CN108317670B - Refrigeration system energy-saving control method and system based on machine learning - Google Patents
Refrigeration system energy-saving control method and system based on machine learning Download PDFInfo
- Publication number
- CN108317670B CN108317670B CN201810174214.9A CN201810174214A CN108317670B CN 108317670 B CN108317670 B CN 108317670B CN 201810174214 A CN201810174214 A CN 201810174214A CN 108317670 B CN108317670 B CN 108317670B
- Authority
- CN
- China
- Prior art keywords
- time point
- training sample
- data processing
- module
- processing module
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Air Conditioning Control Device (AREA)
Abstract
The invention discloses a refrigeration system energy-saving control method and system based on machine learning, which can continuously improve the accuracy and real-time performance of prediction and control, run at optimal efficiency on the premise of ensuring to meet related set cooling targets, and achieve the aim of reducing the energy consumption of the system to the maximum extent. The control method comprises the following steps: s1, initializing operation condition parameters; s2, initializing a training sample set; s3, establishing a machine learning load prediction model; s4, predicting the load at the next time point; s5, finding out the training sample most similar to the test sample at the current time point from the training sample set, and adjusting the adjusting speed of the refrigeration system in the subsequent time period according to the training sample set; s6, judging whether the actual control parameter at the next time point reaches the set control parameter requirement and whether the actual load value is equal to the load predicted value, and correcting or deleting the related training samples in the training sample set according to the actual control parameter and the actual load value; and S7, returning to the step S3 or S4, and entering the next iteration loop. The invention can be widely applied to the field of refrigeration.
Description
Technical Field
The invention relates to a refrigeration system energy-saving control method; the invention also relates to an energy-saving control system of the refrigerating system.
Background
The refrigerating system is widely used for providing cold energy for central air-conditioning and mass cold production processes, and mainly comprises a refrigerating water pump, a refrigerating main machine, cooling equipment (such as a water-cooled refrigerating system comprising a cooling water pump and a cooling tower) and the like. The energy consumption of the refrigeration system is very large, for example, the central air conditioner consists of the refrigeration system and a tail end system, and the energy consumption of the refrigeration system can account for more than 60% of the total energy consumption of the central air conditioner, so that the load prediction and the energy-saving control of the refrigeration system are significant. However, the refrigeration system has the characteristics of strong dynamics such as multivariable, strong coupling, nonlinearity, time-varying property, time lag, and many interference factors, which makes it difficult to perform accurate and real-time load prediction and energy-saving control.
The commonly used energy-saving control method of the refrigeration system mainly takes the variable frequency regulation of a water pump and a fan as a core, and generally carries out smooth regulation on the running frequency of the water pump, the fan and the like according to the deviation between the measured actual cooling effect and the correspondingly set cooling target, so as to regulate the power of the water pump, the fan and a refrigerator, thereby effectively reducing the energy consumption and improving the comfort level. For central air conditioning, these parameters are typically outdoor temperature, humidity and indoor temperature, humidity, and for production chiller systems these parameters are typically the temperature of the cold spot. However, the control method still has some problems, taking a central air conditioner as an example: (1) the central air conditioner has the characteristics of multivariable, strong nonlinearity, strong coupling and the like, so that the cooling capacity cannot be obtained through accurate calculation, and the specific control logic of the control system is seriously dependent on the experience of designers; therefore, the accuracy of the control system is difficult to be ensured, and whether the system achieves the optimal energy-saving effect cannot be judged; (2) the central air conditioner has the characteristic of large time lag, the system needs longer time to reach the set cold output, and the control system is mainly adjusted based on the influence parameters and experience of the current time point; when the set cold output is reached, the influence parameters of the system are likely to have large changes, so that the actual cold consumption (namely the cold actually required to be provided by the central air conditioner) is also likely to have changes, which is not necessarily equal to the actual cold supply of the system at the moment; that is, the control system does not have accurate and real-time load prediction and load matching capabilities.
At present, machine learning techniques (such as support vector machines, artificial neural networks, etc.) have been tried to be used for load prediction of the central air conditioner, and these techniques have unique advantages in handling the strong dynamic characteristics of the central air conditioner. Support vector machines and artificial neural networks are two common machine learning discrimination methods, and are generally used for pattern recognition, classification, and regression analysis. Taking the example of adopting a support vector machine to predict the load of a central air-conditioning refrigeration system, based on the provided running sample set, the support vector machine uses part of samples as training samples for modeling, and then uses the rest samples as test samples for prediction according to the model. The comparison between the predicted value and the actual output value of the test sample shows that the load prediction of the central air conditioner by using the support vector machine is more accurate, so that the modeling difficulty caused by the strong dynamic characteristic of the central air conditioner is effectively solved. However, the existing support vector machine prediction method is established on the basis of known sample data, the existing operation control method does not solve the problem of time lag, the sample output value is the collected actual cooling capacity, but not necessarily the real cooling capacity, namely the sample data for modeling is not accurate, so the existing support vector machine prediction method cannot really and accurately predict the cooling load, and cannot accurately and real-timely guide the central air-conditioning control system to carry out optimal energy-saving control. Similarly, other machine learning prediction methods (such as artificial neural networks) also require an accurate operation sample set for modeling, so accurate and real-time central air conditioning load prediction and optimal energy saving control cannot be really performed.
Additionally, machine learning techniques have been less studied to date for predicting and controlling the refrigeration systems used by the process.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a refrigeration system energy-saving control method based on machine learning. The control method can continuously improve the accuracy and real-time performance of prediction and control, and operates at optimal efficiency on the premise of ensuring that the related set cooling target is met, thereby achieving the purpose of reducing the energy consumption of the system to the maximum extent.
In addition, the invention also provides a refrigerating system energy-saving control system based on machine learning.
