Energy-saving control scheduling method and system oriented to equipment mechanism

Abstract

The invention discloses an equipment mechanism-oriented energy-saving control scheduling method and system. The method comprises the following steps of collecting static data and operation data of core cold and heat source equipment in a building in real time; an energy consumption mechanism model is established and used for describing the rule that equipment energy consumption changes along with the running state; utilizing the energy consumption mechanism model, the historical operation data, the environment data and the people flow data in the building to predict energy consumption and load demand changes in a target time period through a pre-constructed prediction model; according to the prediction result of the energy consumption and the load demand change, in combination with a set optimization objective function, an optimization algorithm is adopted to carry out optimization solution on the start-stop state, the operation load and the water supply and return temperature of the core cold and heat source equipment, and an optimal scheduling scheme is generated; and converting the optimal scheduling scheme into a control instruction and issuing the control instruction to a corresponding building automation system or equipment control unit. According to the invention, the prediction precision and adaptive capacity of energy-saving control scheduling can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of building energy conservation, and in particular relates to an energy conservation control scheduling method and system oriented to equipment mechanism. Background Art

[0002] As the problem of building energy consumption becomes increasingly prominent, especially the increasing dependence of large commercial complexes, hospitals, data centers, etc. on cold and heat source equipment (such as chillers and heat pump units), how to achieve efficient operation and energy-saving control of cold and heat source equipment while ensuring load supply demand has become an important topic in the field of building energy management. In the prior art, the energy consumption optimization scheduling method for cold and heat source equipment mainly relies on empirical settings or simple rule-based strategies. For example, operation control is performed by setting a fixed start and stop schedule, a fixed load distribution ratio, or a preset supply and return water temperature parameter.

[0003] This type of method fails to make full use of real-time operating data and the physical characteristics of equipment, resulting in a disconnect between the scheduling strategy and the actual operating conditions; there is a lack of in-depth modeling and prediction of the laws of energy consumption changes, and it is unable to dynamically adapt to load fluctuations and environmental changes; the optimization goal is single, only considering the lowest energy consumption, ignoring the impact of equipment life and frequent starts and stops on system reliability; the scheduling plan is updated with a lag, making it difficult to promptly reflect the impact of equipment status changes or prediction errors.

[0004] The Chinese invention application with publication number CN118761859A discloses a building energy consumption prediction and control system based on machine learning. The data integration and processing module collects the location, design features and surrounding environment data of the building complex, and performs cleaning and normalization processing; the simulation parameter extraction module uses the computational CFD model to simulate the wind duct effect in the building complex, and extracts microenvironment change parameters from the CFD model data through time-frequency analysis. The microenvironment change parameters include local turbulence intensity and pressure fluctuations; the deep feature analysis module uses multidimensional scaling and principal component analysis to perform deep feature extraction on the extracted microenvironment change parameters, and identifies key variables that affect energy consumption; the key variable network module uses graph neural networks to construct an interaction network based on key variables that affect energy consumption, and predicts how key variables that affect energy consumption jointly affect the energy consumption of buildings; the energy consumption prediction model module develops an energy consumption prediction model, and uses machine learning methods to optimize and cross-validate the model's hyperparameters.

[0005] The above scheme focuses on modeling the impact of micro-environment changes on energy consumption at the building complex level, ignoring the direct driving effect of the operating status of the cold and heat source equipment inside the building on energy consumption; and mainly focuses on offline data collection and energy consumption forecast analysis, lacking the ability to collect equipment operation data in real time and dynamically adjust control instructions; resulting in insufficient level of refinement in the scheduling of the scheme, insufficient adaptability of the scheduling scheme and insufficient prediction accuracy. Summary of the invention

[0006] The present invention provides an energy-saving control scheduling method and system oriented to equipment mechanism, aiming to solve the problems existing in the prior art in energy-saving control scheduling, such as insufficient refinement level, insufficient adaptability of scheduling schemes and insufficient prediction accuracy.

