Method for establishing device model, method for optimizing heating system, and electronic device

Abstract

The embodiment of the application provides a device model establishing method, a heating ventilation system optimizing method and an electronic device. The input of the device model comprises a first working parameter of a device, and the output of the device model is a predicted value of a second working parameter related to the first working parameter. The method comprises the following steps: obtaining first sampling data of the device in a working process and reference working data of the device, wherein the first sampling data and the reference working data both comprise the first working parameter and the corresponding second working parameter; performing iterative training on an initial device model corresponding to the device by using the first sampling data and the reference working data, and correcting the reference working data based on the model obtained through iteration in the iterative training process, so as to obtain the device model, and the parameter optimization effect of the heating ventilation system is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data centers, and more particularly, to a device model establishing method, a heating and ventilation system tuning method and an electronic device. BACKGROUND

[0002] With the development of data center related technologies, how to reduce the energy consumption of the heating and ventilation system of the data center has become a key problem that needs to be solved in the industry. Only based on artificial experience to control and optimize the parameters of the heating and ventilation system has been increasingly unable to meet the energy saving needs.

[0003] At present, in the related art, a neural network model is trained to optimize the device parameters of the heating and ventilation system, and the related data of the heating and ventilation system is taken as input, and the neural network model outputs a predicted power consumption to optimize the device parameters of the heating and ventilation system. However, due to insufficient data of the heating and ventilation system, it is impossible to ensure that the model gives stable and reasonable output results within a feasible value range, resulting in poor parameter optimization effect. SUMMARY

[0004] The present application provides a device model establishing method, a heating and ventilation system tuning method and an electronic device to improve the parameter optimization effect of the heating and ventilation system.

[0005] In a first aspect, the present application provides a device model establishing method, the input of the device model includes a first working parameter of a device, and the output of the device model is a predicted value of a second working parameter related to the first working parameter. The method comprises:

[0006] Obtaining first sampling data of the device in the working process and reference working data of the device, wherein the first sampling data and the reference working data both include the first working parameter and the corresponding second working parameter;

[0007] Using the first sampling data and the reference working data to iteratively train an initial device model corresponding to the device, and correcting the reference working data based on the model obtained by iteration during the iterative training to obtain the device model.

[0008] Optionally, the obtaining of the first sampling data of the device in the working process and the reference working data of the device comprises:

[0009] Obtaining the first sampling data of the device in the working process, and all reference working data of the device within a value range of the first parameter;

[0010] Dividing the value range of the first parameter into a plurality of subintervals;

[0011] For each sub-interval, it is judged whether there is first sampling data in the sub-interval, if there is first sampling data in the sub-interval, the reference working data in the sub-interval is deleted, if there is no first sampling data in the sub-interval, the reference working data in the sub-interval is reserved.

[0012] Optionally, the initial equipment model corresponding to the equipment is iteratively trained by using the first sampling data and the reference working data, and the reference working data is corrected based on the model obtained through iteration in the iterative training process, so as to obtain the equipment model, comprising:

[0013] The initial equipment model is trained by using the first sampling data and the reference working data to obtain a first model, the first working parameter in the reference working data is input into the first model to obtain a predicted value of the second working parameter output by the first model, the second working parameter in the reference working data is updated to the predicted value of the second working parameter output by the first model, and the step is repeatedly executed until the iterative stop condition is met, and the obtained first model is determined as the equipment model.

[0014] Optionally, the method further comprises:

[0015] The second sampling data in the working process of the equipment in a recent first time period is obtained.

[0016] For each sub-interval, it is judged whether there is second sampling data in the sub-interval, if there is second sampling data in the sub-interval, the first sampling data or the reference working data in the sub-interval is deleted, if there is no second sampling data in the sub-interval, the first sampling data or the reference working data in the sub-interval is reserved, and first data is obtained.

[0017] The equipment model is trained by using the first data, and the reference working data in the first data is corrected based on the model obtained through iteration in the iterative training process, so as to update the equipment model.

[0018] Optionally, the preset model of the equipment is one or a combination of a plurality of items in a linear regression model, a polynomial regression model and an empirical model.

[0019] Optionally, the method is applied to a heating and ventilation system, the equipment is an equipment in the heating and ventilation system, and the first working parameter and the second working parameter corresponding to the equipment and the equipment model comprise any one of the following:

[0020] The equipment is a cooling pump, the first working parameter is a cooling water flow, and the second working parameter is a motor frequency of the cooling pump.