The technical scheme adopted by the refrigeration system energy-saving control method based on machine learning is as follows: the refrigeration system energy-saving control method based on machine learning utilizes a refrigeration system energy-saving control system based on machine learning to carry out energy-saving control, the refrigeration system energy-saving control system based on machine learning comprises a data acquisition module, a data processing module and an operation control module, and the method comprises the following steps:
s1: initializing operation condition parameters: the data acquisition module acquires the operating condition parameters of the refrigeration system and the cold using system at each time point according to a certain time step within a certain time range and sends the operating condition parameters to the data processing module;
s2: initializing a training sample set: the data processing module sets a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the proximity degree of output values of the two samples, and other parameters for data processing and calculation, the data processing module forms training samples of each time point according to a certain sample forming rule by using the operating condition parameters of each time point, and the training samples of each time point form an initial training sample set;
s3: establishing a machine learning load prediction model: the data processing module establishes a load prediction model by using machine learning according to the training sample set;
s4: predicting the load at the next time point: the data acquisition module acquires the operating condition parameters of the current time point, and the data processing module forms the input vector of the test sample of the current time point according to the rule of the step S2, substitutes the input vector into the load prediction model and calculates the load prediction value of the next time point;
s5: adjusting the adjusting speed of the refrigerating system: the data processing module compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training sample meeting the requirement exists, the training sample forms a similar training sample set, and finds out the sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point as the most similar training sample;
when the most similar training sample exists, calculating the difference between the load predicted value of the next time point and the actual load value of the next time point recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample;
when the actual load difference of the most similar sample is larger than the approach deviation limit, the data processing module performs acceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module performs deceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the current time point is the next time point, the operation control module does not adjust the adjusting speed of the refrigerating system in the time period between the current time point and the next time point;
s6: correcting a training sample set: at the next time point, the data processing module takes the acquired actual load value at the next time point as the output value of the test sample at the current time point to form a new training sample at the current time point, adds the new training sample into the training sample set, and judges whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module is equal to the set corresponding cooling target and whether the actual load value at the next time point is equal to the load predicted value at the next time point in the step S4;
when the judgment is not true, the data processing module calculates the load prediction deviation at the next time point, adds the load prediction deviation to the output value of the new training sample to be used as the new output value of the new training sample, and performs corresponding processing according to the condition whether the similar training sample set exists in the step S5;
when a similar training sample set exists, the data processing module calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and the absolute value is called as a similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set;
s7: and (3) iterative loop: and the data processing module sets the next time point as the current time point, judges whether the system finishes operating a set time period and calls the set time period as a model correction cycle, and when the system finishes operating the model correction cycle, the data processing module goes to execute step S4, and when the system finishes operating the model correction cycle, the data processing module goes to execute step S3 to continue operating and controlling the refrigeration system.
Preferably, in step S1, the certain time range is a complete cooling operation period.
Preferably, in step S2, the training samples at each time point follow the following sample formation rule: the training sample consists of an input vector and an output value, each element of the input vector of the training sample at the time point is an operation condition parameter which is acquired at the time point and arranged according to a specified sequence, and the output value of the training sample at the time point is a refrigeration load value acquired at the next time point; in step S3, the input of the load prediction model is the input vector, and the output is the output value.
Preferably, in step S2, the degree of similarity of the input vectors of the two samples is determined by using the 2-norm of the difference between the input vectors of the two samples, i.e., the square root of the sum of the squares of all the corresponding element differences of the input vectors of the two samples.
Preferably, the machine learning is a support vector machine or an artificial neural network.
Preferably, in step S3, the machine learning is a support vector machine, the load prediction model is an epsilon-SVR model, and the kernel function used by the epsilon-SVR model is a radial basis kernel function.
Or, the machine learning is an artificial neural network, the load prediction model adopts a BP network model, and the kernel function used by the BP network model is a radial basis kernel function.
Preferably, in step S5, the acceleration adjustment is to increase the frequency adjustment speed of the frequency converters of the water pump and the fan of the refrigeration system, and the deceleration adjustment is to decrease the frequency adjustment speed of the frequency converters of the water pump and the fan of the refrigeration system.
Preferably, in step S7, one of the model correction periods is one day.
The technical scheme adopted by the refrigeration system energy-saving control system based on machine learning is as follows: the refrigeration system energy-saving control system based on machine learning comprises a data acquisition module, a data processing module and an operation control module;
the data acquisition module is connected with the data processing module and used for initializing operating condition parameters, acquiring the operating condition parameters of the refrigeration system and the cold using system at each time point according to a certain time step within a certain time range and sending the operating condition parameters to the data processing module;
the data processing module is connected with the data acquisition module and used for initializing a training sample set, setting a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the proximity degree of output values of the two samples and other parameters for data processing and calculation, forming the operating condition parameters of each time point into the training samples of the time point by the data processing module, and forming the training samples of each time point into an initial training sample set;
the data processing module is also used for establishing a machine learning load prediction model, and the machine learning is used for establishing the load prediction model according to the training sample set;
the data processing module is also used for predicting the load at the next time point, the data acquisition module acquires the operating condition parameters at the current time point, and the data processing module forms the input vector of the test sample at the current time point according to a certain sample forming rule and substitutes the input vector into the load prediction model to calculate the load prediction value at the next time point;
the operation control module is connected with the data processing module and used for adjusting the adjusting speed of the refrigerating system, the data processing module compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training sample meeting the requirement exists, the training samples form a similar training sample set, and the sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point is found out to be used as the most similar training sample;
when the most similar training sample exists, calculating the difference between the load predicted value of the next time point and the actual load value of the next time point recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample;
when the actual load difference of the most similar sample is larger than the approach deviation limit, the data processing module performs acceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module performs deceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the current time point is the next time point, the operation control module does not adjust the adjusting speed of the refrigerating system in the time period between the current time point and the next time point;
the data processing module is further used for correcting the training sample set, when the next time point is reached, the data processing module takes the acquired actual load value of the next time point as the output value of the test sample of the current time point to form a new training sample of the current time point, adds the new training sample into the training sample set, and judges whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module is equal to the set corresponding cooling target or not and whether the actual load value at the next time point is equal to the load predicted value at the next time point or not;
when the judgment is not true, the data processing module calculates the load prediction deviation at the next time point, adds the load prediction deviation to the output value of the new training sample to be used as the new output value of the new training sample, and performs corresponding processing according to the condition whether the similar training sample set exists in the step S5;
when a similar training sample set exists, the data processing module calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and the absolute value is called as a similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set;
the data processing module is also used for iterative cycle, the data processing module sets the next time point as the current time point, judges whether the system finishes operating a set time period and calls the set time period as a model correction period, when the system finishes operating a model correction period, the control of re-correcting or deleting the training sample is executed, when the system finishes operating a model correction period, the control of re-establishing the machine learning load prediction model is executed, the next iterative cycle is entered, and the operation control of the refrigeration system is continued.