[0007] In order to solve the above technical problems, the energy-saving control scheduling method proposed in the present invention comprises the following steps: Real-time collection of static data and operating data of core cold and heat source equipment in the building; Based on the collected data and combined with the physical working mechanism of the equipment, an energy consumption mechanism model is established to describe the law of equipment energy consumption changing with the operating status; Using the energy consumption mechanism model, historical operation data, environmental data and building traffic data, the energy consumption and load demand changes within the target period are predicted through a pre-built prediction model; Based on the predicted results of the energy consumption and load demand changes, combined with the set optimization objective function, an optimization algorithm is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment to generate an optimal scheduling plan; The optimal scheduling plan is converted into control instructions and sent to the corresponding building automation system or equipment control unit.

[0008] Preferably, the method also includes a feedback update step, which updates the prediction model parameters and optimizes the model weight coefficient according to the deviation between the actual operation results of the core cold and heat source equipment and the predicted results after issuing control instructions and collecting the actual operation results of the equipment.

[0009] Preferably, the energy consumption mechanism model adopts a mathematical fitting model based on the equipment thermodynamic performance curve, refrigeration efficiency and power curve.

[0010] Preferably, the energy consumption mechanism model includes an input layer, a physical modeling layer, an empirical correction layer and an output layer; wherein the physical modeling layer calculates the cold and hot loads according to the thermodynamic equations, and further calculates the preliminary energy consumption based on the calculated cold and hot loads combined with the energy efficiency ratio of the equipment; the empirical correction layer trains a regression model through actual collected data to correct the theoretical calculation deviation of the preliminary energy consumption.

[0011] Preferably, the prediction model includes a long short-term memory network, an extreme gradient boosting regression model and a lightweight gradient boosting tree.

[0012] Preferably, the optimization objective function is specifically:

[0013] In the formula, N is the number of devices, T is the number of device scheduling time periods, is the power consumption of the i-th device at time t, is the total load demand predicted for time t, is the load provided by the i-th device at time t, and are the start and stop states of the i-th device at time t and t-1, respectively. and is the weight factor for adjusting the importance of each part.

[0014] Preferably, the optimization objective function satisfies the following conditions:

[0015]

[0016]

[0017] In the formula, is the small tolerance set, is the maximum rated cooling / load of the ith device.

[0018] Preferably, the optimization algorithm includes a mixed integer programming algorithm, a genetic algorithm and a particle swarm optimization algorithm.

[0019] Preferably, the static data includes the model, rated power, cooling / heating capacity, rated supply / return water temperature range and start / stop control parameters of the cold and heat source equipment; the operating data includes the equipment start / stop status, actual supply / return water temperature, instantaneous load rate, energy consumption and environmental parameters.

[0020] Accordingly, the present invention also proposes an energy-saving control scheduling system oriented to equipment mechanism, and the system is used to implement the above energy-saving control scheduling method, including: Data acquisition module, used to collect static data and operation data of core cold and heat source equipment in the building in real time; A mechanism modeling module is used to establish an energy consumption mechanism model based on the static data and the operating data in combination with the physical working mechanism of the equipment, so as to describe the law of the change of the equipment energy consumption with the operating state; An energy consumption prediction module, used to use the energy consumption mechanism model and historical operation data to predict energy consumption and load demand changes within a target period through machine learning; The scheduling optimization module is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment according to the prediction results of the energy consumption and load demand changes, combined with the set optimization objective function, and generate the optimal scheduling plan by using the optimization algorithm; An instruction issuing module is used to convert the optimal scheduling plan into a control instruction and issue it to the corresponding building automation system or equipment control unit to achieve energy-saving operation control of cold and hot source equipment; Feedback update module predicts the model and optimizes model parameters based on the operation results.

[0021] Compared with the prior art, the present invention has the following technical effects: 1. The energy-saving control and scheduling method proposed in the present invention establishes an energy consumption mechanism model by combining the physical working mechanism of the cold and heat source equipment with the real-time operation data. Compared with the traditional method of energy consumption prediction that only relies on historical data, it can more accurately reflect the energy consumption change law of the equipment under different operating states and environmental conditions, and improve the model's adaptability to changes in actual working conditions and prediction accuracy.