[0021] The device is a cooling pump, the first working parameter is a motor frequency of the cooling pump, and the second working parameter is power consumption;

[0022] The device is a refrigeration pump, the first working parameter is a chilled water flow, and the second working parameter is a motor frequency of the refrigeration pump;

[0023] The device is a refrigeration pump, the first working parameter is a motor frequency of the refrigeration pump, and the second working parameter is power consumption;

[0024] The device is a cooling tower, the first working parameter is a fan frequency of the cooling tower, and the second working parameter is power consumption;

[0025] The device is a cooling tower, the first working parameter is a weather temperature of an environment where the heating and ventilation system is located, a temperature difference between inlet and outlet chilled water, and a water vapor ratio, and the second working parameter is an outlet water temperature of the cooling tower;

[0026] The device is a chiller, the first working parameter is a chilled water flow, and the second working parameter is a saturation temperature of an evaporator;

[0027] The device is a chiller, the first working parameter is a chilled water flow, and the second working parameter is a saturation temperature of a condenser;

[0028] The device is a chiller, the first working parameter is a saturation temperature of an evaporator, a saturation temperature of a condenser, and a load rate of the chiller, and the second working parameter is power consumption;

[0029] The device is a plate heat exchanger, the first working parameter is a chilled water flow, a chilled water flow, and a chilled water temperature, and the second working parameter is a chilled water temperature.

[0030] In a second aspect, the application provides a tuning method of a heating and ventilation system, comprising:

[0031] Obtaining device models of devices in the heating and ventilation system, wherein the device models are obtained in advance by using the method in the first aspect, and the device models comprise a model in which the second working parameter is power consumption;

[0032] Connecting the device models of the devices according to topological information of the heating and ventilation system to obtain a system model of the heating and ventilation system, and performing parameter optimization on the system model to obtain target parameters that make power consumption of the heating and ventilation system lowest.

[0033] Optionally, the method further comprises:

[0034] If a difference between the target parameters and field parameters of the heating and ventilation system is greater than or equal to a first preset threshold, adjusting the field parameters of the heating and ventilation system to the target parameters.

[0035] Optionally, further comprising:

[0036] If the difference between the target parameter and the field parameter of the HVAC system is less than the first preset threshold, and the difference between the power consumption of the HVAC system corresponding to the target parameter and the current power consumption of the HVAC system is greater than or equal to a second preset threshold, the field parameter of the HVAC system is adjusted to the target parameter.

[0037] Optionally, the HVAC system has multiple working modes, and the devices in working state are different in different working modes; the method further comprises:

[0038] Parameter optimization is performed on the system model composed of the device models in different working modes, to obtain the optimal working mode and the corresponding target parameter that make the power consumption of the HVAC system lowest in each of the current and future multiple second time periods.

[0039] Optionally, each working mode has a corresponding available condition, and the method further comprises:

[0040] It is determined whether the current working mode of the HVAC system is the optimal working mode;

[0041] If the current working mode is the optimal working mode, and the current working mode is available in one or more future second time periods, the current working mode is output as the recommended working mode;

[0042] If the current working mode is not the optimal working mode, and the current mode is available in multiple future second time periods, if the current working mode is the optimal working mode in any of the future second time periods, the current working mode is output as the recommended working mode;

[0043] If the current working mode is not the optimal working mode, and the current mode is not available in any of the future second time periods or the current working mode is not the optimal working mode in any of the future second time periods, the working mode with the lowest overall power consumption in the multiple future second time periods is output as the recommended working mode.

[0044] Optionally, the method further comprises:

[0045] If the recommended working mode is different from the current working mode, the working mode of the HVAC system is set to the recommended working mode.

[0046] In a third aspect, the present application provides an electronic device, comprising: a memory and a processor;

[0047] The memory is used to store a computer program;

[0048] The processor is configured to execute a computer program stored in the memory, and the computer program, when executed, causes the processor to perform the method of the first aspect or the second aspect.

[0049] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor performs the method of the first aspect or the second aspect.

[0050] In the method for establishing a device model, the method for optimizing a heating and ventilation system, and the electronic device provided in the present application, the reference working data is used to supplement the real data to solve the problem of missing real data in some value ranges. Meanwhile, the reference working data is constantly corrected to gradually reduce the error, so that the device model trained can stably and accurately output results in the complete value range, and the effect of parameter optimization is improved. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a device schematic diagram of a heating and ventilation system provided by an embodiment of the present application;

[0052] Figure 2 is a flowchart of a method for establishing a device model provided by an embodiment of the present application;

[0053] Figure 3 is a data fusion and training schematic diagram provided by an embodiment of the present application;

[0054] Figure 4 is a flowchart of a method for optimizing a heating and ventilation system provided by an embodiment of the present application;

[0055] Figure 5 is a determination flowchart of a recommended working mode provided by an embodiment of the present application;

[0056] Figure 6 is a schematic block diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0057] First, the professional terms involved in the embodiments of the present application are introduced.

[0058] Heating and ventilation system: a system for providing temperature regulation function, which can include one or more devices, such as a cooling pump, a cooling tower, a cooling machine, a refrigeration pump (primary pump), or a plate heat exchanger.

[0059] Cooling pump: a water pump that provides power for cooling water circulation.

[0060] Cooling tower: a device that evaporates and dissipates heat for cooling water through fan rotation.

[0061] Chiller pump: water pump that powers the chiller water circulation.

[0062] Chiller: refrigeration device that transfers chiller-side heat to cooling-side heat through compressed refrigerant.

[0063] Plate heat exchanger: refrigeration device that transfers chiller-side heat to cooling-side heat through heat exchange without consuming electricity.

[0064] Free cooling: mode that relies only on plate heat exchanger for heat exchange without using chiller.

