the surface air layer node. At the same time, the predicted outdoor temperature sequence is used as a boundary condition and input into the outdoor environment node. The heat capacity and thermal resistance parameters of the heat storage body have been determined through previous system identification experiments and do not need to be adjusted in real time.

[0020] Next, the equivalent thermal parameter model is run to complete the recursive simulation calculation time by time to obtain the average temperature of the internal heat storage wall layer at each time in the future period, that is, the average temperature of the heat storage body. Then, the average temperature of the heat storage body at each time is subtracted from the reference equilibrium temperature under the condition of completely releasing heat to obtain the temperature difference of the heat storage body at the corresponding time. Then, the temperature difference is multiplied by the total heat capacity parameter of the heat storage body to calculate the instantaneous stored heat at that time. Finally, all the instantaneous stored heat is arranged in chronological order to obtain the predicted stored heat sequence of the first heating unit in the future period.

[0021] For example, assuming the current baseline equilibrium temperature is 10℃, the average temperature of the heat storage body calculated at a certain moment is 22℃, and the total heat capacity parameter of the heat storage body is 150kJ / ℃, then the temperature difference of the heat storage body at that moment is 22℃-10℃=12℃, and the corresponding instantaneous stored heat is 12℃×150kJ / ℃=1800kJ. After arranging the results calculated in this way for all moments in time, a complete predicted stored heat sequence can be obtained.

[0022] Furthermore, the specific construction process of the equivalent thermal parameter model is as follows: First, the first heating unit is divided into two heat capacity nodes: a surface air layer and an internal heat storage wall layer, corresponding to a first heat capacity parameter and a second heat capacity parameter, respectively. A first thermal resistance parameter is set between the outdoor environment and the surface air layer to characterize the convective heat transfer resistance between the outdoor air and the surface air layer. A second thermal resistance parameter is set between the surface air layer and the internal heat storage wall layer to characterize the solid thermal conductivity resistance between the two layers. A third thermal resistance parameter is set between the internal heat storage wall layer and the indoor environment to characterize the convective heat transfer resistance between the wall layer and the indoor air.

[0023] Secondly, differential equations are established for the two heat capacity nodes based on the principle of thermal balance, thus completing the construction of the equivalent thermal parameter model. For example, the equations are in the following form: , ; Where C1 is the first heat capacity parameter, C2 is the second heat capacity parameter, T1 is the surface air layer temperature, T2 is the internal heat storage wall layer temperature, and T o The outdoor temperature, T i Let R1 be the indoor temperature, R2 be the first thermal resistance parameter, R3 be the third thermal resistance parameter, q be the solar radiation intensity at the current moment, and S be the solar-receiving area of ​​the heat collector. Using this model, the temperature change of the heat storage body at each moment can be recursively calculated, thus yielding a predicted sequence of stored heat.

[0024] S30: Based on the predicted outdoor temperature sequence and the predicted relative humidity sequence, predict and obtain the predicted energy efficiency coefficient sequence of the second heating unit in the future time period; The second heating unit is a heat pump heating unit for prefabricated buildings in cold regions, and the heating equipment of the heat pump heating unit is an ultra-low temperature air source heat pump.

[0025] In this embodiment of the application, based on the measured data of the current outdoor temperature and relative humidity, and combined with the predicted energy efficiency coefficient model, the predicted outdoor temperature sequence and the predicted relative humidity sequence are used as inputs to output the predicted energy efficiency coefficient sequence.

[0026] The coefficient of performance (COP) of an ultra-low temperature air source heat pump is affected by both outdoor ambient temperature and relative humidity. When the outdoor temperature decreases, the evaporation temperature of the heat pump decreases, the air intake decreases, and the COP gradually decreases. When the outdoor relative humidity increases, the evaporator is prone to frost formation, which further reduces the heat exchange efficiency of the heat pump, leading to a decrease in the COP.

[0027] Specifically, step S30 in the method includes: Obtain the standard energy efficiency performance curve calibrated at the factory for the second heating unit; Collect actual operating data of the second heating unit during historical operating periods, including actual outdoor temperature, actual relative humidity, actual power consumption, and actual heat output. The actual energy efficiency coefficients at each historical moment are calculated based on the actual operating data, and the actual energy efficiency coefficients are compared with the benchmark values ​​of the standard energy efficiency performance curve under the same operating conditions to obtain the deviation sequence. The recursive least squares algorithm is used to fit the deviation sequence online and generate real-time correction coefficients. The online fitting process includes: initializing the parameter vector and covariance matrix; whenever a new deviation value is collected, calculating the gain vector based on the current covariance matrix and the input feature vector; iteratively updating the parameter vector and covariance matrix using the prediction error; and using the model coefficients in the updated parameter vector as real-time correction coefficients. The real-time correction coefficient is superimposed on the standard energy efficiency performance curve to obtain the corrected predicted energy efficiency coefficient model. The predicted outdoor temperature sequence and the predicted relative humidity sequence are used as inputs to output the predicted energy efficiency coefficient sequence of the second heating unit in the future time period.

[0028] In this embodiment of the application, firstly, the standard energy efficiency performance curve calibrated at the factory of the second heating unit is obtained. This curve describes the reference value of the energy efficiency coefficient corresponding to different outdoor temperatures and relative humidity under standard operating conditions.

[0029] Secondly, the outdoor temperature, relative humidity, power consumption and heat output of the heat pump heating unit are collected at each moment during actual operation. The actual energy efficiency coefficient is calculated by dividing the output heat output by the power consumption. The actual value is compared with the benchmark value under the same operating conditions given by the standard curve to obtain the deviation sequence.

[0030] Subsequently, a recursive least squares algorithm is used to fit the deviation sequence online. The fitting parameters are iteratively updated each time new operating data is acquired, resulting in real-time correction coefficients updated according to the actual operating state. These correction coefficients are then superimposed on the original standard energy efficiency performance curve to obtain a predicted energy efficiency coefficient model adapted to the current actual operating state of the heat pump. This model characterizes the dynamic relationship between the energy efficiency coefficient of the second heating unit and outdoor temperature and relative humidity. Finally, by inputting the predicted outdoor temperature and relative humidity sequences for future periods, the predicted energy efficiency coefficient sequence for each time period in the future can be output. The least squares algorithm is a method that finds the optimal parameter match by minimizing the sum of squared errors, while the recursive least squares algorithm can recursively calculate on newly acquired data, continuously updating parameter estimates, thus better adapting to the actual characteristic of the heat pump's operating state slowly decaying over time.

[0031] Specifically, the parameter vector and covariance matrix of the recursive least squares algorithm are first initialized. Whenever a new set of actual operating data is collected and the corresponding deviation value is calculated, the outdoor temperature and relative humidity at the current moment are used as the input feature vector. The gain vector is calculated based on the covariance matrix and the input feature vector at the current moment. The deviation value at the current moment is compared with the predicted value of the input feature vector and the parameter vector at the previous moment to obtain the prediction error. The gain vector is then multiplied by the prediction error to obtain the parameter correction amount. The parameter correction amount is then superimposed on the parameter vector at the previous moment to update the parameter vector at the current moment. Based on the updated covariance matrix and gain vector, the covariance matrix is ​​recursively updated, and the model coefficients in the parameter vector at the current moment are used as real-time correction coefficients.