[0022] 2. The energy-saving control scheduling method proposed in the present invention adopts a machine learning algorithm based on the energy consumption mechanism model to predict the energy consumption and load demand in the future period, which can realize the early perception of the load change trend and avoid the problem of passive scheduling based only on the current working conditions in the traditional method, thereby realizing a more proactive and forward-looking energy-saving control strategy.

[0023] 3. The energy-saving control and scheduling method proposed in the present invention comprehensively considers multiple control parameters such as equipment start and stop status, operating load, supply and return water temperature, sets multi-objective optimization functions such as energy saving priority, load satisfaction, and equipment safety, and uses optimization algorithms for global scheduling. It can effectively reduce overall energy consumption while ensuring the internal cooling and heating load requirements of the building and the stable operation of the equipment, solving the problem of limited energy-saving effects under traditional single control strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flow chart of the energy-saving control scheduling method of the present invention. DETAILED DESCRIPTION

[0025] In order to make the objectives, technical solutions and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in combination with specific embodiments of the present application and with reference to the accompanying drawings.

[0026] Embodiment 1 An energy-saving control scheduling method oriented to equipment mechanism, such as Figure 1 As shown, the steps include steps one to five: Step 1: Collect static data and operating data of the core cold and heat source equipment in the building in real time.

[0027] First, static data is collected and entered for the core cold and heat source equipment inside the building (such as chillers, boilers, heat pump units, etc.), including but not limited to the following information: Equipment model, used to identify the specific specifications and models of the equipment, so as to query the factory parameters and performance curves of the equipment later; Rated power refers to the maximum power consumption of the equipment under rated conditions; Cooling / heating capacity, the cooling or heating capacity of the equipment under standard operating conditions; Rated supply and return water temperature range, the recommended supply and return water temperature range for the equipment under normal operating conditions (e.g. 7°C for supply water and 12°C for return water in cooling mode); Start-stop control parameters: including control logic constraints such as the minimum time interval allowed for start-stop, minimum operating time, and minimum shutdown time, which are used to prevent frequent start-stop of equipment from causing reduced efficiency or increased losses.

[0028] The above static data generally comes from the equipment factory documents, product manuals or on-site inspection data, and can be stored in the central database through manual entry or batch import. Static data is the basic characteristic data of the equipment. Once collected and imported, it can be repeatedly called in the subsequent operation process, and usually does not need to be collected repeatedly.

[0029] Secondly, the dynamic operation data generated during the operation of the equipment is collected in real time, including but not limited to the following information: Start / Stop status: indicates whether the device is currently in operation, usually indicated by a switch value (1-operation, 0-stop); Actual water supply temperature, the real-time temperature value of water supply on the output side of the equipment; Actual return water temperature, the real-time temperature value of the return water on the input side of the equipment; Instantaneous load rate, the percentage of the current actual load of the equipment to its rated load, which can be calculated based on parameters such as flow rate and temperature difference or directly output by the control system; Energy consumption power, the actual energy consumption power value of the device in the current unit time; Environmental parameters include temperature, humidity, air pressure and other environmental conditions inside and outside the building. These parameters have an important impact on the operating load and energy consumption of cold and heat source equipment.

[0030] Dynamic operation data can be collected in real time through the Building Management System (BMS), energy consumption monitoring system, Internet of Things sensors, etc., and automatically recorded in the data acquisition module or historical database according to the preset sampling period (such as every 5 minutes, every 15 minutes).

[0031] In practical applications, in order to ensure the accuracy and reliability of data collection, an anomaly detection mechanism can be set up. For example: perform boundary checks on the collected data to eliminate invalid data that exceeds a reasonable range; perform time synchronization on the sampled data to ensure that each data field has a unified timestamp; fill in or mark missing or abnormal values ​​to avoid affecting subsequent model training and prediction accuracy.

[0032] As an example, the following Table 1 is the static data collected in this step.

[0033]

[0034] Table 1 Static data Table 2 below shows the operating data collected in this step.

[0035]

[0036] Table 2 Operation data Step 2: Based on the collected data, an energy consumption mechanism model is established in combination with the physical working mechanism of the equipment to describe the law of equipment energy consumption changing with the operating state. The energy consumption mechanism model should be a multi-input multi-output (MIMO) data-driven model. Its core functions include calculating the energy consumption level at the corresponding time according to the equipment operating state, and supporting the load demand forecast and energy consumption forecast in the subsequent step 3.

