Optimal control method for cold and heat storage air-conditioning system of metering laboratory

Abstract

The invention discloses an optimal control method for metering a cold and heat storage air conditioning system in a laboratory, and relates to the technical field of optimal control, and the method comprises the following steps: predicting hourly cold load and heat load in a future preset period, and generating hourly load prediction data; obtaining time-of-use electricity price information, and generating electricity price time period division data; formulating an energy storage strategy based on hourly load prediction data and electricity price time division data; the method specifically comprises the steps that in the valley electricity price period, when it is predicted that a load demand exists in the future peak electricity price period, a cold and heat source module is controlled to operate, energy is stored in a cold and heat storage module till the rated capacity is reached or the valley period is ended, and the temperature and humidity in a laboratory are maintained within a first set range through a tail end processing module in the energy storage period; and formulating an energy release and conventional operation strategy based on hourly load prediction data, electricity price time division data and the real-time energy storage state of the cold storage and heat storage module. The method has the effect of improving the optimization control efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optimization control, in particular to an optimization control method of a cold and heat storage air conditioning system in a metrological laboratory. BACKGROUND

[0002] The control precision of the environment temperature and humidity in the metrological laboratory is extremely high, and it is usually required to control the temperature and humidity fluctuation in a very small range. A slight temperature and humidity fluctuation may affect the accuracy of the metrological instrument and the reliability of the metrological result. The traditional air conditioning system usually adopts a real-time load matching mode for operation. During the power peak period, the high energy consumption of the air conditioning system not only increases the operation cost, but also may cause an impact on the power grid.

[0003] The cold and heat storage technology is an effective "peak load shifting" means, which can significantly reduce the operation cost of the air conditioning system during the power peak period and reduce the pressure on the power grid. However, the existing cold and heat storage air conditioning control strategy is mainly aimed at commercial buildings or ordinary industrial buildings, and the control target focuses on energy saving and economy, and the control precision requirement is relatively loose. When these strategies are directly applied to the metrological laboratory, the following problems will be encountered: 1. Insufficient control precision: the traditional switching strategy based on the fixed time table or simple logic is easy to cause the step change of the supply air parameters during the energy storage, energy release and energy switching process, resulting in the over-difference fluctuation of the temperature and humidity in the laboratory; 2. Low energy utilization efficiency: lack of accurate load prediction may cause insufficient energy storage to meet the peak load, or excessive energy storage to cause energy waste, and the cold and heat source unit is frequently started and stopped or operated in the low efficiency area, which reduces the overall energy efficiency; 3. Unstable system operation: the dynamic disturbance in the mode switching process lacks effective inhibition means, which affects the continuous stability of the air conditioning system and even the internal environment of the laboratory. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an optimization control method of a cold and heat storage air conditioning system in a metrological laboratory.

[0005] In a first aspect, the present application provides an optimization control method of a cold and heat storage air conditioning system in a metrological laboratory, wherein the air conditioning system comprises a cold and heat source module, a cold and heat storage module, a terminal treatment module, a temperature and humidity monitoring module and a central control module, and the optimization control method comprises the following steps: Step S1: The central control module predicts the hourly cooling and heating load in the future preset period based on the historical operation data of the metrology laboratory, the outdoor weather forecast data in the future preset period, and the internal work plan of the laboratory, generates hourly load prediction data using a machine learning model; Step S2: The central control module obtains time-of-use electricity price information, divides a day into peak electricity price period, flat electricity price period, and valley electricity price period based on the time-of-use electricity price information, and generates electricity price period division data; Step S3: The central control module formulates an energy storage strategy based on the hourly load prediction data and the electricity price period division data; specifically, in the valley electricity price period, when there is a load demand in the predicted future peak electricity price period, the cold and heat source module is controlled to run and store energy in the cold and heat storage module until the rated capacity is reached or the valley period ends, and the end treatment module is used to maintain the temperature and humidity in the laboratory within a first set range during the energy storage period. Step S4: The central control module formulates an energy release and regular operation strategy based on the hourly load prediction data, the electricity price period division data, and the real-time energy storage state of the cold and heat storage module; specifically, it includes: Step S41: In the peak electricity price period, the cold and heat storage module is preferentially controlled to release energy to match the load demand, and the cold and heat source module is started to assist energy supply when the energy release is insufficient; Step S42: In the flat electricity price period, the proportion of direct energy supply of the cold and heat source module and energy release of the cold and heat storage module is dynamically adjusted, and whether to perform supplementary energy storage in the flat period is determined based on the prediction of the energy storage demand in the next peak period; Step S5: The temperature and humidity monitoring module collects temperature and humidity data sets corresponding to multiple monitoring points in the metrology laboratory in real time and feeds back to the central control module; the central control module compares the temperature and humidity data sets with the set values, and when the deviation exceeds a second set range, the supply air parameters of the end treatment module and the operating state of the cold and heat storage module are dynamically adjusted by an optimization control algorithm; Step S6: During the switching process of the cold and heat storage module and the cold and heat source module, the central control module uses a feedforward-feedback composite control strategy to adjust the valve opening and the water pump frequency, so that the change rate of the supply air parameters of the end treatment module during the switching process does not exceed a preset limit value; Step S7: The central control module calculates the prediction error, and modifies and optimizes the machine learning model and the energy storage and energy release strategies according to the error result.

[0006] Preferably, the step S1 includes the following steps: Step S11: Obtain the corresponding historical operation data of the metrology laboratory, the outdoor weather forecast data in the future preset period, and the internal work plan of the laboratory; Step S12: Feature extraction is performed on the historical operation data to obtain historical load feature data; Step S13: Based on the outdoor weather forecast data and the internal work plan of the laboratory, the future environment and internal disturbance characteristics are analyzed to generate future working condition characteristic data; Step S14: The historical load characteristic data and the future working condition characteristic data are input into the machine learning model, and then hourly load prediction data is output.