[0032] For example, the corrected predicted energy efficiency coefficient model is obtained, assuming that the baseline energy efficiency coefficient of the standard energy efficiency performance curve at temperature t and relative humidity φ is... Where t is in °C and φ is in % . Assuming that 5 sets of historical operation data have been collected, resulting in 5 corresponding deviation values ​​ΔCOP1~ΔCOP5, initialize the parameter vector θ0=[0,0,0] for the recursive least squares algorithm. The covariance matrix P0 = 1000I, where I is a 3rd order identity matrix. Now, the 6th set of new actual operating data has been collected: actual outdoor temperature t6 = -10℃, actual relative humidity φ6 = 65%. Under this condition, the standard curve provides the baseline energy efficiency coefficient. The actual calculated coefficient of performance (COP) =1.92, therefore the deviation value for this test is... Input feature vector Calculate the gain vector k6 according to the recursive formula: Where λ is the forgetting factor, usually taken as 0.98, is calculated as follows: Then calculate the prediction error. The previous moment ,therefore , Update parameter vector Then update the covariance matrix. The updated covariance matrix is ​​obtained, completing this iteration.

[0033] Finally, the three elements of the obtained parameter vector θ are used as the constant term, temperature coefficient, and humidity coefficient, respectively, and the corrected predicted energy efficiency coefficient model is: Then, by substituting the predicted outdoor temperature t and predicted relative humidity φ for each time period in the future into the model, the predicted energy efficiency coefficient for the corresponding time period can be calculated. After sorting, the predicted energy efficiency coefficient sequence for the future period can be obtained.

[0034] S40: Based on the preset heating conditions of the cold-region prefabricated building, calculate the target heat that needs to be supplemented by the second heating unit according to the predicted stored heat sequence, and convert the target heat into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. In this embodiment of the application, the preset heating conditions refer to the set temperature range that the indoor air needs to maintain in the future period. The target heat that the second heating unit needs to supplement at each moment in the future period can be calculated based on this constraint.

[0035] Then, based on the target heat at each moment and the predicted energy efficiency coefficient at the corresponding moment, the operating power level of the second heating unit at each moment in the future period is determined. After sorting, a complete operating plan for the future period is obtained, realizing adaptive control based on microclimate prediction results.

[0036] By using the above method, the solar heat stored in the first heating unit can be fully utilized, reducing the energy consumption of the second heating unit, while ensuring that the indoor temperature always meets the preset heating requirements, avoiding situations where the indoor temperature is too low or too high, thus balancing the heating effect and energy-saving needs.

[0037] Specifically, step S40 in the method includes: The preset heating conditions include temperature constraints, start / stop constraints, and cost constraints. A rolling optimization framework is established using a model predictive control algorithm. The prediction time domain of the rolling optimization framework covers the future time period, and the control time domain is a subset of the prediction time domain. Within the rolling optimization framework, the preset heating conditions are used as hard constraints, and the goal is to minimize the operating cost; an optimization problem is then constructed. Using the predicted stored heat sequence as a known free heat source input, the difference between the indoor heat demand at each moment and the corresponding value in the predicted stored heat sequence is calculated to obtain the time-series distribution of the heat gap; The predicted energy efficiency coefficient sequence is substituted into the objective function as the time-varying efficiency parameter of the second heating unit, so that the optimization problem can identify the energy efficiency differences at different times. Solving the optimization problem yields the optimal start-stop state sequence and optimal operating power sequence of the second heating unit in the control time domain, which serve as the operating plan.

[0038] In this embodiment of the application, firstly, the temperature constraint condition in the preset heating conditions is set to the indoor temperature must be maintained within the comfort range set by the user. Specifically, the temperature constraint condition is that the indoor temperature should be maintained within the preset comfort temperature range. The preset comfort temperature range is defined by a lower limit comfort temperature value and an upper limit comfort temperature value. The lower limit comfort temperature value is used to prevent the indoor temperature from being too cold, and the upper limit comfort temperature value is used to prevent the indoor temperature from being too hot.

[0039] The start-stop constraint is that the number of start-stop cycles of the second heating unit within a unit of time does not exceed the preset maximum allowable number of start-stop cycles, which is determined according to the compressor protection requirements of the second heating unit. The cost constraint is that the electricity cost of the second heating unit adopts the time-of-use pricing strategy first. In the time-of-use pricing strategy, the operation tasks are allocated first during the off-peak hours and the shutdown or low-power operation is arranged first during the peak hours.

[0040] When each value in the predicted stored heat sequence individually meets the preset heating conditions, the target heat of the second heating unit is set to zero, and the second heating unit is kept in standby mode.

[0041] Subsequently, a rolling optimization framework was established using model predictive control algorithm, which allows the prediction time domain to cover the future period to be planned, and selects a portion of the prediction time domain as the real-time control interval. The model predictive control algorithm uses online rolling optimization to handle constrained multivariate control problems, and can combine the prediction information of the future period to optimize the control objective under the premise of satisfying various constraints, thus adapting to the thermal inertia hysteresis characteristics of the heating system.

[0042] Secondly, within the rolling optimization framework, the above three types of preset heating conditions are set as hard constraints that must be met in the optimization process. The goal function is to minimize the heating operation cost for the entire future period. A complete optimization problem is then constructed. The predicted stored heat sequence obtained by the first heating unit is then input into the optimization problem as a free heat source. The total heat required for the room to reach the set temperature at the current time is calculated time by time. The stored heat that the first heating unit can provide at the corresponding time is subtracted to obtain the heat gap at the current time. After sorting out the heat gaps at all times, a complete heat gap time series distribution is obtained.

[0043] Furthermore, the predicted energy efficiency coefficient sequence obtained from the second heating unit is substituted into the objective function as a time-varying parameter, enabling the objective function to accurately identify the differences in heat pump energy efficiency under different outdoor environments. Heat production is prioritized during periods of higher energy efficiency, further reducing operating costs. Numerical solutions to the optimization problem are then performed to obtain the optimal start-stop state and optimal operating power of the second heating unit at each moment within the control time domain. The processed results serve as the final operating plan to guide the real-time operation of the second heating unit. After each control time domain operation is completed, the predicted and control time domains are updated continuously, and the above optimization process is repeated, with the operating plan being revised periodically to address potential errors in the prediction process and improve control robustness.