The data acquisition module comprises a data sensing module and a data transformation module, wherein the data sensing module and the data transformation module are installed on an outdoor and/or indoor refrigerating system and a cold system, the data sensing module is used for acquiring the influence parameters influencing the load of the outdoor and/or indoor refrigerating system and the working condition parameters of the running state parameters of the refrigerating system and the cold system, and the data transformation module is used for converting the working condition parameters acquired by the data sensing module into data signals which can be identified by the data processing module and sending the data signals to the data processing module.
The data processing module comprises a data transmission module, a data calculation module, a database and a data processing and displaying module, wherein the data transmission module is used for receiving a data signal sent by the data acquisition module and sending an adjusting speed adjusting parameter in a time period between a current time point and a next time point to the operation control module, the data calculation module is used for establishing and calculating correction data, samples, models, load prediction and adjusting speed adjusting parameters, the database is used for recording data such as sample data and calculating parameters, and the data processing and displaying module is used for visually displaying and operating the data processing module.
Preferably, the data computing module and the database are located in a remote server as a cloud computing end.
The operation control module comprises a control execution module and an operation control display module, the control execution module is used for executing the control system, and the operation control display module is used for inputting and setting a cooling target, operating the control system and displaying control information.
Preferably, the data processing display module and the operation control display module are implemented on the same hardware and software.
The refrigerating system comprises a water pump, a fan and a frequency converter, when the operation control module adjusts the adjusting speed of the refrigerating system, the acceleration adjustment is to accelerate the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system, and the deceleration adjustment is to slow the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system.
Preferably, the machine learning is a support vector machine or an artificial neural network.
The invention has the beneficial effects that: the invention comprehensively considers the advantages of simplicity, maturity and energy-saving effect of the traditional refrigeration system control and the advantages of machine learning load prediction that the load prediction can be carried out without considering various strong dynamic characteristics of the refrigeration system and the prediction result is quite accurate as long as the training data is accurate, thereby overcoming the problems that the traditional refrigeration system control can not achieve the maximum energy saving and the training sample is inaccurate in the traditional machine learning prediction method and meeting the accuracy and real-time requirements of prediction and control; the control method collects the operating condition parameters of the refrigeration system and the cold utilization system to obtain a training sample set, machine learning models the obtained training sample set and carries out load prediction, the control system carries out control mode adjustment according to a prediction result, new training samples are added or the sample set and the model are corrected according to the adjustment result, the energy-saving control system automatically obtains the training sample set for modeling and carries out machine learning modeling, and then a regression function is corrected, so that the accuracy of the training samples and the load prediction model is continuously improved, the control system is guided to carry out adjustment of the adjustment mode, manual intervention is not needed, interference and deviation possibly brought by manual intervention are reduced, and an optimal adjustment method is provided; by repeatedly iterating the modeling, predicting, controlling, adjusting and model correcting process, the accuracy and real-time performance of prediction and control can be continuously improved, and on the premise of ensuring that the related set cooling target such as indoor set temperature for a central air conditioner, process cooling temperature for a cold production process and the like is met, the basic requirements such as comfort level are ensured, and simultaneously, the operation is carried out with optimal efficiency, the purpose of reducing the energy consumption of the system to the maximum extent is achieved, and the optimal energy-saving effect is achieved.
Drawings
FIG. 1 is a schematic flow chart diagram illustrating a method for energy conservation control of a refrigerant system based on machine learning according to an embodiment of the present invention;
FIG. 2 is a detailed flow chart of a refrigeration system energy saving control method based on machine learning according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a refrigeration system energy-saving control system based on machine learning according to an embodiment of the present invention.