[0007] Preferably, the step S3 comprises the following steps: Step S31: Based on the hourly load prediction data, the load demand peak value and duration of the future peak electricity price period are identified; Step S32: Combined with the electricity price period division data and the rated capacity of the cold and heat storage module, the minimum energy storage capacity and the optimal energy storage rate required in the valley electricity price period are calculated; Step S33: Based on the load demand peak value and duration of the future peak electricity price period, an energy storage control instruction is generated to control the cold and heat source module to operate at the optimal energy storage rate, and the energy storage state of the cold and heat storage module and the temperature and humidity in the laboratory are monitored in real time to ensure that the temperature and humidity are maintained within the first set range.

[0008] Preferably, the step S41 comprises the following steps: Step S411: At the beginning of the peak electricity price period, the current available energy storage capacity of the cold and heat storage module and the predicted load in the corresponding period in the hourly load prediction data are obtained; Step S412: The current available energy storage capacity and the predicted load are compared, and if the available energy storage capacity is sufficient, a plan is made to release energy from the cold and heat storage module to meet the load; Step S413: If the available energy storage capacity is insufficient, a control instruction is generated to start the cold and heat source module to supplement the output, while adjusting the energy release rate of the cold and heat storage module.

[0009] Preferably, the step S5 comprises the following steps: Step S51: The temperature and humidity monitoring module collects temperature and humidity data of multiple monitoring points in the metering laboratory at a preset sampling frequency to form a temperature and humidity data set; Step S52: The central control module calculates the deviation of the average value of the temperature and humidity data set from the set value; Step S53: When the absolute value of the temperature deviation exceeds 0.3℃ or the absolute value of the humidity deviation exceeds 2%RH, it is determined that the deviation is out of tolerance, and a dynamic adjustment program is started. The adjustment amount of the air supply amount, air supply temperature and air supply humidity of the end treatment module, and the correction instruction of the output of the cold and heat source module or the energy release rate of the cold and heat storage module are calculated and output according to the deviation size and trend through the optimization control algorithm.

[0010] Preferably, the step S6 comprises the following steps: Step S61: Before switching the cold and heat storage modules, the central control module predicts the parameter variation trend in the switching process through feedforward control based on the hourly load prediction data; Step S62: According to the parameter variation trend, the pre-adjustment instructions for valve opening and water pump frequency are calculated and executed, and the feedback control is used to monitor the air supply parameter variation rate of the terminal processing module in real time during the switching process, so that the terminal air supply parameter variation rate during switching does not exceed the preset limit value.

[0011] Preferably, the step S7 comprises the following steps: S71: The central control module periodically aggregates the actual cold and heat load data, outdoor weather data, operating state data, and corresponding historical prediction data; S72: Calculate the hourly error of the actual cold and heat load data and the corresponding historical prediction data in the time period, and statistically analyze the error distribution characteristics, adjust the weight coefficients in the machine learning model according to the error distribution characteristics, and evaluate the execution effect of the energy storage strategy and the energy release strategy, and optimize the threshold parameters and priority coefficients in the strategy for triggering energy storage, energy release and mode switching.

[0012] Preferably, the first set range is temperature set value ±0.5℃, humidity set value ±3%RH.

[0013] In the second aspect, the application provides a computer readable storage medium storing instructions, when the instructions are run on a computer, the computer executes the optimization control method of the measurement laboratory cold and heat storage air conditioning system according to any one of the above.

[0014] In summary, the application has at least one of the following beneficial technical effects: 1. The application provides an optimization control method of a measurement laboratory cold and heat storage air conditioning system, which uses a machine learning model to accurately predict the future hourly load of the laboratory, lays a foundation for formulating an accurate energy allocation plan in advance, and combines real-time feedback of a temperature and humidity monitoring module and an optimization control algorithm to dynamically fine-tune the air supply parameters and system operating state, ensuring that the laboratory temperature and humidity are stable within the set range, and using a feedforward-feedback composite control strategy during system mode switching, effectively suppressing the air supply parameter fluctuations caused by valve and water pump actions, avoiding the common step changes in traditional strategies, and fundamentally solving the temperature and humidity out-of-tolerance problems that may occur during energy storage, energy release and switching process, ensuring the environmental stability and result reliability of the measurement instruments. 2. The application deeply integrates time-of-use electricity pricing policy and load forecasting, formulates an intelligent "peak load shifting" strategy, stores energy during valley electricity pricing period, releases energy preferentially during peak electricity pricing period, and dynamically optimizes and allocates during flat period, avoids insufficient or excessive energy storage, maximizes the utilization of the capacity of the energy storage device, and simultaneously, through the optimized collaborative operation of the cold and heat source module and the energy storage module, reduces the operation time and the number of frequent start-stop of the cold and heat source unit during the peak electricity pricing period, so that it works more in the high-efficiency operating condition area, thereby significantly reducing the overall energy consumption and electricity cost of the system under the premise of meeting the extremely high control accuracy, and the economic benefit is outstanding. 3. The control instruction is dynamically corrected through real-time monitoring data, and the model and strategy parameters are continuously optimized according to historical prediction errors, so that the system can effectively respond to uncertain disturbances such as weather changes and internal work plan adjustments, and the smooth control of the mode switching process further reduces the equipment impact and parameter disturbance, improves the stability and life of the collaborative operation of the entire air conditioning system and each device, and ensures the continuity and reliability of the laboratory environment control. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0016] Figure 1 The application discloses an optimal control method of a measurement laboratory cold and heat storage air conditioning system. DETAILED DESCRIPTION

[0017] The following will be described in combination with the drawings Figure 1 The application will be further described in detail. EMBODIMENT

[0018] The application discloses an optimal control method of a measurement laboratory cold and heat storage air conditioning system.