[0044] For example, the objective function can be expressed as Where t0 is the current planning time, Np is the prediction time domain length, ΔQk is the heat gap that needs to be supplemented at time k, COPk is the predicted energy efficiency coefficient at time k, η is the pipeline efficiency of the heat pump heat exchange system, and pk is the time-of-use electricity price corresponding to time k. After substituting each parameter into the objective function, a quadratic programming algorithm is used for iterative solution to obtain the start-stop status and operating power of the second heating unit that minimizes the objective function J under all constraints. Finally, the optimal solution in the control time domain is extracted as the operating plan to be executed in the current period. The optimization process is repeated at the next time based on the updated prediction information to complete the rolling adjustment and ensure that the control plan always adapts to the latest prediction results.

[0045] The operation plan includes the target start / stop status, target operating power, and target outlet water temperature of the second heating unit at various times in the future period. The target start / stop status is a binary state variable, including both start-up and shutdown states. The target operating power is a continuous variable, taking values ​​between zero power and rated power, used to determine the specific load rate of the second heating unit in the start-up state. The target outlet water temperature is also a continuous variable, taking values ​​between the minimum safe outlet water temperature and the maximum allowable outlet water temperature, used to control the heating quality of the second heating unit. The target start / stop status, target operating power, and target outlet water temperature are interconnected. The target operating power and target outlet water temperature are set to zero when the target start / stop status is in the shutdown state.

[0046] In this embodiment, the operation plan includes the target start / stop status, target operating power, and target outlet water temperature of the second heating unit at various times within a future period. The start status indicates that the second heating unit needs to operate at that time, while the stop status indicates that the second heating unit does not need to operate at that time. When the target start / stop status is start, the corresponding target operating power and target outlet water temperature are calculated based on the target heat required to be supplemented at the current time, combined with the current pipeline heat exchange characteristics. These three factors are combined to form a complete control command, ensuring that the second heating unit can accurately output heat according to the optimization results, avoiding problems of insufficient or excessive heating.

[0047] For example, if the target heat at the current moment is 0, the target start / stop state is directly set to shutdown, and the target operating power and target outlet water temperature are automatically set to zero; if the target heat at the current moment is 12kW, the corresponding predicted energy efficiency coefficient is 2.4, and the pipeline efficiency is 0.9, then the net power output of the heat pump at this moment can be calculated to be 12kW / (2.4×0.9)≈5.56kW. Combining the power-temperature correspondence of the second heating unit, the target operating power can be set to 55.6% of the rated power, and the corresponding target outlet water temperature can be set to 42℃. Finally, the three parameters are organized into the control command for the current moment and sent to the controller of the second heating unit for execution.

[0048] S50: Based on real-time temperature feedback and building thermal inertia, an adaptive control algorithm is used to correct the operation plan in real time and generate execution instructions to control the second heating unit to provide heating.

[0049] In this embodiment of the application, due to the inertia of the building thermal process, there may be errors in the outdoor microclimate prediction, and the actual stored heat may also deviate from the predicted value. Therefore, it is necessary to make real-time corrections based on the operation plan obtained by rolling optimization, combined with the real-time collected indoor temperature and building thermal inertia state, to avoid the accumulation of errors that cause the indoor temperature to deviate from the set range.

[0050] First, the actual indoor temperature is collected in real time and compared with the set comfortable temperature range to calculate the current temperature deviation. Then, the required additional or reduced heat is estimated based on the current building heat capacity parameters. The corrected heat is added to the original target heat in the operation plan, and the corrected operating power and outlet water temperature are recalculated. Finally, the actual execution command is generated and sent to the second heating unit to ensure that the indoor temperature remains stable within the preset range. New feedback data is collected after each control cycle is completed, and the above correction process is repeated to achieve closed-loop adaptive control.

[0051] Specifically, step S50 in the method includes: The air temperature value inside the cold-region prefabricated building is collected in real time, and the air temperature value is compared with the target temperature value in the preset heating conditions to obtain the real-time temperature error value; Calculate the rate of change of the real-time temperature error value relative to time to obtain the temperature error rate of change value; The characteristic quantity of the building's thermal inertia state is obtained. The real-time temperature error value, the temperature error rate of change value, and the characteristic quantity of the building's thermal inertia state are input into a fuzzy PID controller. The fuzzy PID controller performs fuzzy reasoning on the input quantity according to a predefined fuzzy rule base, outputs a correction coefficient, and adds the correction coefficient to the target operating power in the operation plan to generate the execution instruction. The method for obtaining the characterization of the building's thermal inertia state includes: Multiple temperature sensors are pre-embedded in the inner side of the outer wall and the inner partition wall of the cold-region prefabricated building. The multiple temperature sensors are deployed at different depths inside the wall. The temperature measurement values ​​of each temperature sensor distributed along the depth direction of the wall are read in real time to construct the internal temperature distribution curve of the wall at the set point; The temperature gradient values ​​at different depths inside the wall are calculated based on the internal temperature distribution curve of the wall. The temperature gradient values ​​represent the direction and rate of heat transfer from the inside to the outside or from the outside to the inside of the wall. The temperature gradient value at the depth closest to the indoor surface is used as a characterization of the building's thermal inertia state. A positive gradient indicates that the wall is releasing stored heat into the room, while a negative gradient indicates that the wall is absorbing and storing heat from the room.

[0052] In this embodiment of the application, firstly, the current indoor actual air temperature value is collected once every control cycle, and the difference between the collected actual temperature value and the target comfort temperature value set by the user is calculated to obtain the real-time temperature error value at the current moment.

[0053] Secondly, calculate the temperature error difference between two adjacent control cycles, divide it by the control cycle length, and obtain the temperature error change rate at the current moment.

[0054] Simultaneously, temperature data collected by temperature sensors embedded at different depths within the wall are read in real time. Based on the temperature measurements at different depths, a temperature distribution curve along the wall depth direction at the current moment is obtained. The derivative of this curve is used to obtain the temperature gradient value at the corresponding depth. The temperature gradient value at the depth closest to the indoor surface is taken as a characterization of the current state of the building's thermal inertia. The characterization of the building's thermal inertia reflects the release trend of heat stored inside the prefabricated building envelope in cold regions.

[0055] Furthermore, the real-time temperature error value, temperature error rate of change value, and temperature gradient characterization value are input into the pre-built fuzzy PID controller. The controller performs fuzzification processing on the three input values ​​according to the pre-stored fuzzy rules, performs inference by matching the corresponding fuzzy rules, and then defuzzifies to obtain the correction coefficients of the PID parameters. After adjusting the proportional, integral, and derivative parameters of the PID with the correction coefficients, the correction amount required for the original target operating power at the current moment is calculated. The correction amount is added to the target operating power of the original operating plan. At the same time, the corresponding target outlet water temperature is recalculated based on the corrected operating power. Finally, combined with the original target start-stop state, the execution command actually sent to the second heating unit controller at the current moment is generated to complete this closed-loop correction.