[0037] The energy consumption mechanism model adopts a mathematical fitting model based on the equipment thermodynamic performance curve, refrigeration efficiency and power curve.

[0038] The energy consumption mechanism model includes an input layer, a physical modeling layer, an experience correction layer and an output layer; wherein the physical modeling layer calculates the cold and hot loads according to the thermodynamic equations, and further calculates the preliminary energy consumption based on the calculated cold and hot loads combined with the energy efficiency ratio of the equipment; the experience correction layer trains a regression model through actual collected data to correct the theoretical calculation deviation of the preliminary energy consumption.

[0039] In this embodiment, an energy consumption mechanism model is provided to describe the law of energy consumption of core cold and heat source equipment in a building changing with the operating state. The structure of the energy consumption mechanism model includes: First, the collected static parameters and dynamic operation data of the equipment are received through the input layer, where the static parameters include but are not limited to equipment model, rated power, maximum cooling or heating capacity, energy efficiency ratio, design flow and design head; the dynamic operation data include but are not limited to equipment start and stop status, current load rate, supply and return water temperature, chilled water flow, cooling water flow, inlet and outlet pressure, current, voltage, power and external ambient temperature and humidity.

[0040] Then, enter the modeling layer, which includes the physical modeling module and the experience correction module. The physical modeling module calculates the current cooling and heating load of the equipment based on the basic principles of thermodynamics through the following formula:

[0041] In the formula, Q is the load, m is the flow rate, is the specific heat capacity, and The inlet water temperature and return water temperature are the inlet and outlet water temperature differences respectively; the input power is derived through the performance curve and Q of the equipment.

[0042] After obtaining the preliminary energy consumption value through physical modeling, the empirical correction module is entered. This module constructs a regression model and uses the historically collected equipment operation data to correct the calculation results of the physical model to eliminate the deviations introduced by factors such as equipment aging, environmental changes, and measurement errors. The feature processing unit selects and normalizes the input features and trains an error compensation function based on actual operating conditions, thereby improving the accuracy and robustness of energy consumption prediction.

[0043] Through the output layer, the energy consumption mechanism model can output the following data: the predicted value of energy consumption per unit time of a single cold and heat source equipment in the future target period; the predicted curve of aggregated total energy consumption of multiple equipment; the predicted trend of load rate change of a single device and the overall system; and the predicted demand for cold load or heat load of the target building in the future period.

[0044] Step three, using the energy consumption mechanism model, historical operation data, environmental data and building traffic data, predict the energy consumption and load demand changes within the target period through a pre-built prediction model.

[0045] The prediction models include long short-term memory networks, extreme gradient boosting regression models and lightweight gradient boosting trees.

[0046] Specifically, taking the extreme gradient boosting regression (XGBoost) model as an example, the input of this model includes static feature data of the equipment, real-time operation data of the equipment, output of the equipment energy consumption mechanism model and historical energy consumption load data. Output the predicted energy consumption value at each time point in the future target period and the predicted load value at each time point in the future target period. The output task type setting of XGBoost determines its output type. If the task type is a classification task, the model outputs probability; if the task type is set to a regression task, the output is a continuous real value (i.e., regression prediction value). In the energy consumption load prediction scenario of this step, the regression task of XGBoost is used.

[0047] When training the XGBoost model, it is necessary to extract fixed features related to energy consumption from the static feature data of the equipment, such as rated power, cooling / heating capacity, etc.; extract time series features from the real-time operation data of the equipment, such as current load rate, supply and return water temperature, ambient temperature and humidity; call the energy consumption mechanism model, and derive the output of the energy consumption mechanism model based on the current operation data; combine the output of the above energy consumption mechanism model with the historical energy consumption load data to form a unified training sample set.

[0048] The specific training process is as follows: define the XGBoost regression model and set basic hyperparameters such as learning rate, tree depth, and subsampling ratio; use historical data to train the model with the goal of minimizing the energy consumption prediction error; use cross-validation to tune the hyperparameters to improve the model's generalization ability; save the optimal model after training is completed.