[0019] Referring to Figure 1 An optimal control method of a measurement laboratory cold and heat storage air conditioning system, the air conditioning system comprising a cold and heat source module, a cold and heat storage module, a terminal treatment module, a temperature and humidity monitoring module and a central control module, the optimal control method comprising the following steps: Step S1: The central control module uses a machine learning model to predict the hourly cooling load and heating load in the future preset period based on the historical operation data of the measurement laboratory, the outdoor weather forecast data in the future preset period and the internal work plan of the laboratory, and generates hourly load prediction data; Step S2: The central control module acquires time-of-use electricity price information, divides a day into peak electricity price period, flat electricity price period and valley electricity price period based on the time-of-use electricity price information, and generates electricity price period division data; Step S3: The central control module formulates an energy storage strategy based on the hourly load prediction data and the electricity price period division data; specifically, in the valley electricity price period, when there is a load demand in the predicted future peak electricity price period, the cold and heat source module is controlled to run and store energy in the cold and heat storage module until the rated capacity is reached or the valley period ends, and the temperature and humidity in the laboratory are maintained in the first set range by the terminal processing module during the energy storage period; Step S4: The central control module formulates an energy release and regular operation strategy based on the hourly load prediction data, the electricity price period division data and the real-time energy storage state of the cold and heat storage module; specifically, it includes: Step S41: In the peak electricity price period, the cold and heat storage module is preferentially controlled to release energy to match the load demand, and when the energy release is insufficient, the cold and heat source module is started to assist energy supply; Step S42: In the flat electricity price period, the proportion of direct energy supply of the cold and heat source module and energy release of the cold and heat storage module is dynamically adjusted, and whether to perform supplementary energy storage in the flat period is determined based on the prediction of the energy storage demand in the next peak period; Step S5: The temperature and humidity monitoring module collects temperature and humidity data sets corresponding to multiple monitoring points in the metering laboratory in real time and feeds back to the central control module; the central control module compares the temperature and humidity data sets with the set values, and when the deviation exceeds the second set range, the supply air parameters of the terminal processing module and the operating state of the cold and heat storage module are dynamically adjusted by the optimization control algorithm; Step S6: During the switching process of the cold and heat storage module and the cold and heat source module, the central control module adopts a feedforward-feedback composite control strategy to adjust the valve opening and the water pump frequency, so that the change rate of the supply air parameters of the terminal processing module during the switching process does not exceed the preset limit value; Step S7: The central control module calculates the prediction error, and according to the error result, the machine learning model is corrected and the energy storage strategy and the energy release strategy are optimized.

[0020] Specifically, in the embodiment of the present application, the implementation process of the optimization control method is as follows: the central control module first performs load prediction, and the specific operation principle is as follows: the central control module calls the historical operation data of the measurement laboratory stored in the database, which includes the actual cold and heat load records in different seasons and time periods in the past few months or even years, the corresponding outdoor temperature and humidity and solar radiation intensity historical values, and the laboratory internal equipment start-stop and personnel access logs; at the same time, the central control module obtains the outdoor temperature, relative humidity and solar radiation intensity prediction data within 24 hours or 48 hours in the future, and combines the next day equipment operation schedule and personnel activity arrangement provided by the laboratory management department to form a complete input data set containing external environment and internal disturbance; based on this, the central control module processes by using a pre-trained machine learning model, which learns the complex nonlinear mapping relationship between the meteorological parameters, work plan and load value in the historical data, and calculates the future working condition characteristics hour by hour, the internal calculation process involves the extraction, forgetting and memory of time series characteristics, and finally outputs the predicted cold load value and heat load value of each hour in the future preset period, that is, generates hourly load prediction data with high time resolution; then, the central control module obtains the next day time-of-use electricity price table, which specifies the electricity price level corresponding to different time points; the algorithm built-in the module automatically classifies the 24 hours of a day into "peak period" with expensive electricity price, "flat period" with medium electricity price and "valley period" with low electricity price according to the electricity price, and records the start and end time of each period to generate electricity price period division data, providing a benchmark for subsequent economic scheduling. Subsequently, when formulating the energy storage strategy, the central control module analyzes the hourly load prediction data and the electricity price period division data: the core calculation logic is that in the upcoming valley electricity price period, the central control module scans the load prediction curve of the future peak electricity price period to identify the existing cold / heat load demand peak; then, according to the rated energy storage capacity of the cold and heat storage module, the maximum output capacity of the cold and heat source module in the energy storage working condition and the duration of the valley period, through an optimization calculation taking filling the future peak demand gap as the target and the valley period length as the constraint, the minimum cold / heat quantity required to be stored in the valley period and the recommended energy storage power curve (i.e. the optimal energy storage rate) for realizing the energy storage quantity are solved; after generating the control instruction, the central control module controls the cold and heat source module to start and deliver cold or heat to the cold and heat storage module according to the calculated power curve in the valley period, at the same time, in order to ensure that the energy storage process does not affect the laboratory environment, the central control module strictly maintains the indoor temperature and humidity in the first set range of, for example, temperature set value ±0.5℃ and humidity set value ±3%RH, through real-time adjustment of the terminal processing module (such as air conditioning box), which is usually realized by dynamically adjusting the water valve opening degree of the surface cooler, the humidifier steam output and the frequency of the air supply fan through the proportional-integral-derivative control algorithm.In the energy release and normal operation strategy making phase, in the peak electricity price period, the module preferentially inquires the current energy storage of the cold and heat storage module, and compares it with the hourly predicted load of this period. If the energy storage is sufficient, the instruction is generated to control the cold / heat storage module to release the cold / heat medium on demand, and to provide energy to the terminal through plate exchange or direct mixing. If the real-time calculation finds that the energy storage is lower than the current load demand, the central control module immediately calculates the difference and starts the cold and heat source module to assist the energy supply with the minimum necessary output to ensure that the load is fully met. In the flat electricity price period, the central control module runs a dynamic optimization algorithm, which takes the current real-time load, the remaining energy storage of the cold and heat storage module, and the predicted load of the next peak period as input, and takes the total operation cost and energy efficiency of the system as the optimization target to solve the optimal ratio of direct energy supply of the cold and heat source module and energy release of the energy storage module online. At the same time, the algorithm will evaluate whether the existing remaining energy storage is sufficient to cover the predicted demand of the next peak period. If there is a gap in the prediction and it is economically feasible to perform partial energy storage during the flat period, the central control module will control the cold and heat source module to perform supplementary energy storage during the flat period to avoid high electricity cost or energy shortage in the peak period. In every moment of system operation, real-time feedback and dynamic adjustment are carried out simultaneously: multiple high-precision temperature and humidity sensors deployed in key areas of the laboratory collect data at a set frequency and upload them to the central control module to form a real-time temperature and humidity data set. The module calculates the average value of the data set and compares it with the target value set by the user. Once the absolute value of the temperature deviation exceeds 0.3℃ or the absolute value of the humidity deviation exceeds 2%RH (i.e. exceeds the second set range), the dynamic adjustment program is triggered. At this time, the optimization control algorithm built-in the central control module starts to work. The algorithm takes the control target of quickly eliminating the current temperature and humidity deviation and avoiding large fluctuations of the system. According to the size, direction and trend of the deviation, it performs multi-step prediction and rolling optimization through the internal system dynamic model, calculates the adjustment amount instruction of the terminal treatment module air supply valve, table cooler water valve, humidifier, and cold and heat source unit output power or energy storage module release valve, and executes it to form a fast closed-loop feedback, ensuring that the environmental parameters quickly return to the precise set range.When the system needs to switch between cold and heat storage supply and direct supply of cold and heat sources, the central control module enables a feedforward-feedback composite control strategy to ensure smooth transition: before the switching action is triggered, the feedforward control part will predict the trend of the terminal supply air temperature and humidity changes that may be caused at the switching moment according to the target operating mode to be switched to and the current state of the system through a simplified hydraulic and thermodynamic model; based on this prediction, the module sends frequency adjustment instructions to the related water pump and opening degree gradual change instructions to the waterway switching valve in advance to offset most of the expected disturbance; during the execution of the switching action, the feedback control part is started synchronously, which monitors the actual temperature and humidity change rate of the terminal supply air outlet in real time at a high frequency, and once it is found that the change rate approaches the preset safety limit value, the related actuator is adjusted through a proportional integral algorithm to compensate and correct the feedforward control, thereby ensuring that the entire switching process is smooth and disturbance-free, and the temperature and humidity in the laboratory cannot feel the step impact. Finally, in the system optimization and self-learning stage of step S7, the central control module will regularly start a batch analysis process: the module retrieves all the actual load data, actual weather data, system operating state logs and historical prediction data corresponding thereto in the past week for hourly comparison and analysis; by calculating the root mean square error and mean absolute percentage error between the actual value and the predicted value, the performance of the prediction model is quantified; then, the idea of back propagation or gradient descent is adopted to automatically adjust the feature weight parameters in the machine learning model used, so that the model can better fit the actual law in the next prediction; at the same time, the analysis process also evaluates the execution effect of the energy storage strategy and the energy release strategy in the past week, such as whether the energy storage amount in the valley period is appropriate, whether the energy release in the peak period is sufficient, whether the mode switching is smooth, etc., and based on the evaluation results, the key parameters in the strategy are adaptively optimized, such as the load prediction threshold for triggering the start of energy storage, the priority coefficient for determining the output ratio of the cold and heat source and the energy storage module, etc., so that the entire control system has the ability to continuously improve itself, constantly improving the control accuracy and economy.