[0056] Specifically, acquiring the characteristic quantity of the building's thermal inertia state involves first pre-embedding temperature sensors at different depths in the wall and reading the temperature measurements from each sensor distributed along the wall's depth direction in real time. Based on these measurements, a temperature distribution curve inside the wall at a set point is constructed. Then, the temperature gradient value at different depths inside the wall is calculated based on this curve. This temperature gradient value reflects the direction and intensity of heat storage and release within the wall. Finally, the temperature gradient value at the depth closest to the indoor surface is used as the characteristic quantity of the building's thermal inertia state. Before the indoor temperature deviates significantly, the wall's temperature gradient has already changed. By inputting the dynamic trend of the building's thermal inertia into the controller in advance, the controller can detect changes in the wall's heat storage and release, making timely adjustments to the operating power, avoiding control lag issues caused by building thermal inertia, and further improving the stability of indoor temperature control.

[0057] Specifically, the fuzzy PID controller also has a rule-adaptive adjustment function, the execution steps of which include: The control performance indicators of the fuzzy PID controller are continuously monitored. The control performance indicators include temperature overshoot and settling time. The temperature overshoot is the maximum deviation of the real-time temperature from the target temperature value. The settling time is the time taken from the change of the executed command to the indoor temperature entering the preset tolerance range. The temperature overshoot is compared with a preset overshoot threshold. When the temperature overshoot exceeds the overshoot threshold multiple times in a row, it is determined to be an inertial prediction underestimation state. In the state of insufficient inertial prediction, the weighting coefficient of the temperature error change rate input in the fuzzy rule base is increased by a first predetermined ratio, so that the fuzzy PID controller can respond to the temperature change trend in advance. The adjustment stabilization time is compared with a preset stabilization time threshold. When the adjustment stabilization time exceeds the stabilization time threshold multiple times consecutively, it is determined to be a slow response state. In the state of slow response, the weighting coefficient of the real-time temperature error input in the fuzzy rule base is increased by a second predetermined ratio, so that the fuzzy PID controller enhances its response to the current deviation.

[0058] In this embodiment of the application, firstly, the temperature overshoot and adjustment stabilization time are continuously recorded during each temperature deviation from the target value. The results of multiple consecutive monitoring are used to determine whether the current fuzzy rule is suitable for the thermal inertia characteristics of the current building.

[0059] Specifically, when the temperature overshoot exceeds the preset overshoot threshold multiple times consecutively, such as exceeding the preset overshoot threshold three times in a row, it indicates that the controller is not making enough predictions about the dynamic trend of the building's thermal inertia and is failing to respond in advance. In this case, the weight coefficient of the fuzzy domain corresponding to the temperature error rate of change input is increased by a first predetermined ratio to strengthen the influence of the temperature change trend in the inference process, so that the controller can adjust the output according to the temperature change trend earlier and suppress the generation of overshoot.

[0060] The first predetermined ratio refers to a preset multiple or percentage used to increase the weight coefficient of the temperature error change rate input in the fuzzy rule base when the inertial estimation is insufficient. The specific value can be determined through on-site debugging or simulation optimization, and the value range is usually greater than 1.0, for example, between 1.2 and 2.0. Further, the specific adjustment method is to multiply the original temperature error change rate weight coefficient by the first predetermined ratio. For example, if the original weight is 0.3 and the first predetermined ratio is 1.5, then the new weight is 0.45. By increasing the weight of the change rate, the controller pays more attention to the rate of temperature change rather than the current absolute deviation, thereby reducing the output in advance when the temperature is about to exceed the set value, suppressing overshoot.

[0061] Secondly, if the adjustment stabilization time exceeds the preset stabilization time threshold multiple times consecutively, such as exceeding the preset stabilization time threshold three times in a row, it indicates that the controller's response to the existing temperature deviation is insufficient. At this time, the weight coefficient of the real-time temperature error input in the fuzzy rule base is increased by a second predetermined ratio to enhance the controller's response to the current actual temperature deviation and accelerate the return of the temperature to the set range.

[0062] The second predetermined ratio refers to a preset multiple or percentage used to increase the weight coefficient of the real-time temperature error input in the fuzzy rule base when the response is sluggish. The specific value is set according to the response characteristics, and the range is usually greater than 1.0, for example, between 1.1 and 1.6, to avoid over-adjustment. Further, the adjustment method is to multiply the original real-time temperature error weight coefficient by the second predetermined ratio. For example, if the original weight is 0.5 and the second predetermined ratio is 1.3, then the new weight is 0.65. By increasing the weight of the current error, the controller becomes more sensitive to temperature deviations, enhancing its ability to eliminate steady-state error and accelerate response, thus shortening the settling time.

[0063] By adaptively adjusting the weights of fuzzy rules online, the controller can adapt to the differences in thermal characteristics of different buildings and building envelopes, and can automatically optimize control performance during long-term operation without the need for manual on-site adjustment of rule parameters.

[0064] Furthermore, a heating adaptive control method based on microclimate prediction also forms a closed-loop feedback update mechanism during execution, which includes: After executing the execution command and controlling the second heating unit to provide heating, the actual indoor temperature change trajectory of this heating response is collected, and the actual indoor temperature change trajectory is compared with the expected temperature change trajectory predicted in the operation plan to calculate the prediction deviation value. Based on the predicted deviation value, the equivalent thermal parameter model of the first heating unit and the predicted energy efficiency coefficient model of the second heating unit are corrected online. The corrected model parameters are then used for microclimate prediction and heating adaptive control in the next future period to achieve rolling updates of model parameters. The calibration steps of the equivalent thermal parameter model include: using the prediction deviation value as a feedback signal, adjusting the heat capacity and thermal resistance parameters of the heat storage body in the equivalent thermal parameter model using the gradient descent method; calculating the residual between the adjusted model output and the actual temperature; if the residual exceeds a preset tolerance range, iteratively correcting the heat capacity and thermal resistance parameters along the gradient direction until the residual converges to within the tolerance range; and using the converged heat capacity and thermal resistance parameters as the model parameters for the next round of prediction. The calibration steps of the predicted energy efficiency coefficient model include: mapping the predicted deviation value to an error term for energy efficiency coefficient prediction; collecting the error term sequence within the most recent preset time period using a sliding window method; performing statistical analysis on the error term sequence to calculate the mean drift and variance change of the error term; adjusting the baseline value of the real-time correction coefficient based on the mean drift and adjusting the forgetting factor of the recursive least squares algorithm based on the variance change; and using the adjusted real-time correction coefficient and forgetting factor for the next round of online fitting of the energy efficiency coefficient.

[0065] In this embodiment, firstly, after each control cycle, the actual indoor temperature change trajectory is compared with the predicted expected temperature change trajectory, and the deviation value generated in the prediction process is calculated. This deviation essentially originates from the parameter mismatch of the building equivalent thermal parameter model or the heat pump energy efficiency coefficient model. Therefore, the deviation needs to be used as a feedback signal to correct the parameters of the two models in reverse, so as to ensure that the accuracy of subsequent predictions remains stable.