Detailed Description
The first embodiment is as follows:
as shown in fig. 1 to fig. 3, the energy saving control method for a refrigeration system based on machine learning of the present embodiment discloses an energy saving control method for a refrigeration system based on a support vector machine for a central air conditioner, the method performs energy saving control by using a refrigeration system energy saving control system based on machine learning (support vector machine), the refrigeration system energy saving control system based on machine learning (support vector machine) includes a data acquisition module 1, a data processing module 2 and an operation control module 3, that is, the machine learning is a support vector machine, and the method includes the following steps:
s1: initializing operation condition parameters: the data acquisition module acquires the operating condition parameters of the refrigeration system and the cold utilization system at various time points according to a certain time step within a certain time range and sends the operating condition parameters to the data processing module, namely step 101 in fig. 3 (hereinafter, only the step code in fig. 3 is written); the cold utilization system is a tail end system of the central air conditioner, and the operation working condition parameters include but are not limited to outdoor temperature, humidity and solar radiation intensity, indoor temperature and humidity, input power of a refrigerator, chilled water flow, chilled water inlet and outlet water temperature, chilled water pump operation power, chilled water pump operation frequency, cooling water flow, cooling water inlet and outlet water temperature, cooling water pump operation power, cooling water pump operation frequency and cooling tower power; the certain time range is a complete cooling operation period;
s2: initializing a training sample set: the data processing module sets a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the similarity degree of output values of the two samples, and other parameters for data processing and calculation, namely 102, the data processing module forms training samples of each time point according to a certain sample forming rule by using the operating condition parameters of each time point, and the training samples of each time point form an initial training sample set; the training samples at each time point obeyed the following sample formation rules: the training sample consists of an input vector and an output value, each element of the input vector of the training sample at the time point is an operation condition parameter which is acquired at the time point and is arranged according to a specified sequence, the output value of the training sample at the time point is a refrigeration load value acquired at the next time point, the similarity degree of the input vectors of the two samples uses a 2-norm of the difference of the input vectors of the two samples, namely the square root of the sum of squares of all the corresponding element differences of the input vectors of the two samples, the smaller the 2-norm of the difference of the input vectors of the two samples is, the larger the similarity degree of the input vectors of the two samples is, the smallest 2-norm of the difference of the input vectors of the two samples is, and the largest similarity degree of the input vectors of the two samples is indicated;
s3: establishing a load prediction model of a support vector machine: the data processing module establishes a load prediction model, namely 201, by using a support vector machine according to a training sample set, wherein the input of the load prediction model is the input vector in the step S2, and the output is the output value; the load prediction model adopts an epsilon-SVR model, and the kernel function used by the epsilon-SVR model is a radial basis kernel function;
s4: predicting the load at the next time point: the data acquisition module acquires the operating condition parameters at the current time point, namely 202, and the data processing module forms the input vector of the test sample at the current time point according to the rule of the step S2, substitutes the input vector into the load prediction model, and calculates the load prediction value at the next time point, namely 203;
s5: adjusting the adjusting speed of the refrigerating system: the data processing module compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training sample meeting the requirement exists, the training samples form a similar training sample set, and finds out a sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point as the most similar training sample, namely 301;
when the most similar training sample exists, calculating the difference between the load predicted value at the next time point and the actual load value at the next time point recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample, namely 302;
when the actual load difference of the most similar sample is larger than the approach deviation limit, the data processing module performs acceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample; the acceleration adjustment is to accelerate the frequency regulation speed of frequency converters of a water pump and a fan of the refrigeration system, namely 303;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module performs deceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample; the speed reduction adjustment is to reduce the frequency regulation speed of frequency converters of a water pump and a fan of the refrigeration system, namely 304;
when the other situation is the case, the operation control module does not adjust the adjustment speed of the refrigeration system in the time period between the current time point and the next time point, that is, 305;
s6: correcting a training sample set: at the next time point, the data processing module takes the acquired actual load value at the next time point as the output value of the test sample at the current time point to form a new training sample at the current time point, and adds the new training sample to a training sample set, namely 401, and judges whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module is equal to the set corresponding cooling target, and whether the actual load value at the next time point is equal to the predicted load value at the next time point in the step S4, wherein the cooling target is the indoor temperature;
when the judgment is not true, the data processing module calculates the load prediction deviation at the next time point, adds the load prediction deviation to the output value of the new training sample to be used as the new output value of the new training sample, and performs corresponding processing according to the condition whether the similar training sample set exists in the step S5, namely 402;
when a similar training sample set exists, the data processing module calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and refers to the similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set, namely 403;
s7: and (3) iterative loop: the data processing module sets the next time point as the current time point, judges whether the system finishes operating a set time period and calls the set time period as a model correction cycle, and when the system finishes operating the model correction cycle, the data processing module goes to execute step S4 (501), and when the system finishes operating the model correction cycle, the data processing module goes to execute step S3 (502), and continues to control the operation of the refrigeration system; one such model modification period is taken to be one day.