[0021] Further, the step S1 comprises the following steps: Step S11: obtaining the corresponding historical operating data in the metering laboratory, the outdoor weather forecast data in the future preset period and the internal work plan of the laboratory; Step S12: feature extraction is performed on the historical operating data to obtain historical load feature data; Step S13: future environment and internal disturbance feature analysis is performed based on the outdoor weather forecast data and the internal work plan of the laboratory to generate future working condition feature data; Step S14: the historical load feature data and the future working condition feature data are input into the machine learning model, and then hourly load prediction data is output.

[0022] Specifically, in the embodiments of the present application, the load prediction process is implemented as follows: the central control module first calls the historical operation data of the measurement laboratory in the past at least one complete annual cycle from the built-in historical database, which contains the hourly actual cold load and heat load values recorded by time stamp, the corresponding outdoor dry bulb temperature, wet bulb temperature, total solar radiation intensity and internal state record, the latter covers the on-off timing of the main measurement equipment, heating power curve and internal thermal disturbance caused by personnel access; at the same time, the central control module obtains the outdoor weather forecast data for the next 24 to 48 hours through the data interface in real time, the time resolution is not less than 1 hour, and the elements at least include temperature, relative humidity, cloud cover and solar radiation intensity forecast value; in addition, the central control module also imports the detailed work plan of the same period from the laboratory management system, which clearly shows the scheduled operation time period of each precision equipment, the experimental project to be executed and the heat and humidity load grade it may generate. Then, deep feature extraction is performed on the historical operation data, the operating principle of which is to use time series analysis method to decompose the historical load data into baseline component reflecting long-term trend, cycle component with daily period and random fluctuation component, and calculate the dynamic correlation coefficient between historical load and different outdoor meteorological parameters, identify the sensitivity change rule of load to a certain meteorological factor under a certain season or weather mode; at the same time, this step also identifies and quantifies the internal load mutation points in the historical data caused by equipment start-stop or personnel gathering, extracts the typical intensity, duration and contribution proportion of these internal disturbance events to the total load, and finally integrates the regularity parameters and feature patterns obtained through statistical analysis, frequency domain transformation and pattern recognition into structured historical load feature data. Subsequently, based on the obtained outdoor weather forecast data and laboratory internal work plan, the future environment and internal disturbance features are analyzed, and the calculation process is as follows: first, according to the temperature, humidity and solar radiation intensity in the weather forecast data, combined with the thermal parameters of the building envelope, the heat load calculation core algorithm based on the transfer function method or response coefficient method is used to calculate the heat transfer load of the envelope caused by the change of outdoor weather conditions and the solar radiation heat gain load in the future every hour; then, combined with the work plan, the running period of each equipment and the personnel activity period in the plan are mapped into the corresponding internal heat and humidity load intensity time series, which is completed according to the equipment nameplate power, personnel density and internal disturbance model summarized in the historical feature data; finally, the calculated external disturbance load time series and internal disturbance load time series are superimposed, and the nonlinear interaction between them is considered to generate future working condition feature data.Finally, the historical load characteristic data and the future working condition characteristic data are input into the machine learning model which is trained in advance, and the model has established a complex nonlinear mapping relationship from the "historical characteristic mode" and the "future working condition characteristic" to the "future load value" by learning a large number of historical samples. When the model runs, firstly, the memory unit is used to capture the long-term dependence and periodicity in the historical load characteristic data, and form a code for the current "state" of the system; then, the code and the input future working condition characteristic data are fused and nonlinearly transformed in multiple fully connected layers of the model, and the final accurate cold load prediction value and heat load prediction value of each hour in the future preset period are obtained in the output layer through forward propagation calculation, layer-by-layer weighted summation and application of the activation function, that is, the hourly load prediction data.