[0065] Pre-cooling: mode that uses both chiller and plate heat exchanger for heat exchange.

[0066] Mechanical cooling: mode that uses chiller for heat exchange.

[0067] HVAC topology diagram: real-time running condition of HVAC equipment mapped to control interface in the background, which can be viewed by on-site personnel and remotely controlled.

[0068] Reference working data: data provided by equipment manufacturers and / or data determined based on empirical formula.

[0069] Real data: data obtained by sampling equipment.

[0070] Figure 1 is a device schematic diagram of a HVAC system provided by an embodiment of the present application. In a HVAC system parameter optimization scheme using a neural network model, the neural network model is trained based on relevant data of the HVAC system obtained by sampling, such as wet-bulb temperature, load, and internal parameters of the HVAC system, such as fan frequency and cooling pump frequency, so that the neural network model outputs predicted power consumption information of the HVAC system, and then the HVAC system is parameter optimized based on the predicted power consumption information. However, since the neural network model is relatively sensitive, and the relevant data of the HVAC system often has a problem of missing in a partial value range (which can also be referred to as a value space or a data space), which leads to difficulty in maintaining stable and reasonable output of the trained model in the complete value range, and the model has a serious overfitting problem, resulting in poor parameter optimization effect of the HVAC system. In addition, the neural network model also has a poor interpretability problem. The above problems lead to great resistance in the deployment of this optimization scheme.

[0071] In some other HVAC system parameter optimization schemes, in order to solve the problem of missing data in a partial value range, the data is supplemented by field experiments during model training. However, in this scheme, a large number of experiments will cause fault risks and additional consumption, and since the performance of each device is different, the transferability of this method is poor.

[0072] In view of this, the embodiment of the present application provides a device model establishment method, which can be applied to a device or a system that needs to model the relationship between the working parameters of the device, for example, can be applied to a heating and ventilation system. The device model of each device can be one or more. The input of the device model includes a first working parameter of the device, which can be one or more. The output of the device model is a predicted value of a second working parameter related to the first working parameter. The device model is used to represent the correlation between the working parameters of the device. Without relying on on-site experiments of the device, real data and reference working data of the device are used to form training data in a complete value range. At the same time, the reference working data is continuously corrected during the model training process to reduce the error of the reference working data. Therefore, the device model obtained by training can stably and accurately output results in the complete value range, and the effect of parameter optimization is improved.

[0073] The device model establishment method and the optimization method of the heating and ventilation system provided by the embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0074] Figure 2 FIG. 1 is a flowchart of a device model establishment method provided by the embodiment of the present application. In the following embodiments, the method is applied to a heating and ventilation system. As shown in FIG. 1, the device model establishment method includes the following steps. Figure 2

[0075] S201, obtaining first sampling data of the device in the working process and reference working data of the device. The first sampling data and the reference working data both include a first working parameter and a corresponding second working parameter.

[0076] ​The first sampling data is real data sampled during operation of the device, and the reference working data is data provided by the device manufacturer and / or data determined based on an empirical formula. Since the device is usually maintained in a relatively stable working state during operation, the first sampling data obtained by sampling the device may be missing in part of the value range, for example, the value range of the first working parameter of the device is between 0 and 100, but the first working parameter of the device is mostly maintained between 40 and 60 during operation. Thus, the first sampling data is likely to be missing data between 0 and 40 and between 60 and 100, while the reference working data can cover the complete value range, for example, the device manufacturer provides reference working data of the device in the complete value range, or the reference working data in the complete value range can be calculated based on an empirical formula. Thus, the reference working data can be used to make up for the missing problem of the first sampling data in part of the value range, for example, in the case of the foregoing example, the reference working data between 0 and 40 and between 60 and 100 can be used to supplement the first sampling data, so as to obtain data between 0 and 100 as training data.

[0077] Since the first sampling data is real data obtained from the heating system, it is more accurate than the reference working data. When obtaining the reference working data, if all the reference working data of the device in the value range of the first parameter is obtained, the reference working data can be partially deleted based on the first sampling data.

[0078] Optionally, the first sampling data of the device during operation and all the reference working data of the device in the value range of the first parameter are obtained; the value range of the first parameter is divided into a plurality of subintervals; for each subinterval, it is determined whether there is first sampling data in the subinterval, if there is first sampling data in the subinterval, the reference working data in the subinterval is deleted, and if there is no first sampling data in the subinterval, the reference working data in the subinterval is retained.

[0079] Through the above method, the reference working data can be used to supplement the first sampling data. The data obtained after the supplement can be referred to as fusion data, which fully retains the real data of the device and ensures that there is data in each subinterval in the entire value range.