[0066] For the equivalent thermal parameter model, the prediction bias is used as the input to the loss function. Gradient descent is employed to progressively adjust the heat capacity and thermal resistance parameters in the model. After each adjustment, the residual between the model output and the actual temperature is calculated until the residual falls within a preset tolerance range. The iteration stops, and the updated parameters are used for the next round of prediction. This allows the equivalent thermal parameter model to gradually match the actual changes in the building's thermal characteristics, such as the slow changes in wall insulation performance over time. For example, after five years of use, the wall insulation material may experience slight performance degradation due to moisture, resulting in a lower equivalent thermal resistance than the initial design value. Closed-loop correction can gradually adjust the thermal resistance parameters in the model to values ​​that reflect actual conditions, preventing a continuous decline in prediction accuracy due to model parameter aging.

[0067] For the predicted energy efficiency coefficient (EEC) model of the second heating unit, the prediction deviation is first converted into the prediction error term of the EEC. A sliding window is used to retain the error sequence of the most recent N control cycles. Statistical analysis is performed on the sequence to obtain the mean drift and variance change of the error. Then, the baseline of the real-time correction coefficient for energy efficiency prediction is adjusted according to the mean drift to eliminate the overall deviation of the prediction results. Next, the forgetting factor of the recursive least squares algorithm is adjusted according to the variance change. If the error fluctuation increases, the forgetting factor is reduced, giving more weight to recent data to keep up with changes in the EEC more quickly. Ultimately, the accuracy of the energy efficiency prediction is kept within a reasonable range, thus completing the closed-loop update of the entire control process. For example, after three years of operation of the heat pump unit, dust accumulation on the evaporator surface causes a slow decline in heat exchange efficiency, and the actual EEC will be lower than the initial model prediction. Closed-loop correction can automatically capture the mean drift, adjust the correction coefficient baseline to eliminate the overall deviation, and ensure the accuracy of subsequent prediction results.

[0068] In summary, compared with existing technologies, this application optimizes the operation plan of the heating unit in advance by combining outdoor microclimate forecast data of cold regions. At the same time, it introduces the internal temperature gradient of the building walls to characterize the thermal inertia state of the building. In the real-time control stage, it combines the thermal inertia trend to correct the operating parameters in advance. It can also adaptively adjust the fuzzy control rules online and continuously update the prediction model parameters. Under the premise of meeting the user's thermal comfort needs, it can maximize the operating efficiency of the heating system, reduce unnecessary heat waste, adapt to the thermal characteristics of prefabricated buildings in cold regions, and achieve adaptive and precise heating.

[0069] In summary, the embodiments of this application have at least the following technical effects: This application provides a heating adaptive control method based on microclimate prediction. First, it acquires microclimate prediction data for future periods to predict the impact of the outdoor environment on the indoor thermal environment. Combined with the predicted stored heat sequence of the first heating unit, it calculates the heat that free heat sources such as solar energy can provide. Based on this, it plans the operation of the second heating unit to maximize the utilization of free heat sources and reduce unnecessary energy consumption. Second, after generating the initial operation plan, this application uses real-time temperature feedback and building thermal inertia to correct the plan in real time, offsetting the deviation between the prediction results and actual needs, improving the control accuracy of indoor temperature, and avoiding excessive temperature fluctuations. Finally, a closed-loop feedback update mechanism is set up to continuously correct model parameters based on actual heating response results, adapting to the slow changes in the thermal properties of building components over time. This ensures stable control performance even after long-term operation, solving the problem of performance degradation after long-term operation in traditional methods.

[0070] Through the above technical solution, this application can reduce the energy consumption of the second heating unit by making full use of natural heat sources such as solar energy while ensuring the indoor thermal comfort of prefabricated buildings in cold regions. At the same time, by combining adaptive correction of the building's real-time thermal inertia and closed-loop rolling model parameter updates, it can maintain high control accuracy and energy-saving effect in the long term. Compared with traditional control methods, it can effectively reduce heating energy consumption, improve the stability of the indoor thermal environment, and meet the heating control needs of prefabricated buildings in severely cold regions.

[0071] Example 2, as Figure 2 As shown, based on the same inventive concept as the microclimate prediction-based adaptive heating control method provided in Embodiment 1, this application also provides a microclimate prediction-based adaptive heating control system, including: The prediction data acquisition module 11 is used to acquire microclimate prediction data for a future period, wherein the microclimate prediction data includes a predicted solar radiation intensity sequence, a predicted outdoor temperature sequence, and a predicted relative humidity sequence. The stored heat sequence prediction module 12 is used to predict and obtain the predicted stored heat sequence of the first heating unit in the future time period based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence. The energy efficiency coefficient sequence prediction module 13 is used to predict and obtain the predicted energy efficiency coefficient sequence of the second heating unit in the future time period based on the predicted outdoor temperature sequence and the predicted relative humidity sequence. The operation plan generation module 14 is used to calculate the target heat that needs to be supplemented by the second heating unit based on the predicted stored heat sequence, constrained by the preset heating conditions of the cold-region prefabricated building, and convert the target heat into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. The operation plan correction module 15 is used to correct the operation plan in real time based on real-time temperature feedback and building thermal inertia state using an adaptive control algorithm, and generate execution instructions to control the second heating unit to provide heating.

[0072] The first heating unit is the solar energy storage and heat collection unit of the prefabricated building in cold regions, and the second heating unit is the heat pump heating unit of the prefabricated building in cold regions. The heating equipment of the heat pump heating unit is an ultra-low temperature air source heat pump.

[0073] In one embodiment, the heat sequence prediction module 12 is specifically used for: Establish an equivalent thermal parameter model for the first heating unit; The predicted solar radiation intensity sequence is used as the solar radiation heat flow input of the equivalent thermal parameter model, the predicted outdoor temperature sequence is used as the boundary temperature input of the equivalent thermal parameter model, and the heat storage body heat capacity parameter and thermal resistance parameter in the equivalent thermal parameter model are set as fixed parameters determined in advance through system identification experiments. Run the equivalent thermal parameter model to perform time-by-time simulation calculations and output the average temperature of the heat storage body of the first heating unit at each time in the future time period; The average temperature of the heat storage body is subtracted from the preset reference temperature value to obtain the heat storage body temperature difference sequence. Each temperature difference value in the heat storage body temperature difference sequence is multiplied by the heat capacity parameter of the heat storage body to obtain the instantaneous stored heat value at each time. The reference temperature value is the equilibrium temperature of the first heating unit in the state of completely releasing heat. Arrange the instantaneous stored heat values ​​at each moment in chronological order to generate the predicted stored heat sequence of the first heating unit in the future time period; The equivalent thermal parameter model adopts a three-thermal-resistance, two-heat-capacity model structure, and its construction method includes: The first heating unit is divided into two heat capacity nodes: a surface air layer and an internal heat storage wall layer. The surface air layer corresponds to the first heat capacity parameter, and the internal heat storage wall layer corresponds to the second heat capacity parameter. A first thermal resistance parameter is set between the outdoor environment and the surface air layer to characterize the first convective heat transfer resistance between the outdoor air and the surface air layer. A second thermal resistance parameter is provided between the surface air layer and the internal heat storage wall layer to characterize the thermal resistance between the surface air layer and the internal heat storage wall layer; A third thermal resistance parameter is set between the internal heat storage wall layer and the indoor environment to characterize the second convective heat transfer resistance between the internal heat storage wall layer and the indoor air. The predicted solar radiation intensity sequence is converted into a heat flow source term and applied to the surface air layer node, and the predicted outdoor temperature sequence is applied as a boundary condition to the outdoor environment node.