[0023] Further, the step S3 comprises the following steps: Step S31: identifying the load demand peak value and duration of the future peak electricity price period based on the hourly load prediction data; Step S32: calculating the minimum energy storage amount and the optimal energy storage rate required in the valley electricity price period in combination with the electricity price period division data and the rated capacity of the cold and heat storage module; Step S33: generating an energy storage control instruction based on the load demand peak value and duration of the future peak electricity price period, controlling the cold and heat source module to operate at the optimal energy storage rate, and monitoring the energy storage state of the cold and heat storage module and the temperature and humidity in the laboratory in real time to ensure that the temperature and humidity are maintained in the first set range.

[0024] Specifically, in the embodiment of the present application, the energy storage strategy formulation process is implemented as follows: the central control module first performs step S31, and the operating principle is to scan and analyze all the predicted load data falling within the interval based on the future 24-hour hourly load prediction data curve generated in step S1 and the peak electricity price period interval divided in step S2; the algorithm built in the module identifies the highest point of the predicted cooling load or heating load value in the peak period, i.e. the peak load demand, through traversal comparison, and records the specific time point at which the peak value occurs; at the same time, the algorithm not only focuses on the single-point peak value, but also needs to determine the duration of the peak load, and the calculation process is to set a load threshold, and then trace back forward and backward in the peak electricity price period to find the start time and end time at which the predicted load value is continuously higher than the threshold, and the difference between the two time points is the substantial duration of the peak load, so as to accurately quantify the core load total amount and time span that need to be coped with in the future peak period. Then, the core optimization calculation is performed, and the operating process is: taking the peak load demand and duration of the future peak period as the target, taking the length of the upcoming valley electricity price period as the time window, and taking the rated energy storage capacity of the cold and heat storage module as the physical upper limit; first, the minimum required energy storage amount is calculated, which is not simply the integral of the peak period predicted load, but considers the efficiency loss during actual energy release and the possible base load, and through an algorithm containing the energy storage module energy release efficiency coefficient and the valley period base load deduction, the minimum cold or heat value that must be stored in the valley period in order to cover the future peak period peak demand is obtained; then, the optimal energy storage rate is calculated, which is a dynamic optimization problem, and the module needs to solve what power curve to charge the energy storage module in the given valley period length, so as to meet the minimum energy storage amount on the premise, and at the same time, consider three constraints: one is the rated maximum output limit of the cold and heat source module, two is that the energy storage process itself cannot cause the laboratory temperature and humidity to exceed the tolerance, and three is to make the cold and heat source unit run in the load interval with the highest energy efficiency as much as possible; the algorithm usually adopts a linear programming or model predictive control framework, and the optimization goal is that the energy storage module state reaches the target at the end of energy storage, and the comprehensive energy efficiency of the cold and heat source module is the highest in the entire valley period, and the recommended energy storage power at each time is iteratively solved, i.e. the optimal energy storage rate curve.Finally, the central control module converts the optimal energy storage rate curve into specific and executable control instruction sequences; at the beginning of the valley electricity price period, the module sends start instructions and initial target output values to the main controller of the cold and heat source module, and controls the pipeline valve to switch to the energy storage circuit; during the energy storage process, the module collects key state parameters of the cold and heat storage module in real time, calculates the stored energy in real time through integral operation, and compares it with the expected energy storage curve to form a closed-loop feedback, and fine-tune the output of the cold and heat source module to track the optimal energy storage rate curve; at the same time, the temperature and humidity monitoring module continuously feeds back the environmental data in the laboratory to the central control module, which compares these data with the first set range, and once it is found that the energy storage process causes the indoor temperature and humidity to approach the limit, it will immediately intervene, and the adjustment principle is to dynamically redistribute the output of the cold and heat source module: part of the energy continues to be used for energy storage, and another part of the energy is used to prioritize the immediate needs of the laboratory environment through bypass or direct adjustment of the end treatment device to ensure that the environmental parameters are always stable within the permitted range. This process is achieved through a fast-response multivariable decoupling control algorithm that can calculate and coordinate the control amount of the energy storage rate and the end treatment device online, thereby ensuring the core environmental control accuracy of the metrology laboratory while achieving the economic energy storage goal.

[0025] Further, the step S41 comprises the following steps: Step S411: At the beginning of the peak electricity price period, the current available energy storage amount of the cold and heat storage module and the predicted load of the corresponding period in the hourly load prediction data are obtained; Step S412: The current available energy storage amount is compared with the predicted load, and if the available energy storage amount is sufficient, a plan for releasing energy from the cold and heat storage module to meet the load is made; Step S413: If the available energy storage amount is insufficient, control instructions are generated to start the cold and heat source module to supplement the output, and the energy release rate of the cold and heat storage module is adjusted.

[0026] Further, the step S5 comprises the following steps: Step S51: The temperature and humidity monitoring module collects temperature and humidity data of multiple monitoring points in the metrology laboratory at a preset sampling frequency to form a temperature and humidity data set; Step S52: The central control module calculates the deviation of the average value of the temperature and humidity data set from the set value; Step S53: When the absolute value of the temperature deviation exceeds 0.3℃ or the absolute value of the humidity deviation exceeds 2%RH, it is determined that the deviation is out of tolerance, and a dynamic adjustment program is started, which calculates and outputs adjustment amounts of the supply air volume, supply air temperature and supply air humidity of the end treatment module, and correction instructions of the output of the cold and heat source module or the energy release rate of the cold and heat storage module according to the deviation size and trend through an optimization control algorithm.