[0080] The device model is a model between a first working parameter and a second working parameter of the device, and the first working parameter and the second working parameter can include temperature, flow, load, frequency of the device, power consumption and the like, wherein the load refers to heat generated by the device in the data center where the HVAC system is located. The specific data contained in the first working parameter and the second working parameter of each device model can be determined according to the actual situation of each device. In order to predict the power consumption of the HVAC system, for the chilled pump, the device model to be established is a model between the chilled water flow and the motor frequency of the chilled pump and a model between the motor frequency of the chilled pump and the power consumption; for the cooling pump, the device model to be established is a model between the cooling water flow and the motor frequency of the cooling pump and a model between the motor frequency of the cooling pump and the power consumption; for the cooling tower, the device model to be established is a model between the fan frequency of the cooling tower and the power consumption, and a model between the weather temperature of the environment where the HVAC system is located, the cooling side temperature difference (the temperature difference between the inlet and outlet cooling water), the water vapor ratio (LGR) and the cooling tower outlet water temperature (cooling water temperature); for the chiller, the device model to be established is a model between the chilled water flow and the evaporator saturation temperature, a model between the cooling water flow and the condenser saturation temperature, and a model between the evaporator saturation temperature, the condenser saturation temperature and the chiller load rate and the power consumption; for the plate heat exchanger, the device model to be established is a model between the chilled water flow, the cooling water flow, the cooling water temperature and the chilled water temperature. Based on these models, the first sampling data to be collected and the reference working data to be obtained for each device can be determined.

[0081] In S202, the initial device model corresponding to the device is iteratively trained by using the first sampling data and the reference working data, and the reference working data is corrected based on the model obtained through iteration in the iterative training process, so as to obtain the device model.

[0082] Optionally, the initial device model in the embodiment of the present application can use a black box algorithm such as random forest, eXtreme Gradient Boosting (XGBoost) and neural network, which can make the model have higher accuracy. Optionally, the initial device model can be a combination of one or more of a linear regression model, a polynomial regression model and an empirical model, and the empirical model can be an expert formula model, an expert parameter selection model and the like. For example, the preset model of the device is a linear regression + expert formula model; the preset model of the device is a polynomial regression + expert parameter selection model, and compared with the black box algorithm, the model has better interpretability. In actual application, the initial device model can be determined according to the type of the device.

[0083] Since there will be some deviation between the reference working data and the actual data of the device, when training the model using the first sampled data and the reference working data, the reference working data can be continuously corrected to gradually reduce the error of the reference working data. Optionally, in the embodiments of this application, a residual correction can be performed on the reference working data after each round of training iterations.

[0084] Optionally, the initial device model is trained using the first sampled data and the reference working data to obtain the first model. The first working parameters in the reference working data are input into the first model to obtain the predicted values ​​of the second working parameters output by the first model. The second working parameters in the reference working data are updated with the predicted values ​​of the second working parameters output by the first model. This step is repeated until the iteration stopping condition is met, and the obtained first model is determined as the device model.

[0085] The initial device model is trained using fused data consisting of first sampled data and reference working data, resulting in a first model after the first round of training. The model loss is determined for the first sampled data. Then, the first working parameters from the reference working data are input into the first model after the first round of training to obtain the predicted values ​​of the second working parameters output by the first model. The second working parameters from the reference working data are then updated to the predicted values ​​of the second working parameters output by the first model after the first round of training, thus achieving the first residual correction of the reference working data and obtaining updated reference working data. The first round of training is then performed using the first sampled data and the updated reference working data. The first model is trained to obtain the first model after the second round of training. The model loss is determined for the first sampled data. Then, the first working parameters in the reference working data are input into the first model after the second round of training to obtain the predicted value of the second working parameter output by the first model after the second round of training. The second working parameters in the reference working data are updated with the predicted value of the second working parameter output by the first model after the second round of training, thus realizing the second residual correction of the reference working data. The updated reference working data is obtained again. The above iterative process is repeated until the model loss meets the iteration stopping condition. The first model obtained at this time is used as the trained device model.

[0086] By training the initial equipment model using the above method on each device in the HVAC system, the corresponding equipment model can be established. The trained equipment model, along with the corresponding first sampling data and reference working data, can be stored to facilitate subsequent updates to the equipment model.

[0087] In the embodiments of the present application, the real data of the heating system and the reference working data are fused, the reference working data is used to supplement the real data to solve the problem of missing real data in some value range, and the error of the reference working data is continuously corrected to gradually reduce the error, so that the device model trained can stably and accurately output results in the complete value range, and the effect of parameter optimization is improved.

[0088] In addition, since the environment of the device during operation is dynamically changing, the error of the device model at different times can be different, and the error of the trained device model can be large after a period of time. Therefore, in the model establishing method of the embodiments of the present application, dynamic updating of the device model is further proposed.

[0089] Optionally, the second sampling data of the device in the working process in the first recent time period is obtained; for each sub-interval, it is judged whether the second sampling data exists in the sub-interval, if the second sampling data exists in the sub-interval, the first sampling data or the reference working data in the sub-interval is deleted, if the second sampling data does not exist in the sub-interval, the first sampling data or the reference working data in the sub-interval is retained, and the first data is obtained; the device model is trained by using the first data, and the reference working data in the first data is corrected based on the model obtained by each iteration in the iterative training process to update the device model.