[0074] In one embodiment, energy efficiency coefficient sequence prediction 13 is specifically used for: Obtain the standard energy efficiency performance curve calibrated at the factory for the second heating unit; Collect actual operating data of the second heating unit during historical operating periods, including actual outdoor temperature, actual relative humidity, actual power consumption, and actual heat output. The actual energy efficiency coefficients at each historical moment are calculated based on the actual operating data, and the actual energy efficiency coefficients are compared with the benchmark values ​​of the standard energy efficiency performance curve under the same operating conditions to obtain the deviation sequence. The recursive least squares algorithm is used to fit the deviation sequence online and generate real-time correction coefficients. The online fitting process includes: initializing the parameter vector and covariance matrix; whenever a new deviation value is collected, calculating the gain vector based on the current covariance matrix and the input feature vector; iteratively updating the parameter vector and covariance matrix using the prediction error; and using the model coefficients in the updated parameter vector as real-time correction coefficients. The real-time correction coefficient is superimposed on the standard energy efficiency performance curve to obtain the corrected predicted energy efficiency coefficient model. The predicted outdoor temperature sequence and the predicted relative humidity sequence are used as inputs to output the predicted energy efficiency coefficient sequence of the second heating unit in the future time period.

[0075] In one embodiment of the application, the runtime plan generation module 14 is specifically used for: The preset heating conditions include temperature constraints, start / stop constraints, and cost constraints. A rolling optimization framework is established using a model predictive control algorithm. The prediction time domain of the rolling optimization framework covers the future time period, and the control time domain is a subset of the prediction time domain. Within the rolling optimization framework, the preset heating conditions are used as hard constraints, and the goal is to minimize the operating cost; an optimization problem is then constructed. Using the predicted stored heat sequence as a known free heat source input, the difference between the indoor heat demand at each moment and the corresponding value in the predicted stored heat sequence is calculated to obtain the time-series distribution of the heat gap; The predicted energy efficiency coefficient sequence is substituted into the objective function as the time-varying efficiency parameter of the second heating unit, so that the optimization problem can identify the energy efficiency differences at different times. Solving the optimization problem yields the optimal start-stop state sequence and optimal operating power sequence of the second heating unit in the control time domain, which serve as the operating plan.

[0076] The operation plan includes the target start / stop status, target operating power, and target outlet water temperature of the second heating unit at various times in the future period. The target start / stop status is a binary state variable, including both start-up and shutdown states. The target operating power is a continuous variable, taking values ​​between zero power and rated power, used to determine the specific load rate of the second heating unit in the start-up state. The target outlet water temperature is also a continuous variable, taking values ​​between the minimum safe outlet water temperature and the maximum allowable outlet water temperature, used to control the heating quality of the second heating unit. The target start / stop status, target operating power, and target outlet water temperature are interconnected. The target operating power and target outlet water temperature are set to zero when the target start / stop status is in the shutdown state.

[0077] In one embodiment of the application, the operation plan correction module 15 is specifically used for: The air temperature value inside the cold-region prefabricated building is collected in real time, and the air temperature value is compared with the target temperature value in the preset heating conditions to obtain the real-time temperature error value; Calculate the rate of change of the real-time temperature error value relative to time to obtain the temperature error rate of change value; The characteristic quantity of the building's thermal inertia state is obtained. The real-time temperature error value, the temperature error rate of change value, and the characteristic quantity of the building's thermal inertia state are input into a fuzzy PID controller. The fuzzy PID controller performs fuzzy reasoning on the input quantity according to a predefined fuzzy rule base, outputs a correction coefficient, and adds the correction coefficient to the target operating power in the operation plan to generate the execution instruction. The method for obtaining the characterization of the building's thermal inertia state includes: Multiple temperature sensors are pre-embedded in the inner side of the outer wall and the inner partition wall of the cold-region prefabricated building. The multiple temperature sensors are deployed at different depths inside the wall. The temperature measurement values ​​of each temperature sensor distributed along the depth direction of the wall are read in real time to construct the internal temperature distribution curve of the wall at the set point; The temperature gradient values ​​at different depths inside the wall are calculated based on the internal temperature distribution curve of the wall. The temperature gradient values ​​represent the direction and rate of heat transfer from the inside to the outside or from the outside to the inside of the wall. The temperature gradient value at the depth closest to the indoor surface is used as a characterization of the building's thermal inertia state. A positive gradient indicates that the wall is releasing stored heat into the room, while a negative gradient indicates that the wall is absorbing and storing heat from the room.

[0078] Furthermore, the fuzzy PID controller also has a rule-adaptive adjustment function, the execution steps of which include: The control performance indicators of the fuzzy PID controller are continuously monitored. The control performance indicators include temperature overshoot and settling time. The temperature overshoot is the maximum deviation of the real-time temperature from the target temperature value. The settling time is the time taken from the change of the executed command to the indoor temperature entering the preset tolerance range. The temperature overshoot is compared with a preset overshoot threshold. When the temperature overshoot exceeds the overshoot threshold multiple times in a row, it is determined to be an inertial prediction underestimation state. In the state of insufficient inertial prediction, the weighting coefficient of the temperature error change rate input in the fuzzy rule base is increased by a first predetermined ratio, so that the fuzzy PID controller can respond to the temperature change trend in advance. The adjustment stabilization time is compared with a preset stabilization time threshold. When the adjustment stabilization time exceeds the stabilization time threshold multiple times consecutively, it is determined to be a slow response state. In the state of slow response, the weighting coefficient of the real-time temperature error input in the fuzzy rule base is increased by a second predetermined ratio, so that the fuzzy PID controller enhances its response to the current deviation.