[0027] Specifically, in the embodiment of the present application, the real-time feedback and dynamic adjustment process of step S5 is implemented as follows: the central control module first relies on a temperature and humidity monitoring network composed of multiple high-precision temperature and humidity sensors deployed in key areas of the metrology laboratory. These sensors synchronously collect the temperature and relative humidity raw electrical signals at their locations at a preset high frequency. After analog-to-digital conversion and filtering processing, real-time data points with time stamps and spatial position labels are formed. These data collected from all monitoring points are arranged into a temperature and humidity data set in the central control module. Subsequently, the central control module performs real-time statistical analysis on the data set. The calculation process is as follows: for the temperature parameter, the temperature readings of all effective monitoring points at the same sampling time are taken to calculate the arithmetic mean value, which represents the overall temperature level of the laboratory at that time. The average value is subtracted from the user's preset precise target temperature value to obtain the real-time overall temperature deviation value. For the humidity parameter, the same calculation logic is used. The average value of the humidity readings of all monitoring points is first calculated, and then subtracted from the preset target humidity value to obtain the overall humidity deviation value.Subsequently, the core logic of step S53 is triggered, the central control module compares the calculated absolute value of the real-time temperature deviation with the preset threshold value, and at the same time compares the absolute value of the humidity deviation with another preset threshold value, as long as the absolute value of any one of the deviation exceeds the corresponding threshold value, the system determines that the laboratory environment is in the state of "out of tolerance", and automatically starts the advanced dynamic adjustment program; the execution of the program depends on the optimization control algorithm built in the central control module, the operating principle of the algorithm is: in each control period, not only the size and positive and negative direction of the deviation at the current time are considered, but also the change trend is estimated by analyzing the deviation data of a recent time sequence; then, a pre-established simplified mathematical model reflecting the dynamic characteristics of the air conditioning system is called, which describes the causal and lag relationship between the control instructions (such as the air supply valve opening, the water valve opening of the cooling coil, the power of the humidifier, and the output power of the cold and heat source unit) and the controlled variables (the average temperature and humidity of the laboratory); based on the model, the current system state and the predicted load disturbance in the near future, the algorithm solves an optimization problem in a rolling time domain, the objective function of the problem is to make the predicted future laboratory temperature and humidity values return to the set range as soon as possible and as smoothly as possible, while reducing the intensity of control action and energy consumption as much as possible; by solving this constrained optimization problem, the algorithm calculates the number of hertz that the frequency of the air supply fan frequency converter should be adjusted at the current time, the percentage of the opening that the cooling coil regulating valve should be increased, the percentage of the steam flow that the humidifier regulating valve should be adjusted, and if necessary, the output power of the cold and heat source unit should be increased in kilowatts or the opening of the energy release valve of the cold and heat storage module should be increased in percentage; these calculated specific adjustment amounts and correction instructions are immediately converted into standard control signals and issued to the corresponding actuators, thereby forming a fast, accurate and system-wide dynamic closed-loop feedback control, ensuring that any small environmental deviation can be quickly detected and effectively suppressed, maintaining the high precision and stability of the temperature and humidity in the metrology laboratory.

[0028] Further, the step S6 comprises the following steps: Step S61: Before the switching of the cold and heat storage module and the cold and heat source module, the central control module predicts the parameter change trend in the switching process through feedforward control based on the hourly load prediction data; Step S62: According to the parameter change trend, calculate and execute the pre-adjustment instructions for valve opening and pump frequency, and use feedback control to monitor the air supply parameter change rate of the terminal processing module in real time during the switching process, so that the air supply parameter change rate of the terminal during the switching process does not exceed the preset limit value.

[0029] Specifically, in the embodiment of the present application, the switching smooth control process of step S6 is implemented as follows: the central control module executes step S61 before the system is about to switch from the cold and heat storage module energy supply mode to the cold and heat source module direct energy supply mode, and the operation principle is to estimate the impact trend of the sudden change of the thermal parameters of the energy supply side on the air supply parameters of the terminal processing module at the switching action execution moment and the subsequent short period of time based on the future short-time high-precision hourly load prediction data, the residual energy storage and energy release state of the current cold and heat storage module, and the real-time performance curve of the cold and heat source unit, through a built-in feedforward control prediction model; the prediction model is essentially a simplified transfer function model or a state space model reflecting the dynamic relationship between the waterway valve opening and the water pump frequency and the terminal air supply temperature and humidity; the module uses this model to simulate the predicted trajectory of the change of the air supply temperature and humidity with time after only executing the valve switching instruction without any pre-compensation, taking the planned switching time as the starting point, so as to quantify the predicted maximum change rate of the air supply parameters and the time point at which it occurs. Then, the central control module starts feedforward compensation calculation according to the parameter change trend, and the core is to solve an optimization problem with the minimum switching impact as the target: that is, to find a set of opening change curves of the related regulating valves and adjustment curves of the primary side / secondary side water pump frequency, so that when the set of “pre-adjustment instructions” and “mode switching instructions” are superimposed on the system, the predicted air supply parameter disturbance can be maximally offset; the calculation is usually carried out in the framework of a model predictive control, based on the feedforward model, with the air supply temperature change rate not exceeding 0.2℃ / min and the air supply humidity change rate not exceeding 1%RH / min as the hard constraint conditions, and the smoothness of the valve and water pump action as one of the optimization targets, to solve the optimal pre-adjustment instruction sequence on-line and rolling; these instructions will be step by step executed within a pre-set time window before the switching action is triggered, for example, gradually closing the energy storage and release valve while gradually opening the cold and heat source water supply valve, and synchronously fine-tuning the speed of the related water pump, so that the hydraulic and thermal working conditions of the system are smoothly transitioned to the state close to the target after switching.During the whole process of the formal execution of the switching instruction and the superposition of the feedforward action, the feedback control loop runs synchronously at high speed: the high-response speed temperature and humidity sensors deployed at the air supply outlets of the end treatment devices collect real-time values of the air supply parameters at a frequency much higher than the conventional monitoring, and the central control module continuously calculates the instantaneous change rate; the real-time change rate data are continuously compared with the preset limit change rate threshold; once the monitored change rate approaches but has not yet exceeded the threshold, the feedback control algorithm immediately intervenes, which calculates the valve opening degree or water pump frequency correction amount according to the real-time deviation; this correction amount is superimposed in real time to online calibrate and compensate the feedforward control, so as to cope with random disturbances or model errors that the feedforward model fails to accurately predict, thereby forming a composite control structure of “feedforward rough prediction compensation and feedback fine tuning calibration”, and finally ensuring that the change rate of the air supply temperature and humidity delivered to the laboratory is strictly limited within the permitted range during the whole dynamic switching transition period, so as to fundamentally avoid the step fluctuation or out-of-tolerance of the laboratory internal environment parameters caused by the energy side mode switching, and to ensure the continuity of the metrological activities and the reliability of the data.