[0090] The second sampling data is similar to the first sampling data, but the second sampling data is the real data obtained by the latest sampling, and is closer to the current time than the first sampling data. For example, in daily maintenance, the error of the device model can be determined based on the real-time real data of each device, if the error is greater than a preset error value, the model can be retrained, the second sampling data of the first recent time period is obtained, for example, the second sampling data of the last month, the second sampling data is fused with the fusion data after the last training, for example, the first sampling data and the reference working data after the last training, to obtain new fusion data, that is, the first data. For example, each device model can be retrained according to certain rules, for example, the device model is trained and updated monthly, and the second sampling data of each device in the heating system in the last month is obtained each time the training is performed, and the second sampling data is fused with the fusion data after the last training to obtain new fusion data. Each time the data is fused, the latest sampling data is used to replace the sampling data or the reference working data of the fusion data after the last training. The method of training the device model by using the new fusion data is similar to the foregoing, and the method of correcting the reference working data in the new fusion data is also similar to the foregoing.

[0091] Optionally, with reference to Figure 3As shown, after the first sampling data and the second sampling data are acquired, the first sampling data and the second sampling data can be subjected to abnormal data detection and screening according to the device models, and the screened first detection data and the second detection data are subjected to data fusion with the reference working data in the subsequent data fusion process.

[0092] In the embodiment of the present application, the method of data fusion and residual correction is used in model training, and adaptive model updating can be realized, which is more suitable for the real use environment of the data center and saves a lot of cost.

[0093] On the basis of the device model, the method for optimizing the HVAC system provided in the embodiment of the present application is described.

[0094] Figure 4 is a flowchart of a method for optimizing the HVAC system provided in the embodiment of the present application. As shown, Figure 4 the method comprises:

[0095] S401, acquiring device models of devices in the HVAC system, the device models being obtained by training in advance by using the method of the foregoing embodiment, and the device models including a model with the second working parameter being power consumption.

[0096] Optionally, after the device models of the devices are acquired, the error of the device models can be determined based on real-time real data of the devices, so that the error can be eliminated in the result of model prediction in the subsequent optimization process.

[0097] S402, connecting the device models of the devices according to the topological information of the HVAC system to obtain a system model of the HVAC system, and performing parameter optimization on the system model to obtain target parameters that make the power consumption of the HVAC system lowest.

[0098] The topological information of the HVAC system includes the connection relationship between the devices in the HVAC system, the water flow direction and other information. By using the topological information of the HVAC system, the device models of the devices can be connected to obtain a coupled model, i.e., the system model of the HVAC system. The topological information can also be displayed in a graphical manner, for example, the HVAC topology diagram can be viewed on the control platform of the HVAC system.

[0099] Based on the temperature information, the load information of the heating and ventilation system, the refrigerating capacity distribution information of the heating and ventilation system, and the upper and lower limits of parameters of each device in the heating and ventilation system, an optimization algorithm is used to optimize the system model. The temperature information can be the wet-bulb temperature, i.e. the weather temperature of the environment in which the heating and ventilation system is located, and the load information is the total heat generated by the devices in the data center. The heating and ventilation system can include one or more subsystems, and multiple subsystems can be partially or entirely in working condition. The refrigerating capacity distribution information of the heating and ventilation system refers to the refrigerating capacity distribution ratio of the subsystems in the heating and ventilation system, which can be determined based on real-time actual data (refrigerated water inlet and outlet temperature, flow rate, etc.) of the heating and ventilation system. For example, if two subsystems of the heating and ventilation system are working, the refrigerating capacity distribution ratio is 0.4:0.6, i.e. one subsystem bears 40% of the load and the other subsystem bears 60% of the load. If a single subsystem is working, the refrigerating capacity distribution ratio is 100%. The refrigerating capacity distribution ratio can be stored in a table, which is updated after each calculation, and can be read from the table during optimization.

[0100] The device models of each device are interrelated and can form a coupled system, i.e. the system model of the entire heating and ventilation system. For the system model, a set of parameters is assumed, including the cooling water temperature difference, the chilled water temperature difference, and the cooling tower frequency (cooling tower fan frequency). The cooling water flow rate can be determined based on the cooling water temperature difference and the load, and the chilled water flow rate can be determined based on the chilled water temperature difference and the load. For the case of multiple subsystems, the load of each subsystem is the product of the total load of the heating and ventilation system and the refrigerating capacity distribution ratio.

[0101] The power consumption of the cooling pump is determined based on the chilled water flow rate and two equipment models of the cooling pump: a model between the chilled water flow rate and the frequency and a model between the frequency and the power consumption. The power consumption of the primary pump is determined based on the chilled water flow rate and two equipment models of the primary pump: a model between the chilled water flow rate and the frequency and a model between the frequency and the power consumption. The power consumption of the cooling tower is determined based on an equipment model of the cooling tower: a model between the frequency and the power consumption. The LGR is determined based on the chilled water flow rate and the cooling tower fan frequency, and the cooling tower outlet water temperature is determined based on an equipment model of the cooling tower: a model between the wet bulb temperature, the chilled water temperature difference, the LGR, and the cooling tower outlet water temperature (the chilled water temperature). The chilled water temperature is determined based on the cooling tower outlet water temperature and an equipment model of the plate heat exchanger: a model between the chilled water flow rate, the chilled water flow rate, the chilled water temperature, and the chilled water temperature. The evaporator saturation temperature is determined based on the chilled water temperature, the chilled water flow rate, and an equipment model of the chiller: a model between the chilled water flow rate and the evaporator saturation temperature, and the condenser saturation temperature is determined based on the chilled water flow rate, the chilled water temperature, and an equipment model of the chiller: a model between the chilled water flow rate and the condenser saturation temperature. The power consumption of the chiller is determined based on an equipment model of the chiller: a model between the evaporator saturation temperature, the condenser saturation temperature, the chiller load rate, and the power consumption. The chiller load rate is the ratio of the load to the rated power of the chiller.