[0049] The trained XGBoost model can be used to predict energy consumption and load in future time periods: obtain the expected static and dynamic input features within the prediction target period; use the energy consumption mechanism model to perform feature inference on the operating status of the prediction period, generate operating status features, and fuse the static and dynamic input features with the operating status features to form the final prediction feature vector; call the optimal model for prediction and output the predicted energy consumption and predicted load.

[0050] In this embodiment, two types of predictors are trained for each device: predicting the future hourly load rate and predicting the future hourly energy consumption using the load rate. As an example, the following Table 3 provides real-time data for device A to input into the optimal model for prediction.

[0051]

[0052] Table 3 Real-time data of device A Step 4: Based on the predicted results of the energy consumption and load demand changes and the set optimization objective function, an optimization algorithm is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment to generate an optimal scheduling plan.

[0053] The optimization objective function is specifically:

[0054] In the formula, N is the number of devices, T is the number of device scheduling time periods, is the power consumption of the i-th device at time t, is the total load demand predicted for time t, is the load provided by the i-th device at time t, and are the start and stop states of the i-th device at time t and t-1, respectively. and is the weight factor for adjusting the importance of each part.

[0055] The optimization objective function satisfies the following conditions:

[0056]

[0057]

[0058] In the formula, is the small tolerance set, is the maximum rated cooling / load of the ith device.

[0059] The optimization algorithms include mixed integer programming (MIP), genetic algorithm (GA) and particle swarm optimization (PSO). The mixed integer programming algorithm can be solved using the gurobipy library, the genetic algorithm can be implemented using DEAP and PyGMO libraries, and the particle swarm optimization algorithm can be implemented using PySwarms and SciPy libraries.

[0060] As an example, the load demand forecast for the next 24 hours is obtained through step 3, with 1 data point per hour, recorded as . Establish equipment mechanism models for the three chillers in the building (numbered A, B, and C), including: rated cooling capacity , which are 500kW, 600kW, and 400kW respectively; the load rate and power consumption relationship curve provided by the energy consumption mechanism model; and the small tolerance (minimum operating load limit) is 50%. Based on the above functions, a specific optimization objective function is established:

[0061] In the above formula, the weight factor and Set them to 100 and 5 respectively, call the mixed integer programming optimizer, input equipment modeling parameters, load forecast value, optimization objective function, load satisfaction constraints and load rate limit. After solving, output the start and stop status of each equipment at each time and the corresponding load rate.

[0062] Step 5: Convert the optimal scheduling solution into control instructions and send them to the corresponding building automation system or equipment control unit. Based on the example provided in step 4, this step converts the scheduling solution obtained by optimization into executable control instructions, such as: Unit A runs at 70% load at 0-8; Unit B runs at 65% load from 8 to 18 o'clock; Unit C is turned on as an auxiliary during peak hours, with a load of 40%.

[0063] In some other embodiments of the present invention, the method also includes step six, which is specifically: a feedback update step, in which after issuing control instructions and collecting actual operating results of the equipment, the feedback update step updates the prediction model parameters and optimizes the model weight coefficient according to the deviation between the actual operating results of the core cold and heat source equipment and the predicted results.

[0064] Embodiment 2 This embodiment is an energy-saving control scheduling system oriented to equipment mechanism, and the system is used to implement the energy-saving control scheduling method as described in the first embodiment, including: Data acquisition module, used to collect static data and operation data of core cold and heat source equipment in the building in real time; A mechanism modeling module is used to establish an energy consumption mechanism model based on the static data and the operating data in combination with the physical working mechanism of the equipment, so as to describe the law of the change of the equipment energy consumption with the operating state; An energy consumption prediction module, used to use the energy consumption mechanism model and historical operation data to predict energy consumption and load demand changes within a target period through machine learning; The scheduling optimization module is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment according to the prediction results of the energy consumption and load demand changes, combined with the set optimization objective function, and generate the optimal scheduling plan by using the optimization algorithm; An instruction issuing module is used to convert the optimal scheduling plan into a control instruction and issue it to the corresponding building automation system or equipment control unit to achieve energy-saving operation control of cold and hot source equipment; Feedback update module predicts the model and optimizes model parameters based on the operation results.