The refrigeration system energy-saving control system based on machine learning is a refrigeration system energy-saving control system based on a support vector machine and used for a central air conditioner, and comprises a data acquisition module 1, a data processing module 2 and an operation control module 3, wherein the refrigeration system comprises a water pump, a fan and a frequency converter;
the data acquisition module 1 is connected with the data processing module 2 and is used for initializing operating condition parameters, acquiring the operating condition parameters of the refrigeration system and the cold utilization system at each time point according to a certain time step within a certain time range and sending the operating condition parameters to the data processing module 2; the data acquisition module 1 comprises a data sensing module 11 and a data conversion module 12 which are arranged on an outdoor and/or indoor refrigerating system and a cold using system, wherein the data sensing module 11 is used for acquiring outdoor and/or indoor influencing parameters influencing the load of the refrigerating system and working condition parameters of running state parameters of the refrigerating system and the cold using system, and the data conversion module 12 is used for converting the working condition parameters acquired by the data sensing module 11 into data signals which can be identified by the data processing module 2 and sending the data signals to the data processing module 2;
the data processing module 2 is connected with the data acquisition module 1 and is used for initializing a training sample set, setting a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the proximity degree of output values of the two samples, and other parameters for data processing and calculation, forming the training sample of each time point by the operation working condition parameter of each time point according to a certain sample forming rule by the data processing module, and forming an initial training sample set by the training sample of each time point; the data processing module 2 comprises a data transmission module 21, a data calculation module 22, a database 23 and a data processing display module 24, wherein the data transmission module 21 is configured to receive a data signal sent by the data acquisition module 1 and send an adjustment speed adjustment parameter in a time period between a current time point and a next time point to the operation control module 3, the data calculation module 22 is configured to establish and calculate correction data, a sample, a model, load prediction, and an adjustment speed adjustment parameter, the database 23 is configured to record data such as sample data and calculation parameters, the data processing display module 24 is configured to perform visual display and operation on the data processing module 2, and the data calculation module 22 and the database 23 are located in a remote server as a cloud computing end;
the data processing module 2 is further configured to establish a support vector machine load prediction model, and establish the load prediction model using the support vector machine according to a training sample set, where an input of the load prediction model is the input vector and an output is the output value;
the data processing module 2 is further used for predicting the load at the next time point, the data acquisition module 1 acquires the operating condition parameters at the current time point, and the data processing module 2 forms the input vector of the test sample at the current time point according to a sample forming rule and substitutes the input vector into the load prediction model to calculate the load prediction value at the next time point;
the operation control module 3 is connected with the data processing module 2 and is used for adjusting the adjusting speed of the refrigeration system, the operation control module 3 comprises a control execution module 31 and an operation control display module 32, the control execution module 31 is used for executing the control system, and the operation control display module 32 is used for inputting and setting a cold supply target, operating the control system and displaying control information; the data processing display module 24 and the operation control display module 32 are implemented on the same hardware and software; the data processing module 2 compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training samples meeting the requirement exist, the training samples form a similar training sample set, and finds out the sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point as the most similar training sample;
when the most similar training sample exists, calculating the difference between the load predicted value at the next time point, which is recorded in the test sample at the current time point and calculated by the data processing module 2, and the actual load value at the next time point, which is recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample;
when the actual load difference of the most similar sample is greater than the approach deviation limit, the data processing module 2 performs acceleration adjustment on the adjustment speed of the refrigeration system through the operation control module 3 in a time period between the current time point and the next time point on the basis of the actual adjustment speed related parameter recorded in the most similar sample;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module 2 performs deceleration adjustment on the adjustment speed of the refrigeration system through the operation control module 3 in a time period between the current time point and the next time point on the basis of the actual adjustment speed related parameter recorded in the most similar sample;
when the operation control module 3 adjusts the adjusting speed of the refrigerating system, the acceleration adjustment is to accelerate the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system, and the deceleration adjustment is to slow the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system;
if the current time point is the next time point, the operation control module 3 does not adjust the adjusting speed of the refrigeration system in the time period between the current time point and the next time point;
the data processing module 2 is further configured to modify a training sample set, at a next time point, the data processing module 2 forms a new training sample at the current time point after taking the acquired actual load value at the next time point as an output value of the test sample at the current time point, adds the new training sample to the training sample set, and determines whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module 1 is equal to a set corresponding cooling target and whether the actual load value at the next time point is equal to the load predicted value at the next time point calculated by the data processing module 2;
when the judgment is not true, the data processing module 2 calculates the load prediction deviation at the next time point, adds the load prediction deviation to the output value of the new training sample to be used as the new output value of the new training sample, and performs corresponding processing according to the condition whether the similar training sample set exists or not;
when a similar training sample set exists, the data processing module 2 calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and refers to the difference as the similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set;
the data processing module 2 is further configured to perform an iterative loop, the data processing module 2 sets a next time point as a current time point, determines whether the system has finished operating a set time period and refers to the set time period as a model correction period, when the system has not finished operating a model correction period, the control of re-correcting or deleting the training sample is executed, when the system has finished operating a model correction period, the control of re-establishing the support vector machine load prediction model is executed, the next iterative loop is entered, and the operation control of the refrigeration system is continued.
Example two:
the present embodiment is different from the first embodiment in that: the embodiment is a refrigeration system energy-saving control method and system based on machine learning for a production cold process; the machine is learned as an artificial neural network, in step S1, the cooling system is a process cooling system, and the operation condition parameters include, but are not limited to, outdoor temperature, humidity and solar radiation intensity, process cooling temperature, flow rate of a material to be cooled, input power of a refrigerator, chilled water flow rate, chilled water inlet and outlet temperature, chilled water pump operation power, chilled water pump operation frequency, cooling water flow rate, cooling water inlet and outlet temperature, cooling water pump operation power, cooling water pump operation frequency, and cooling tower power; in step S3, the load prediction model adopts a BP network model, and the kernel function used by the BP network model is a radial basis kernel function; in step S6, the cooling target is a process cooling temperature.
The remaining features of this embodiment are the same as those of the first embodiment.
The invention comprehensively considers the advantages of simplicity, maturity and energy-saving effect of the traditional refrigeration system control and the advantages of machine learning load prediction that the load prediction can be carried out without considering various strong dynamic characteristics of the refrigeration system and the prediction result is quite accurate as long as the training data is accurate, thereby overcoming the problems that the traditional refrigeration system control can not achieve the maximum energy saving and the training sample is inaccurate in the traditional machine learning prediction method and meeting the accuracy and real-time requirements of prediction and control; the control method collects the operating condition parameters of the refrigeration system and the cold utilization system to obtain a training sample set, machine learning models the obtained training sample set and carries out load prediction, the control system carries out control mode adjustment according to a prediction result, new training samples are added or the sample set and the model are corrected according to the adjustment result, the energy-saving control system automatically obtains the training sample set for modeling and carries out machine learning modeling, and then a regression function is corrected, so that the accuracy of the training samples and the load prediction model is continuously improved, the control system is guided to carry out adjustment of the adjustment mode, manual intervention is not needed, interference and deviation possibly brought by manual intervention are reduced, and an optimal adjustment method is provided; by repeatedly iterating the modeling, predicting, controlling, adjusting and model correcting process, the accuracy and real-time performance of prediction and control can be continuously improved, and on the premise of ensuring that the related set cooling target such as indoor set temperature for a central air conditioner, process cooling temperature for a cold production process and the like is met, the basic requirements such as comfort level are ensured, and simultaneously, the operation is carried out with optimal efficiency, the purpose of reducing the energy consumption of the system to the maximum extent is achieved, and the optimal energy-saving effect is achieved.