[0102] In the above method, the predicted power consumption of each device is obtained by substituting the parameters into the system model composed of each equipment model, that is, the predicted power consumption of the HVAC system. The grid search + gradient descent search method is used to repeatedly perform the above operation to optimize the parameters in the entire feasible parameter value range, so as to obtain the target parameters that make the power consumption of the HVAC system lowest. In addition, for each model, the corresponding intermediate variable can be output to improve the interpretability of the model.

[0103] Optionally, if the difference between the target parameters and the field parameters of the HVAC system is greater than or equal to a first preset threshold, the field parameters of the HVAC system are adjusted to the target parameters. Optionally, if the difference between the target parameters and the field parameters of the HVAC system is less than the first preset threshold, and the difference between the power consumption of the HVAC system corresponding to the target parameters and the current power consumption of the HVAC system is greater than or equal to a second preset threshold, the field parameters of the HVAC system are adjusted to the target parameters. Thus, while improving the parameter optimization effect, the frequent changes to the equipment parameters are reduced.

[0104] Optionally, the HVAC system has multiple working modes, and the devices in working state are different in different working modes, such as the free cooling mode, the pre-cooling mode and the mechanical cooling mode mentioned above, different working modes are suitable for different environments, and the corresponding power consumption is also different, each working mode has corresponding available conditions, for example, when the temperature is less than the first temperature, the free cooling mode, the pre-cooling mode and the mechanical cooling mode are all available, and when the temperature is greater than or equal to the first temperature, the pre-cooling mode and the mechanical cooling mode are available. In addition to parameter optimization, the method of the embodiment of the application can also optimize the working mode. The system model composed of the device model in different working modes is optimized to obtain the optimal working mode and the corresponding target parameter of each second time period in the current and future multiple second time periods, which makes the power consumption of the HVAC system lowest. Among them, the temperature information can include real-time temperature information and future temperature prediction information, and the load information can include real-time load information and future load prediction information. Based on the temperature prediction information and the load prediction information of the multiple second time periods, the optimal working mode of each second time period in the future multiple second time periods can be predicted. For example, for each second time period, the target parameter of the working mode with the lowest power consumption is determined for each working mode using the method described above, and then the multiple working modes are compared to determine the working mode with the lowest power consumption, which is the optimal working mode. For example, based on the real-time data of the HVAC device, the current optimal working mode is determined, and based on the temperature prediction information and the load prediction information of the future n hours, the optimal working mode of each hour in the future n hours is determined.

[0105] Optionally, it is determined whether the current working mode of the HVAC system is the optimal working mode; if the current working mode is the optimal working mode and the current working mode is available in one or more second time periods in the future, the current working mode is output as the recommended working mode; if the current working mode is not the optimal working mode and the current mode is available in multiple second time periods in the future, if the current working mode is the optimal working mode in any second time period in the future, the current working mode is output as the recommended working mode; if the current working mode is not the optimal working mode and the current mode is not available in any second time period in the future or the current working mode is not the optimal working mode in any second time period in the future, the working mode with the lowest overall power consumption in the multiple second time periods in the future is output as the recommended working mode. Optionally, if the recommended working mode is different from the current working mode, the working mode of the HVAC system is set to the recommended working mode, and the parameters of the HVAC system are set to the target parameters corresponding to the recommended working mode. The working mode available means that the device can work normally in the working mode, and the available working modes in different environmental conditions can be set in advance.

[0106] For example, as Figure 5As shown, it is determined whether the current working mode of the HVAC system is the optimal working mode; if the current working mode is the optimal working mode and the current working mode is available in the next hour, the current working mode is output as the recommended working mode; if the current working mode is not the optimal working mode and the current mode is available in each of the next n hours, if the current working mode is the optimal working mode in any of the next n hours, the current working mode is output as the recommended working mode; if the current working mode is not the optimal working mode and the current mode is not available in any of the next n hours or the current working mode is not the optimal working mode in each of the next n hours, the working mode with the lowest overall power consumption in the next n hours is output as the recommended working mode. In actual application, the frequency of parameter optimization and working mode optimization can be set as needed, for example, parameter optimization is triggered every 5 minutes and working mode optimization is triggered every hour.

[0107] Optionally, in the embodiments of the present application, the determined target parameters, recommended working mode, predicted power consumption and the like can be uploaded to the control platform of the HVAC system, and the control platform can issue instructions to the HVAC system for optimization. The control platform can also draw curves according to the real-time running conditions of the HVAC system and the predicted power consumption, so as to enable maintenance personnel to intuitively observe the parameter optimization effect.