[0065] The above is only a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several modifications and improvements without departing from the inventive concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. An energy-saving control scheduling method oriented to equipment mechanism, characterized in that: The following steps are involved: Real-time collection of static data and operating data of core cold and heat source equipment in the building; Based on the collected data and combined with the physical working mechanism of the equipment, an energy consumption mechanism model is established to describe the law of equipment energy consumption changing with the operating status; Using the energy consumption mechanism model, historical operation data, environmental data and building traffic data, the energy consumption and load demand changes within the target period are predicted through a pre-built prediction model; Based on the predicted results of the energy consumption and load demand changes, combined with the set optimization objective function, an optimization algorithm is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment to generate an optimal scheduling plan; The optimal scheduling plan is converted into control instructions and sent to the corresponding building automation system or equipment control unit.

2. The method according to claim 1, characterized in that The method also includes a feedback update step, which updates the prediction model parameters and optimizes the model weight coefficient according to the deviation between the actual operation results of the core cold and heat source equipment and the predicted results after issuing control instructions and collecting the actual operation results of the equipment.

3. The method according to claim 1, characterized in that The energy consumption mechanism model adopts a mathematical fitting model based on the equipment thermodynamic performance curve, refrigeration efficiency and power curve.

4. The method according to claim 1, characterized in that The energy consumption mechanism model includes an input layer, a physical modeling layer, an experience correction layer and an output layer; wherein the physical modeling layer calculates the cold and hot loads according to the thermodynamic equations, and further calculates the preliminary energy consumption based on the calculated cold and hot loads combined with the energy efficiency ratio of the equipment; the experience correction layer trains a regression model through actual collected data to correct the theoretical calculation deviation of the preliminary energy consumption.

5. The method according to claim 1, characterized in that The prediction models include long short-term memory networks, extreme gradient boosting regression models and lightweight gradient boosting trees.

6. The method according to claim 1, characterized in that The optimization objective function is specifically: In the formula, N is the number of devices, T is the number of device scheduling time periods, is the power consumption of the i-th device at time t, is the total load demand predicted for time t, is the load provided by the i-th device at time t, and are the start and stop states of the i-th device at time t and t-1, respectively. and is the weight factor for adjusting the importance of each part.

7. The method according to claim 6, characterized in that The optimization objective function satisfies the following conditions: In the formula, is the small tolerance set, is the maximum rated cooling / load of the ith device.

8. The method according to claim 1, characterized in that The optimization algorithms include mixed integer programming algorithm, genetic algorithm and particle swarm optimization algorithm.

9. The method according to claim 1, characterized in that: The static data includes the model, rated power, cooling / heating capacity, rated supply and return water temperature range and start / stop control parameters of the cold and heat source equipment; the operating data includes the equipment start / stop status, actual supply and return water temperature, instantaneous load rate, energy consumption and environmental parameters.

10. An energy-saving control and scheduling system oriented to equipment mechanism, characterized in that: The system is used to implement the energy-saving control scheduling method according to any one of claims 1 to 9, including: Data acquisition module, used to collect static data and operation data of core cold and heat source equipment in the building in real time; A mechanism modeling module is used to establish an energy consumption mechanism model based on the static data and the operating data in combination with the physical working mechanism of the equipment, so as to describe the law of the change of the equipment energy consumption with the operating state; An energy consumption prediction module, used to use the energy consumption mechanism model and historical operation data to predict energy consumption and load demand changes within a target period through machine learning; The scheduling optimization module is used to optimize the start and stop status, operating load and supply and return water temperature of the core cold and heat source equipment according to the prediction results of the energy consumption and load demand changes, combined with the set optimization objective function, and generate the optimal scheduling plan by using the optimization algorithm; An instruction issuing module is used to convert the optimal scheduling plan into a control instruction and issue it to the corresponding building automation system or equipment control unit to achieve energy-saving operation control of cold and hot source equipment; Feedback update module predicts the model and optimizes model parameters based on the operation results.

Images