[0079] Furthermore, a heating adaptive control method based on microclimate prediction also forms a closed-loop feedback update mechanism during execution, which includes: After executing the execution command and controlling the second heating unit to provide heating, the actual indoor temperature change trajectory of this heating response is collected, and the actual indoor temperature change trajectory is compared with the expected temperature change trajectory predicted in the operation plan to calculate the prediction deviation value. Based on the predicted deviation value, the equivalent thermal parameter model of the first heating unit and the predicted energy efficiency coefficient model of the second heating unit are corrected online. The corrected model parameters are then used for microclimate prediction and heating adaptive control in the next future period to achieve rolling updates of model parameters. The calibration steps of the equivalent thermal parameter model include: using the prediction deviation value as a feedback signal, adjusting the heat capacity and thermal resistance parameters of the heat storage body in the equivalent thermal parameter model using the gradient descent method; calculating the residual between the adjusted model output and the actual temperature; if the residual exceeds a preset tolerance range, iteratively correcting the heat capacity and thermal resistance parameters along the gradient direction until the residual converges to within the tolerance range; and using the converged heat capacity and thermal resistance parameters as the model parameters for the next round of prediction. The calibration steps of the predicted energy efficiency coefficient model include: mapping the predicted deviation value to an error term for energy efficiency coefficient prediction; collecting the error term sequence within the most recent preset time period using a sliding window method; performing statistical analysis on the error term sequence to calculate the mean drift and variance change of the error term; adjusting the baseline value of the real-time correction coefficient based on the mean drift and adjusting the forgetting factor of the recursive least squares algorithm based on the variance change; and using the adjusted real-time correction coefficient and forgetting factor for the next round of online fitting of the energy efficiency coefficient.

Claims

1. A heating adaptive control method based on microclimate prediction, characterized in that, Applied to prefabricated buildings in cold regions, wherein the prefabricated buildings in cold regions include a first heating unit and a second heating unit, the method includes: Obtain microclimate prediction data for future time periods, wherein the microclimate prediction data includes predicted solar radiation intensity sequences, predicted outdoor temperature sequences, and predicted relative humidity sequences; Based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence, the predicted stored heat sequence of the first heating unit in the future time period is obtained. The first heating unit is a solar energy storage and heat collection unit of a prefabricated building in cold regions. Based on the predicted outdoor temperature sequence and the predicted relative humidity sequence, the predicted energy efficiency coefficient sequence of the second heating unit in the future time period is obtained. The second heating unit is a heat pump heating unit for prefabricated buildings in cold regions, and the heating equipment of the heat pump heating unit is an air source heat pump. Constrained by the preset heating conditions of prefabricated buildings in cold regions, the target heat that needs to be supplemented by the second heating unit is calculated based on the predicted stored heat sequence, and the target heat is converted into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. Based on real-time temperature feedback and building thermal inertia, an adaptive control algorithm is used to correct the operation plan in real time and generate execution instructions to control the second heating unit to provide heating. Based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence, the predicted stored heat sequence of the first heating unit in the future time period is predicted and obtained, including: Establish an equivalent thermal parameter model for the first heating unit; The predicted solar radiation intensity sequence is used as the solar radiation heat flow input of the equivalent thermal parameter model, the predicted outdoor temperature sequence is used as the boundary temperature input of the equivalent thermal parameter model, and the heat storage body heat capacity parameter and thermal resistance parameter in the equivalent thermal parameter model are set as fixed parameters determined in advance through system identification experiments. Run the equivalent thermal parameter model to perform time-by-time simulation calculations and output the average temperature of the heat storage body of the first heating unit at each time in the future time period; The average temperature of the heat storage body is subtracted from the preset reference temperature value to obtain the heat storage body temperature difference sequence. Each temperature difference value in the heat storage body temperature difference sequence is multiplied by the heat capacity parameter of the heat storage body to obtain the instantaneous stored heat value at each time. The reference temperature value is the equilibrium temperature of the first heating unit in the state of completely releasing heat. Arrange the instantaneous stored heat values ​​at each moment in chronological order to generate the predicted stored heat sequence of the first heating unit in the future time period; The equivalent thermal parameter model adopts a three-thermal-resistance, two-heat-capacity model structure, and its construction method includes: The first heating unit is divided into two heat capacity nodes: a surface air layer and an internal heat storage wall layer. The surface air layer corresponds to the first heat capacity parameter, and the internal heat storage wall layer corresponds to the second heat capacity parameter. A first thermal resistance parameter is set between the outdoor environment and the surface air layer to characterize the first convective heat transfer resistance between the outdoor air and the surface air layer. A second thermal resistance parameter is provided between the surface air layer and the internal heat storage wall layer to characterize the thermal resistance between the surface air layer and the internal heat storage wall layer; A third thermal resistance parameter is set between the internal heat storage wall layer and the indoor environment to characterize the second convective heat transfer resistance between the internal heat storage wall layer and the indoor air. The predicted solar radiation intensity sequence is converted into a heat flow source term and applied to the surface air layer node, and the predicted outdoor temperature sequence is applied as a boundary condition to the outdoor environment node. Based on real-time temperature feedback and building thermal inertia, an adaptive control algorithm is used to correct the operation plan in real time, generating execution commands to control the second heating unit to provide heating, including: The air temperature value inside the cold-region prefabricated building is collected in real time, and the air temperature value is compared with the target temperature value in the preset heating conditions to obtain the real-time temperature error value; Calculate the rate of change of the real-time temperature error value relative to time to obtain the temperature error rate of change value; The characteristic quantity of the building's thermal inertia state is obtained. The real-time temperature error value, the temperature error rate of change value, and the characteristic quantity of the building's thermal inertia state are input into a fuzzy PID controller. The fuzzy PID controller performs fuzzy reasoning on the input quantity according to a predefined fuzzy rule base, outputs a correction coefficient, and adds the correction coefficient to the target operating power in the operation plan to generate the execution instruction. The method for obtaining the characterization of the building's thermal inertia state includes: Multiple temperature sensors are pre-embedded in the inner side of the outer wall and the inner partition wall of the cold-region prefabricated building. The multiple temperature sensors are deployed at different depths inside the wall. The temperature measurement values ​​of each temperature sensor distributed along the depth direction of the wall are read in real time to construct the internal temperature distribution curve of the wall at the set point; The temperature gradient values ​​at different depths inside the wall are calculated based on the internal temperature distribution curve of the wall. The temperature gradient values ​​represent the direction and rate of heat transfer from the inside to the outside or from the outside to the inside of the wall. The temperature gradient value at the depth closest to the indoor surface is used as a characterization of the building's thermal inertia state. A positive gradient indicates that the wall is releasing stored heat into the room, while a negative gradient indicates that the wall is absorbing and storing heat from the room.

2. The heating adaptive control method based on microclimate prediction according to claim 1, characterized in that, The first heating unit is the solar energy storage and heat collection unit of the prefabricated building in cold regions, and the second heating unit is the heat pump heating unit of the prefabricated building in cold regions, wherein the heating equipment of the heat pump heating unit is an ultra-low temperature air source heat pump.