The above disclosure is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is defined by the appended claims.
The invention can be widely applied to the field of refrigeration.
Claims (10)
1. A refrigeration system energy-saving control method based on machine learning is characterized in that: the method utilizes a refrigeration system energy-saving control system based on machine learning to carry out energy-saving control, the refrigeration system energy-saving control system based on machine learning comprises a data acquisition module, a data processing module and an operation control module, and the method comprises the following steps:
s1: initializing operation condition parameters: the data acquisition module acquires the operating condition parameters of the refrigeration system and the cold using system at each time point according to a certain time step within a certain time range and sends the operating condition parameters to the data processing module;
s2: initializing a training sample set: the data processing module sets a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the similarity degree of output values of the two samples, and other parameters for data processing and calculation, the data processing module forms the training sample of each time point according to a certain sample forming rule by using the operating condition parameter of each time point, the training sample of each time point forms an initial training sample set, and the training sample of each time point follows the following sample forming rule: the training sample consists of an input vector and an output value, each element of the input vector of the training sample at the time point is an operation condition parameter which is acquired at the time point and arranged according to a specified sequence, and the output value of the training sample at the time point is a refrigeration load value acquired at the next time point;
s3: establishing a machine learning load prediction model: the data processing module establishes a load prediction model by using machine learning according to the training sample set;
s4: predicting the load at the next time point: the data acquisition module acquires the operating condition parameters of the current time point, and the data processing module forms the input vector of the test sample of the current time point according to the rule of the step S2, substitutes the input vector into the load prediction model and calculates the load prediction value of the next time point;
s5: adjusting the adjusting speed of the refrigerating system: the data processing module compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training sample meeting the requirement exists, the training sample forms a similar training sample set, and finds out the sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point as the most similar training sample;
when the most similar training sample exists, calculating the difference between the load predicted value of the next time point and the actual load value of the next time point recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample;
when the actual load difference of the most similar sample is larger than the approach deviation limit, the data processing module performs acceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module performs deceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the current time point is the next time point, the operation control module does not adjust the adjusting speed of the refrigerating system in the time period between the current time point and the next time point;
s6: correcting a training sample set: at the next time point, the data processing module takes the acquired actual load value at the next time point as the output value of the test sample at the current time point to form a new training sample at the current time point, adds the new training sample into the training sample set, and judges whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module is equal to the set corresponding cooling target and whether the actual load value at the next time point is equal to the load predicted value at the next time point in the step S4;
when the judgment is not true, the data processing module calculates the load prediction deviation at the next time point, adds the load prediction deviation to the output value of the new training sample to be used as the new output value of the new training sample, and performs corresponding processing according to the condition whether the similar training sample set exists in the step S5;
when a similar training sample set exists, the data processing module calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and the absolute value is called as a similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set;
s7: and (3) iterative loop: and the data processing module sets the next time point as the current time point, judges whether the system finishes operating a set time period and calls the set time period as a model correction cycle, and when the system finishes operating the model correction cycle, the data processing module goes to execute step S4, and when the system finishes operating the model correction cycle, the data processing module goes to execute step S3 to continue operating and controlling the refrigeration system.
2. The machine learning based refrigerant system energy conservation control method of claim 1 wherein: in step S3, the input of the load prediction model is the input vector, and the output is the output value.
3. The machine learning based refrigerant system energy conservation control method of claim 1 wherein: in step S2, the degree of similarity of the input vectors of the two samples uses the 2-norm of the difference between the input vectors of the two samples, i.e., the square root of the sum of the squares of all the corresponding element differences of the input vectors of the two samples.
4. The machine learning based refrigerant system energy conservation control method of claim 1 wherein: the machine learning is a support vector machine, in step S3, an epsilon-SVR model is adopted as a load prediction model, and a kernel function used by the epsilon-SVR model is a radial basis kernel function; or, the machine is learned as an artificial neural network, and in step S3, the load prediction model adopts a BP network model, and the kernel function used by the BP network model is a radial basis kernel function.
5. The machine learning based refrigerant system energy conservation control method of claim 1 wherein: in step S5, the acceleration adjustment is to accelerate the frequency adjustment speed of the frequency converters of the water pump and the blower of the refrigeration system, and the deceleration adjustment is to decelerate the frequency adjustment speed of the frequency converters of the water pump and the blower of the refrigeration system.