[0030] Further, the step S7 comprises the following steps: S71: The central control module periodically aggregates the actual cooling and heating load data, outdoor weather data, operating state data and corresponding historical prediction data; S72: Calculate the hourly error of the actual cooling and heating load data and the corresponding historical prediction data in the time period, and statistically analyze the error distribution characteristics, and adjust the weight coefficients in the machine learning model according to the error distribution characteristics, and evaluate the execution effect of the energy storage strategy and the energy release strategy, and optimize the threshold parameters and priority coefficients in the strategy for triggering energy storage, energy release and mode switching.

[0031] Specifically, the system optimization and self-learning process in step S7 of this embodiment is implemented as follows: The central control module automatically starts a background analysis and optimization routine according to a pre-set cycle, such as weekly or monthly. When the program is executed, the module first performs the data collection step, that is, extracts all timestamp-aligned actual measurement data from the historical database of distributed storage, including hourly actual cooling and heating load data recorded by the energy metering device, actual outdoor temperature, humidity and solar radiation intensity data recorded by the meteorological station, and time stamp data of energy storage and release status of cold and heat storage modules, start-up and shutdown times and load rate of cold and heat source units, opening degree of terminal valves and all mode switching events recorded in the system's own operation log. At the same time, the module simultaneously extracts hourly load forecast data, electricity price period pollen data and strategy execution plan data within the same time period, and arranges the actual dataset and historical forecast plan dataset in chronological order to form a complete data pair sequence for subsequent analysis. The system then enters the core error calculation and feature analysis stage. The central control module calculates the hourly absolute error by measuring the difference between the actual load value and the predicted load value for each hour, and further calculates the relative error percentage. At the same time, the error sequence is classified and aggregated for analysis under different conditions. For example, the mean and variance of the error are calculated separately for weekdays and weekends, the error distribution is calculated separately for sunny days and rainy days, and the error characteristics are calculated separately for high load periods and low load periods. Through this multi-dimensional statistical analysis, the system identifies the predictive model under what external conditions or operating conditions it has a systematic bias tendency. For example, it is found that the model continuously underestimates the cooling load when the afternoon solar radiation suddenly increases, or continuously overestimates the heating load when large laboratory equipment is shut down at the same time. These quantified error distribution characteristics include the central tendency of the bias, the degree of dispersion, and the correlation with other variables, providing accurate input for subsequent model correction. Based on the above error analysis results, the system initiates a parameter correction process for the machine learning model. The operating principle utilizes the error backpropagation mechanism. Specifically, the module uses the identified systematic error features as supplementary training samples and re-inputs them into the adjustment algorithm of the machine learning model. This algorithm calculates the loss function between the predicted output and the actual value of these error samples under the current internal weight coefficients of the model. Using the principle of gradient descent, it iteratively adjusts the weight coefficients and bias terms on each neuron connection in the model along the direction of decreasing loss function. The adjustment magnitude is controlled by the learning rate parameter and is proportional to the contribution of the corresponding input feature to the error. For example, if the analysis finds that the outdoor radiation intensity feature significantly contributes to the underestimation error of afternoon cooling load, the algorithm will increase the weights of the neural connections associated with this feature, making the model more sensitive to changes in radiation intensity in future predictions. This achieves adaptive calibration of the model, making it more closely match the actual load response characteristics of the specific metrology laboratory.While revising the model, the system simultaneously optimizes the control strategy parameters. The central control module evaluates the actual performance of the energy storage and release strategies in the previous cycle. Key evaluation indicators include the matching degree between the total energy storage during off-peak hours and the demand during peak hours, the number of mode switching times and stability, and the overall operating cost. By comparing the expected goals of the strategy with the actual results, the module uses heuristic search or Bayesian optimization methods to adjust key threshold parameters in the strategy. For example, it raises or lowers the percentage of load prediction threshold that triggers energy storage, adjusts the priority coefficient used to determine insufficient energy storage and the need for cold or heat source assistance, and optimizes the length of the safety time window used in the feedforward control for mode switching. The optimization of these parameters is based on the principle of maximizing economy, ensuring environmental control accuracy, and minimizing equipment wear. The optimal parameter combination for the new round is determined through simulation combined with historical data playback testing. Finally, all the revised model weight coefficients and optimized strategy parameters are updated to the online operating database of the central control module, replacing the old parameter set. This enables the entire control system to have more accurate prediction capabilities, more economical scheduling strategies, and more stable control performance in the following operating cycles, completing a complete self-learning and performance improvement closed loop.

[0032] Furthermore, the first setting range is a temperature setting value ±0.5℃ and a humidity setting value ±3%RH.

[0033] Furthermore, the machine learning model is one of the following: a long short-term memory network model, a support vector regression model, or a gradient boosting decision tree model.

[0034] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.