3. The heating adaptive control method based on microclimate prediction according to claim 1, characterized in that, Based on the predicted outdoor temperature sequence and the predicted relative humidity sequence, the predicted energy efficiency coefficient sequence of the second heating unit for the future time period is obtained, including: Obtain the standard energy efficiency performance curve calibrated at the factory for the second heating unit; Collect actual operating data of the second heating unit during historical operating periods, including actual outdoor temperature, actual relative humidity, actual power consumption, and actual heat output. The actual energy efficiency coefficients at each historical moment are calculated based on the actual operating data, and the actual energy efficiency coefficients are compared with the benchmark values ​​of the standard energy efficiency performance curve under the same operating conditions to obtain the deviation sequence. The recursive least squares algorithm is used to fit the deviation sequence online and generate real-time correction coefficients. The online fitting process includes: initializing the parameter vector and covariance matrix; whenever a new deviation value is collected, calculating the gain vector based on the current covariance matrix and the input feature vector; iteratively updating the parameter vector and covariance matrix using the prediction error; and using the model coefficients in the updated parameter vector as real-time correction coefficients. The real-time correction coefficient is superimposed on the standard energy efficiency performance curve to obtain the corrected predicted energy efficiency coefficient model. The predicted outdoor temperature sequence and the predicted relative humidity sequence are used as inputs to output the predicted energy efficiency coefficient sequence of the second heating unit in the future time period.

4. The heating adaptive control method based on microclimate prediction according to claim 1, characterized in that, Constrained by the preset heating conditions of prefabricated buildings in cold regions, the target heat that needs to be supplemented by the second heating unit is calculated based on the predicted stored heat sequence, and the target heat is converted into an operating plan for the second heating unit in the future period based on the predicted energy efficiency coefficient sequence, including: The preset heating conditions include temperature constraints, start / stop constraints, and cost constraints. A rolling optimization framework is established using a model predictive control algorithm. The prediction time domain of the rolling optimization framework covers the future time period, and the control time domain is a subset of the prediction time domain. Within the rolling optimization framework, the preset heating conditions are used as hard constraints, and the goal is to minimize the operating cost; an optimization problem is then constructed. Using the predicted stored heat sequence as a known free heat source input, the difference between the indoor heat demand at each moment and the corresponding value in the predicted stored heat sequence is calculated to obtain the time-series distribution of the heat gap; The predicted energy efficiency coefficient sequence is substituted into the objective function as the time-varying efficiency parameter of the second heating unit, so that the optimization problem can identify the energy efficiency differences at different times. Solving the optimization problem yields the optimal start-stop state sequence and optimal operating power sequence of the second heating unit in the control time domain, which serve as the operating plan.

5. The heating adaptive control method based on microclimate prediction according to claim 4, characterized in that, The operation plan includes the target start / stop status, target operating power, and target outlet water temperature of the second heating unit at various times in the future period. The target start / stop status is a binary state variable, including both start-up and shutdown states. The target operating power is a continuous variable, taking values ​​between zero power and rated power, used to determine the specific load rate of the second heating unit in the start-up state. The target outlet water temperature is also a continuous variable, taking values ​​between the minimum safe outlet water temperature and the maximum allowable outlet water temperature, used to control the heating quality of the second heating unit. The target start / stop status, target operating power, and target outlet water temperature are interrelated. The target operating power and target outlet water temperature are set to zero when the target start / stop status is in the shutdown state.

6. The heating adaptive control method based on microclimate prediction according to claim 1, characterized in that, The fuzzy PID controller also has a rule-adaptive adjustment function, the execution steps of which include: The control performance indicators of the fuzzy PID controller are continuously monitored. The control performance indicators include temperature overshoot and settling time. The temperature overshoot is the maximum deviation of the real-time temperature from the target temperature value. The settling time is the time taken from the change of the executed command to the indoor temperature entering the preset tolerance range. The temperature overshoot is compared with a preset overshoot threshold. When the temperature overshoot exceeds the overshoot threshold multiple times in a row, it is determined to be an inertial prediction underestimation state. In the state of insufficient inertial prediction, the weighting coefficient of the temperature error change rate input in the fuzzy rule base is increased by a first predetermined ratio, so that the fuzzy PID controller can respond to the temperature change trend in advance. The adjustment stabilization time is compared with a preset stabilization time threshold. When the adjustment stabilization time exceeds the stabilization time threshold multiple times consecutively, it is determined to be a slow response state. In the state of slow response, the weighting coefficient of the real-time temperature error input in the fuzzy rule base is increased by a second predetermined ratio, so that the fuzzy PID controller enhances its response to the current deviation.

7. The heating adaptive control method based on microclimate prediction according to claim 1, characterized in that, The method also forms a closed-loop feedback update mechanism during execution, which includes: After executing the execution command and controlling the second heating unit to provide heating, the actual indoor temperature change trajectory of this heating response is collected, and the actual indoor temperature change trajectory is compared with the expected temperature change trajectory predicted in the operation plan to calculate the prediction deviation value. Based on the predicted deviation value, the equivalent thermal parameter model of the first heating unit and the predicted energy efficiency coefficient model of the second heating unit are corrected online. The corrected model parameters are then used for microclimate prediction and heating adaptive control in the next future period to achieve rolling updates of model parameters. The calibration steps of the equivalent thermal parameter model include: using the prediction deviation value as a feedback signal, adjusting the heat capacity and thermal resistance parameters of the heat storage body in the equivalent thermal parameter model using the gradient descent method; calculating the residual between the adjusted model output and the actual temperature; if the residual exceeds a preset tolerance range, iteratively correcting the heat capacity and thermal resistance parameters along the gradient direction until the residual converges to within the tolerance range; and using the converged heat capacity and thermal resistance parameters as the model parameters for the next round of prediction. The calibration steps of the predicted energy efficiency coefficient model include: mapping the predicted deviation value to an error term for energy efficiency coefficient prediction; collecting the error term sequence within the most recent preset time period using a sliding window method; performing statistical analysis on the error term sequence to calculate the mean drift and variance change of the error term; adjusting the baseline value of the real-time correction coefficient based on the mean drift and adjusting the forgetting factor of the recursive least squares algorithm based on the variance change; and using the adjusted real-time correction coefficient and forgetting factor for the next round of online fitting of the energy efficiency coefficient.

8. A heating adaptive control system based on microclimate prediction, characterized in that, A heating adaptive control method based on microclimate prediction as described in any one of claims 1-7 includes: The prediction data acquisition module is used to acquire microclimate prediction data for future periods, wherein the microclimate prediction data includes a predicted solar radiation intensity sequence, a predicted outdoor temperature sequence, and a predicted relative humidity sequence. A stored heat sequence prediction module is used to predict and obtain the predicted stored heat sequence of the first heating unit in the future time period based on the predicted solar radiation intensity sequence and the predicted outdoor temperature sequence. An energy efficiency coefficient sequence prediction module is used to predict and obtain the predicted energy efficiency coefficient sequence of the second heating unit in the future time period based on the predicted outdoor temperature sequence and the predicted relative humidity sequence. The operation plan generation module is used to calculate the target heat that needs to be supplemented by the second heating unit based on the predicted stored heat sequence, constrained by the preset heating conditions of the cold-region prefabricated building, and to convert the target heat into the operation plan of the second heating unit in the future period based on the predicted energy efficiency coefficient sequence. The operation plan correction module is used to correct the operation plan in real time based on real-time temperature feedback and building thermal inertia state using an adaptive control algorithm, and generate execution instructions to control the second heating unit to provide heating.

Images