6. The utility model provides a refrigerating system energy-saving control system based on machine learning which characterized in that: the system comprises a data acquisition module, a data processing module and an operation control module;
the data acquisition module is connected with the data processing module and used for initializing operating condition parameters, acquiring the operating condition parameters of the refrigeration system and the cold using system at each time point according to a certain time step within a certain time range and sending the operating condition parameters to the data processing module;
the data processing module is connected with the data acquisition module and used for initializing a training sample set, setting a non-negative similarity deviation limit for restricting the similarity degree of input vectors of two samples, a non-negative proximity deviation limit for restricting the proximity degree of output values of the two samples, and other parameters for data processing and calculation, wherein the data processing module forms the operating condition parameters of each time point into the training samples of the time point, the training samples of each time point form an initial training sample set, and the training samples of each time point follow the following sample forming rules: the training sample consists of an input vector and an output value, each element of the input vector of the training sample at the time point is an operation condition parameter which is acquired at the time point and arranged according to a specified sequence, and the output value of the training sample at the time point is a refrigeration load value acquired at the next time point;
the data processing module is also used for establishing a machine learning load prediction model, and the machine learning is used for establishing the load prediction model according to the training sample set;
the data processing module is also used for predicting the load at the next time point, the data acquisition module acquires the operating condition parameters at the current time point, and the data processing module forms the input vector of the test sample at the current time point according to a certain sample forming rule and substitutes the input vector into the load prediction model to calculate the load prediction value at the next time point;
the operation control module is connected with the data processing module and used for adjusting the adjusting speed of the refrigerating system, the data processing module compares the test sample at the current time point with each training sample in the training sample set, judges whether the similarity degree of the input vector of the training sample and the input vector of the test sample at the current time point is within the similar deviation limit, when the training sample meeting the requirement exists, the training samples form a similar training sample set, and the sample with the maximum similarity degree of the input vector in the similar training sample set and the input vector of the test sample at the current time point is found out to be used as the most similar training sample;
when the most similar training sample exists, calculating the difference between the load predicted value of the next time point and the actual load value of the next time point recorded in the most similar training sample, and calling the difference as the actual load difference of the most similar sample;
when the actual load difference of the most similar sample is larger than the approach deviation limit, the data processing module performs acceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the actual load difference of the most similar sample is smaller than the negative approach deviation limit, the data processing module performs deceleration adjustment on the adjusting speed of the refrigeration system through the operation control module in a time period between the current time point and the next time point on the basis of the actual adjusting speed related parameter recorded in the most similar sample;
when the current time point is the next time point, the operation control module does not adjust the adjusting speed of the refrigerating system in the time period between the current time point and the next time point;
the data processing module is further used for correcting the training sample set, when the next time point is reached, the data processing module takes the acquired actual load value of the next time point as the output value of the test sample of the current time point to form a new training sample of the current time point, adds the new training sample into the training sample set, and judges whether the actual cooling effect of the cooling system at the next time point acquired by the data acquisition module is equal to the set corresponding cooling target or not and whether the actual load value at the next time point is equal to the load predicted value at the next time point or not;
when the judgment is not true, the data processing module calculates the load prediction deviation of the next time point, the load prediction deviation is added to the output value of the new training sample to serve as the new output value of the new training sample, and corresponding processing is carried out according to the condition whether a similar training sample set exists or not;
when a similar training sample set exists, the data processing module calculates the absolute value of the difference between the output value of each training sample in the similar training sample set and the output value of a new training sample and the absolute value is called as a similar sample output difference, finds out the training sample with the largest similar sample output difference and larger than the approximate deviation limit, and deletes the training sample from the training sample set;
the data processing module is also used for iterative cycle, the data processing module sets the next time point as the current time point, judges whether the system finishes operating a set time period and calls the set time period as a model correction period, when the system finishes operating a model correction period, the control of re-correcting or deleting the training sample is executed, when the system finishes operating a model correction period, the control of re-establishing the machine learning load prediction model is executed, the next iterative cycle is entered, and the operation control of the refrigeration system is continued.
7. The machine learning based refrigerant system economizer control system of claim 6 wherein: the data acquisition module comprises a data sensing module and a data transformation module, wherein the data sensing module and the data transformation module are installed on an outdoor and/or indoor refrigerating system and a cold system, the data sensing module is used for acquiring the influence parameters influencing the load of the outdoor and/or indoor refrigerating system and the working condition parameters of the running state parameters of the refrigerating system and the cold system, and the data transformation module is used for converting the working condition parameters acquired by the data sensing module into data signals which can be identified by the data processing module and sending the data signals to the data processing module.
8. The machine learning based refrigerant system economizer control system of claim 6 or 7 wherein: the data processing module comprises a data transmission module, a data calculation module, a database and a data processing and displaying module, wherein the data transmission module is used for receiving a data signal sent by the data acquisition module and sending an adjusting speed adjusting parameter in a time period between the current time point and the next time point to the operation control module, the data calculation module is used for establishing and calculating correction data, samples, models, load prediction and adjusting speed adjusting parameters, the database is used for recording sample data and calculating parameter data, and the data processing and displaying module is used for visually displaying and operating the data processing module; the data computing module and the database are located in a remote server and serve as cloud computing terminals.
9. The machine learning based refrigerant system economizer control system of claim 8 wherein: the operation control module comprises a control execution module and an operation control display module, the control execution module is used for executing a control system, and the operation control display module is used for inputting a set cooling target, operating the control system and displaying control information; the data processing display module and the operation control display module are realized on the same hardware and software.