[0108] The method of the embodiments of the present application realizes online automatic operation of device model training, working mode recommendation, parameter recommendation, platform information pushing and display of the HVAC system, does not require manual intervention, can realize 24-hour uninterrupted pushing, can automatically control the working state of the field, greatly saves manpower, has good scalability, can realize low-cost scheme migration, and is conducive to wide application in various data centers.

[0109] Figure 6 is a schematic block diagram of an electronic device provided by the embodiments of the present application. As Figure 6 shown, the electronic device 600 can include at least one processor 610 for implementing the device model establishment method of the HVAC system or the optimization method of the HVAC system provided by the embodiments of the present application.

[0110] Optionally, the electronic device 600 further includes at least one memory 620 for storing program instructions and / or data. The memory 620 and the processor 610 are coupled. The coupling in the embodiments of the present application is indirect coupling or communication connection between devices, units or modules, which can be electrical, mechanical or other forms, for information interaction between devices, units or modules. The processor 610 can operate cooperatively with the memory 620. The processor 610 can execute the program instructions stored in the memory 620. At least one of the at least one memory can be included in the processor.

[0111] Optionally, the electronic device 600 further includes a communication interface 630 for communicating with other devices via a transmission medium, thereby enabling the electronic device 600 to communicate with other devices. The communication interface 630 may be, for example, a transceiver, interface, bus, circuit, or a device capable of transmitting and receiving functions. The processor 610 can utilize the communication interface 630 to transmit and receive data and / or information, and to implement the device model establishment method or HVAC system optimization method provided in the embodiments of this application.

[0112] This application embodiment does not limit the specific connection medium between the processor 610, memory 620, and communication interface 630. This application embodiment... Figure 6 The processor 610, memory 620, and communication interface 630 are connected via bus 640. Bus 640 is... Figure 6 The connections between other components are shown in thick lines only and are not intended to be limiting. This bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, Figure 6 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0113] It should be understood that the processor in the embodiments of this application can be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method embodiments can be completed by the integrated logic circuitry in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method.

[0114] It should also be appreciated that the memory in the embodiments of the present application can be a volatile memory or a nonvolatile memory, or can include both volatile and nonvolatile memory. Among them, the nonvolatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example, and not limitation, many forms of RAM can be used, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct Rambus RAM (DR RAM). It should be noted that the memory of the system and method described herein is intended to include, but not be limited to, these and any other suitable types of memory.

[0115] The present application also provides a computer readable storage medium, which stores a computer program (also referred to as code or instructions). When the computer program is executed, it makes the computer execute the method in the foregoing embodiments.

[0116] The terms "unit", "module" and the like used in the specification can be used to represent a computer-related entity, hardware, firmware, a combination of hardware and software, software, or software in execution.

[0117] Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. The choice of hardware or software, or combinations of both, would be dependent on the specific application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application. In several embodiments provided in the present application, it will be apparent that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the division of the units described above is merely illustrative, and for example, the division of the units is merely a logical function division, and actual implementation can have another division, for example, a plurality of units or components can be combined or integrated into another system, or some features can be omitted or not implemented. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0118] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. can be located in one place or can be distributed to a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0119] In addition, the functional units in each of the embodiments of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit.

[0120] In the above embodiments, the functions of the various functional units can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, the whole or part of the computer program instructions (program) can be implemented in the form of a computer program product. When the computer program instructions (program) are loaded and executed on a computer, the whole or part of the flow or function according to the embodiments of the present application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL) or wireless (such as infrared, wireless, microwave, etc.)) way. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media. The available media can be magnetic media (for example, floppy disk, hard disk, magnetic tape), optical media (for example, digital video disc (DVD)), or semiconductor media (for example, solid state disk (SSD)) and the like.

[0121] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part of the prior art that contributes to the technical solutions or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk and various media that can store program codes.

[0122] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for establishing a device model, characterized in that, The input to the device model includes a first operating parameter of the device, and the output of the device model is a predicted value of a second operating parameter related to the first operating parameter. The method includes: Acquire first sampling data of the device during operation, and all reference operating data of the device within the value range of the first operating parameter; wherein, the first sampling data is real data sampled during the operation of the device, and the reference operating data is data provided by the device manufacturer and / or data determined based on empirical formulas, and both the first sampling data and the reference operating data include the first operating parameter and the corresponding second operating parameter; The range of values ​​for the first working parameter is divided into multiple sub-intervals; For each sub-interval, if the first sampled data exists in the sub-interval, the reference working data in the sub-interval is deleted; if the first sampled data does not exist in the sub-interval, the reference working data in the sub-interval is retained. The initial device model corresponding to the device is iteratively trained using the first sampled data and the reference working data, and the reference working data is corrected based on the model obtained during the iterative training process to obtain the device model.

2. The method according to claim 1, characterized in that, The step of iteratively training an initial device model corresponding to the device using the first sampled data and the reference working data, and correcting the reference working data based on the iteratively obtained model during the iterative training process to obtain the device model, includes: The initial device model is trained using the first sampled data and the reference working data to obtain a first model. The first working parameter in the reference working data is input into the first model to obtain the predicted value of the second working parameter output by the first model. The second working parameter in the reference working data is updated with the predicted value of the second working parameter output by the first model. This step is repeated until the iteration stopping condition is met, and the obtained first model is determined as the device model.

3. The method according to claim 1, characterized in that, Also includes: Acquire second sampling data of the device during its most recent first time period of operation; For each sub-interval, determine whether there is second sampled data in the sub-interval. If there is second sampled data in the sub-interval, delete the first sampled data or reference working data in the sub-interval. If there is no second sampled data in the sub-interval, retain the first sampled data or reference working data in the sub-interval to obtain the first data. The device model is trained using the first data, and during the iterative training process, the reference working data in the first data is corrected based on the model obtained through iteration, so as to update the device model.

4. The method according to any one of claims 1-3, characterized in that, The preset model of the device is one or a combination of linear regression model, multinomial regression model and empirical model.

5. The method according to any one of claims 1-3, characterized in that, The method is applied to a heating, ventilation, and air conditioning (HVAC) system, and the device is a device within the HVAC system. The first and second operating parameters corresponding to the device and the device model include any one of the following: The device is a cooling pump, the first operating parameter is the cooling water flow rate, and the second operating parameter is the motor frequency of the cooling pump. The device is a cooling pump, the first operating parameter is the motor frequency of the cooling pump, and the second operating parameter is the power consumption; The device is a chilled water pump, the first operating parameter is the chilled water flow rate, and the second operating parameter is the motor frequency of the chilled water pump; The device is a refrigeration pump, the first operating parameter is the motor frequency of the refrigeration pump, and the second operating parameter is the power consumption; The device is a cooling tower, the first operating parameter is the fan frequency of the cooling tower, and the second operating parameter is the power consumption; The device is a cooling tower. The first operating parameter is the ambient temperature of the environment where the HVAC system is located, the temperature difference between the inlet and outlet cooling water, and the water-vapor ratio. The second operating parameter is the outlet water temperature of the cooling tower. The equipment is a chiller, the first operating parameter is the chilled water flow rate, and the second operating parameter is the evaporator saturation temperature; The device is a chiller, the first operating parameter is the cooling water flow rate, and the second operating parameter is the condenser saturation temperature; The device is a chiller, the first operating parameter is the evaporator saturation temperature, the condenser saturation temperature and the chiller load rate, and the second operating parameter is the power consumption. The device is a plate heat exchanger, and the first operating parameters are chilled water flow rate, cooling water flow rate, and cooling water temperature, while the second operating parameter is chilled water temperature.

6. A method for optimizing a heating, ventilation, and air conditioning system, characterized in that, include: Obtain the equipment model of each device in the HVAC system. The equipment model is obtained in advance using the method described in any one of claims 1-4. The equipment model includes a model whose second operating parameter is power consumption. The device models of each device are connected according to the topology information of the HVAC system to obtain the system model of the HVAC system. The system model is then optimized to obtain the target parameters that minimize the power consumption of the HVAC system.

7. The method according to claim 6, characterized in that, Also includes: If the difference between the target parameter and the field parameters of the HVAC system is greater than or equal to a first preset threshold, then the field parameters of the HVAC system are adjusted to the target parameter.

8. The method according to claim 7, characterized in that, Also includes: If the difference between the target parameter and the field parameter of the HVAC system is less than the first preset threshold, and the difference between the power consumption of the HVAC system corresponding to the target parameter and the current power consumption of the HVAC system is greater than or equal to the second preset threshold, then the field parameter of the HVAC system will be adjusted to the target parameter.

9. The method according to claim 6, characterized in that, The HVAC system has multiple operating modes, and the equipment in different operating modes is in different states; the method further includes: The system model composed of equipment models under different working modes is optimized by parameter search to obtain the optimal working mode and corresponding target parameters that minimize the power consumption of the HVAC system for the current and multiple future second time periods.

10. The method according to claim 9, characterized in that, Each working mode has corresponding available conditions, and the method further includes: Determine whether the current operating mode of the HVAC system is the optimal operating mode; If the current working mode is the optimal working mode, and the current working mode is available in one or more future second time periods, then the current working mode will be output as the recommended working mode. If the current working mode is not the optimal working mode, and the current working mode is available in multiple future second time periods, then if the current working mode is the optimal working mode in any future second time period, the current working mode will be output as the recommended working mode. If the current operating mode is not the optimal operating mode, and the current operating mode is unavailable in any future second time period, or the current operating mode is not the optimal operating mode in any future second time period, then the operating mode with the lowest overall power consumption in multiple future second time periods will be output as the recommended operating mode.

11. The method according to claim 10, characterized in that, Also includes: If the recommended operating mode is different from the current operating mode, then the operating mode of the HVAC system is set to the recommended operating mode.

12. An electronic device, characterized in that, include: Memory and processor; The memory is used to store computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the method described in any one of claims 1-11.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to perform the method as described in any one of claims 1-11.

Images