[0035] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0036] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. An optimized control method for a thermal energy storage air conditioning system in a metrology laboratory, characterized in that, The air conditioning system includes a cold and heat source module, a cold and heat storage module, a terminal processing module, a temperature and humidity monitoring module, and a central control module. The optimized control method includes the following steps: Step S1: Based on the historical operating data of the metrology laboratory, the outdoor weather forecast data for the future preset period, and the laboratory's internal work plan, the central control module uses a machine learning model to predict the hourly cooling load and heating load for the future preset period, generating hourly load prediction data. Step S2: The central control module acquires time-of-use electricity price information, and divides a day into peak electricity price period, flat electricity price period and valley electricity price period based on the time-of-use electricity price information, and generates electricity price period division data; Step S3: The central control module formulates an energy storage strategy based on the hourly load forecast data and the electricity price period division data; specifically: during the off-peak electricity price period, when it is predicted that there will be load demand during the future peak electricity price period, the cold and heat source module is controlled to operate and store energy into the cold and heat storage module until the rated capacity is reached or the off-peak period ends. During the energy storage period, the temperature and humidity in the laboratory are maintained within the first set range through the terminal processing module. Step S4: Based on the hourly load forecast data, the electricity price period division data, and the real-time energy storage status of the cold and heat storage modules, the central control module formulates energy release and normal operation strategies; specifically including: Step S41: During peak electricity price periods, prioritize controlling the energy release of the cold and heat storage modules to match load demand. When the energy release is insufficient, start the cold and heat source modules to provide auxiliary energy supply. Step S42: During the period of flat electricity price, dynamically adjust the ratio of direct energy supply from cold and heat source modules to energy release from cold and heat storage modules, and decide whether to supplement energy storage during normal periods based on the prediction and judgment of energy storage demand in the next peak period. Step S5: The temperature and humidity monitoring module collects temperature and humidity data sets corresponding to multiple monitoring points in the metrology laboratory in real time and feeds them back to the central control module; the central control module compares the temperature and humidity data sets with the set values, and when the deviation exceeds the second set range, it dynamically adjusts the air supply parameters of the terminal processing module and the operating status of the cold and heat storage module through the optimized control algorithm. Step S6: During the switching process between the cold and heat storage module and the cold and heat source module, the central control module adopts a feedforward-feedback composite control strategy to adjust the valve opening and water pump frequency so that the rate of change of the air supply parameters of the terminal processing module does not exceed the preset limit during the switching process. Step S7: The central control module calculates the prediction error and corrects and optimizes the energy storage and release strategies based on the error results.

2. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Obtain the corresponding historical operating data, outdoor weather forecast data for the future preset period, and internal work plan of the metrology laboratory; Step S12: Extract features from the historical operating data to obtain historical load feature data; Step S13: Based on the outdoor weather forecast data and the laboratory's internal work plan, analyze the characteristics of the future environment and internal disturbances to generate future working condition characteristic data; Step S14: Input the historical load characteristic data and future operating condition characteristic data into the machine learning model, and then output hourly load prediction data.

3. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Based on the hourly load forecast data, identify the peak load demand and duration during future peak electricity price periods; Step S32: Based on the electricity price period segmentation data and the rated capacity of the cold and heat storage modules, calculate the minimum energy storage required and the optimal energy storage rate for off-peak electricity price periods; Step S33: Generate energy storage control commands based on the peak load demand and duration during future peak electricity price periods, control the cold and heat source modules to operate at the optimal energy storage rate, and monitor the energy storage status of the cold and heat storage modules and the temperature and humidity in the laboratory in real time to ensure that the temperature and humidity are maintained within the first set range.

4. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S41 includes the following steps: Step S411: At the start of the peak electricity price period, obtain the current available energy storage of the cold and heat storage modules and the predicted load for the corresponding period in the hourly load prediction data; Step S412: Compare the currently available energy storage with the predicted load. If the available energy storage is sufficient, formulate a plan for the energy storage modules to release energy to meet the load. Step S413: If the available energy storage is insufficient, a control command is generated to start the cold and heat source modules to supplement the output, and at the same time the energy release rate of the cold and heat storage modules is adjusted.

5. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S5 includes the following steps: Step S51: The temperature and humidity monitoring module collects temperature and humidity data corresponding to multiple monitoring points in the metrology laboratory at a preset sampling frequency to form a temperature and humidity dataset; Step S52: The central control module calculates the deviation between the average value of the temperature and humidity dataset and the set value; Step S53: When the absolute value of the temperature deviation exceeds 0.3℃ or the absolute value of the humidity deviation exceeds 2%RH, it is determined to be out of tolerance. The dynamic adjustment program is started. The optimized control algorithm calculates and outputs the adjustment amount of the air volume, air temperature and air humidity of the terminal processing module according to the magnitude and trend of the deviation, as well as the correction command of the output of the cold and heat source module or the energy release rate of the cold and heat storage module.

6. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S6 includes the following steps: Step S61: Before switching between the cold and heat storage module and the cold and heat source module, the central control module predicts the parameter change trend during the switching process based on the hourly load prediction data through feedforward control. Step S62: Calculate and execute pre-adjustment commands for valve opening and pump frequency based on the parameter change trend, and use feedback control to monitor the change rate of air supply parameters of the terminal processing module in real time during the switching process, so that the change rate of terminal air supply parameters does not exceed the preset limit during the switching period.

7. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, Step S7 includes the following steps: S71: The central control module regularly summarizes actual cooling and heating load data, outdoor meteorological data, operating status data, and corresponding historical forecast data; S72: Calculate the hourly error between the actual heating and cooling load data and the corresponding time period in the historical prediction data, statistically analyze the error distribution characteristics, adjust the weight coefficients in the machine learning model according to the error distribution characteristics, evaluate the execution effect of the energy storage strategy and the energy release strategy, and optimize the threshold parameters and priority coefficients used to trigger energy storage, energy release and mode switching in the strategy.

8. The optimized control method for a metrology laboratory cold and heat storage air conditioning system according to claim 1, characterized in that, The first setting range is the temperature setting value ±0.5℃ and the humidity setting value ±3%RH.

9. A computer-readable storage medium, characterized in that: The system stores instructions that, when executed on a computer, cause the computer to perform an optimized control method for a metrology laboratory cold and heat storage air conditioning system as described in any one of claims 1 to 8.

Images