10. The machine learning based refrigerant system economizer control system of claim 6 wherein: the refrigerating system comprises a water pump, a fan and a frequency converter, when the operation control module adjusts the adjusting speed of the refrigerating system, the acceleration adjustment is to accelerate the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system, and the deceleration adjustment is to slow the frequency adjusting speed of the frequency converters of the water pump and the fan of the refrigerating system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810174214.9A CN108317670B (en) | 2018-03-02 | 2018-03-02 | Refrigeration system energy-saving control method and system based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810174214.9A CN108317670B (en) | 2018-03-02 | 2018-03-02 | Refrigeration system energy-saving control method and system based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108317670A CN108317670A (en) | 2018-07-24 |
| CN108317670B true CN108317670B (en) | 2020-01-07 |
Family
ID=62900242
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810174214.9A Active CN108317670B (en) | 2018-03-02 | 2018-03-02 | Refrigeration system energy-saving control method and system based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108317670B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109213098B (en) * | 2018-08-29 | 2021-11-26 | 西门子电力自动化有限公司 | Method, device, electronic equipment and computer readable medium for adjusting operation parameters |
| JP6760348B2 (en) | 2018-10-11 | 2020-09-23 | 株式会社富士通ゼネラル | Air conditioner, data transmission method and air conditioner system |
| CN109520071A (en) * | 2018-10-18 | 2019-03-26 | 珠海东之尼电子科技有限公司 | A kind of air-conditioning self-adaptation control method and system based on support vector machines study |
| CN109882996A (en) * | 2019-01-25 | 2019-06-14 | 珠海格力电器股份有限公司 | A kind of method and apparatus of control |
| CN110145904A (en) * | 2019-04-18 | 2019-08-20 | 广州汇电云联互联网科技有限公司 | A kind of cold group artificial intelligence cloud control system |
| CN110186156A (en) * | 2019-06-03 | 2019-08-30 | 西安锦威电子科技有限公司 | Refrigeration plant Fuzzy control system |
| EP3764173A1 (en) * | 2019-07-08 | 2021-01-13 | ABB Schweiz AG | Human-plausible automated control of an industrial process |
| CN111059722A (en) * | 2019-11-27 | 2020-04-24 | 广东美的制冷设备有限公司 | Control method and device of air conditioning equipment and server |
| CN110864400A (en) * | 2019-11-27 | 2020-03-06 | 广东美的制冷设备有限公司 | Control method and device of air conditioning equipment and air conditioning equipment |
| CN112304436B (en) * | 2020-10-23 | 2022-07-19 | 北京百度网讯科技有限公司 | Method, apparatus, electronic device, and medium for creating temperature correction model |
| CN113570066B (en) * | 2021-07-23 | 2024-03-29 | 中国恩菲工程技术有限公司 | Data processing method, system, electronic device and storage medium |
| CN113608596B (en) * | 2021-07-29 | 2024-05-24 | 上海德衡数据科技有限公司 | Intelligent cooling method and system for server |
| CN115388586B (en) * | 2022-08-24 | 2023-05-12 | 南栖仙策(南京)高新技术有限公司 | Refrigeration control model, device, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201463177U (en) * | 2009-01-16 | 2010-05-12 | 深圳职业技术学院 | Central air conditioner optimized energy-saving control system based on load forecasting |
| CN102997374A (en) * | 2012-12-31 | 2013-03-27 | 深圳市奥宇控制系统有限公司 | Method and device for forecasting air-conditioning load and air-conditioner |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH06147598A (en) * | 1992-11-09 | 1994-05-27 | Takasago Thermal Eng Co Ltd | Air conditioning load prediction method |
-
2018
- 2018-03-02 CN CN201810174214.9A patent/CN108317670B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201463177U (en) * | 2009-01-16 | 2010-05-12 | 深圳职业技术学院 | Central air conditioner optimized energy-saving control system based on load forecasting |
| CN102997374A (en) * | 2012-12-31 | 2013-03-27 | 深圳市奥宇控制系统有限公司 | Method and device for forecasting air-conditioning load and air-conditioner |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
Non-Patent Citations (2)
| Title |
|---|
| Applying support vector machine to predict hourly cooling load in the building;Qiong Li等;《Applying support vector machine to predict hourly cooling load in the building》;20091031;第2249-2256页 * |
| 不同核函数的支持向量机用于空调负荷预测的对比研究;王东等;《不同核函数的支持向量机用于空调负荷预测的对比研究》;20150901;第531-535页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108317670A (en) | 2018-07-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108317670B (en) | Refrigeration system energy-saving control method and system based on machine learning | |
| CN111237989A (en) | Building ventilation air conditioner control method and device based on load prediction | |
| CN104019526B (en) | Improve PSO algorithm Fuzzy Adaptive PID temperature and humidity control system and method | |
| CN110392515A (en) | A kind of Cooling and Heat Source equipment room energy-conserving control method and system based on historical data | |
| CN110410942B (en) | Energy-saving optimization control method and system for cold and heat source machine room | |
| CN105571073B (en) | A kind of subway station air-conditioner water system frequency conversion control energy-saving method | |
| CN113326651B (en) | Refrigerating station load and energy efficiency ratio dynamic modeling method based on T-S fuzzy model | |
| CN108489013A (en) | Central air-conditioner control method based on genetic algorithm and load on-line amending and device | |
| CN107940679B (en) | Group control method based on performance curve of water chilling unit of data center | |
| CN111649457B (en) | Dynamic predictive machine learning type air conditioner energy-saving control method | |
| CN103912966A (en) | Optimal control method for ground source heat pump refrigerating system | |
| CN113739365A (en) | Central air-conditioning cold station group control energy-saving control method, device, equipment and storage medium | |
| CN111047117B (en) | Differential pressure bypass valve energy-saving optimization method based on terminal load prediction | |
| CN111306728B (en) | Air volume regulating and controlling method and device for air conditioning system and computer equipment | |
| CN114154677A (en) | Air conditioner operation load model construction and prediction method, device, equipment and medium | |
| CN112070353A (en) | Method and system for accurately detecting energy efficiency of data center | |
| CN113778215A (en) | Method for realizing data center PUE prediction and consumption reduction strategy based on big data | |
| CN114046593A (en) | Dynamic predictive machine learning type air conditioner energy-saving control method and system | |
| CN110966714B (en) | Intelligent control method for air conditioner, computer readable storage medium and air conditioner | |
| CN116907036A (en) | Deep reinforcement learning water chilling unit control method based on cold load prediction | |
| CN115682324A (en) | Energy-saving optimization control method, device, equipment and storage medium for central air-conditioning system | |
| CN117146382B (en) | Intelligent adaptive system optimization method | |
| CN112180733B (en) | Fuzzy logic-based building energy consumption system prediction control parameter setting method | |
| CN113446713A (en) | Digital twin-based intelligent data optimal control and energy saving method and system | |
| CN111520874A (en) | Water system air conditioner room energy efficiency detection optimization system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |