1. A cold machine number control method based on GRNN is characterized by comprising the following steps:

acquiring the running characteristics of a water chilling unit according to a preset rule, wherein the running characteristics of the water chilling unit comprise at least one of the current time, the number of the current working coolers, the load rate and the flow rate of the interior of the cooler, the return water temperature of chilled water, the outlet water temperature of the chilled water, the condensation temperature, the evaporation temperature, the compressor frequency, the refrigerant tank filling amount, the refrigeration amount and the power consumption;

screening relevant characteristics influencing the energy efficiency ratio of the water chilling unit at the Ti moment to form a first characteristic set, and calculating the energy efficiency ratio of the water chilling unit at the Ti moment;

screening out relevant characteristics influencing the external efficiency and the internal efficiency of the water chilling unit Ti moment to form a second characteristic set, and calculating the external efficiency and the internal efficiency of the water chilling unit Ti moment;

establishing an internal efficiency prediction model of the water chilling unit; the internal efficiency prediction model is obtained by calculation through a GRNN network training model;

and according to the operation characteristics of the water chilling unit, in combination with an internal efficiency prediction model of the water chilling unit, predicting and selecting the working condition with the highest internal efficiency to start the corresponding water chilling unit according to the air conditioner load demand at the moment of Ti + 1.

2. The GRNN-based chiller number control method according to claim 1, wherein the preset rule is data of operating characteristics of the chiller for not less than six months, and sampling intervals of the operating characteristics are five minutes.

3. The GRNN-based number of cold machines control method according to claim 2, wherein after setting the operation characteristics of the cold water machine set according to a preset rule, before screening the first characteristic set and the second characteristic set from the operation characteristics, the method further comprises:

and processing the missing value and the abnormal value in the operation characteristic, and updating the operation characteristic of the water chilling unit.

4. The GRNN-based number of cold machines control method according to claim 3, wherein calculating the energy efficiency ratio of the cold water machine set at the time Ti according to the first feature set includes:

screening relevant operation characteristics from the operation characteristics of the water chilling unit to form a first characteristic set, wherein the first characteristic set comprises refrigerating capacity and power consumption;

calculating the energy efficiency ratio at the moment Ti by using the following formula:

wherein: COP_tiRepresents the energy efficiency ratio at the time of Ti, Q_tiShows the cooling capacity at the moment of Ti, W_tiRepresents the power consumption at time Ti.

5. The GRNN-based number of cold machines control method according to claim 4, wherein the calculating of the external efficiency and the internal efficiency of the cold water machine set at the time Ti according to the second feature set specifically includes:

screening relevant operation characteristics from the operation characteristics of the water chilling unit to form a second characteristic set, wherein the second characteristic set comprises an evaporation temperature and a condensation temperature;

the external efficiency was calculated using the following formula:

wherein: ICOP_TiRepresents the external efficiency at the moment of Ti, Ti_evDenotes the evaporation temperature at the time of Ti, Ti_edRepresents the condensation temperature at the time of Ti;

calculating internal efficiency according to the energy efficiency ratio and the external efficiency:

wherein: DCOP_TiIndicating the internal efficiency at time Ti.

6. The GRNN-based cold machine number control method according to claim 5, wherein the GRNN network training model establishing process is:

screening out operation characteristics influencing internal efficiency to form a training set and a testing set, wherein the training set and the testing set respectively comprise at least one of flow, return water temperature of chilled water, outlet water temperature of chilled water, the number of cold machines to be started, condensing temperature, evaporating temperature and compressor frequency;

inputting the training set as a neuron into a GRNN network training model;

calculating the distance between the training set and the test set, and outputting a training value by using a radial-based activation function;

wherein: p is a radical of_iFor activating the output of the functional neuron i, σ is the smoothing parameter, n is the number of training sets, d_iThe distance between the training set and the test set;

wherein: x ═ X₁，x₂，…，x_m) For the input of test setsInput vector, x_jIs the value of the jth element therein, X_ij＝(x_i1，x_i2，…，x_im) As input vector, x, of the ith training set_ijIs the jth element value, m is the number of elements of the input vector, w_jIs the weight of the jth element;

calculating the sum of products of the training output data and the weight values of the activation function:

S＝{Y}^T·p

wherein Y ═ { Y ═ Y₁，Y₂，…，Y_nThe matrix is an output vector matrix of the training set, and p is the output of the activation function;

calculate the sum of all weight values:

where n is the number of neurons in the model layer, p_iThe output of the ith neuron of the mode layer;

obtaining an output value according to the sum of products of the training output data and the weight values of the activation function and the sum of the ownership weight values

Wherein: y represents the estimated DCOP value.

7. The GRNN-based cold machine number control method according to claim 6, wherein a training process of the GRNN network training model specifically includes:

continuously updating the characteristics of an input training set and a test set of the GRNN network training model, and performing loop training iteration until the smoothing parameter sigma meets a set threshold value;

storing the GRNN network training model meeting the smoothing parameter sigma;

testing the model by using the test set, calculating the error of the test set, and storing the GRNN network training model when the error meets a threshold range;

and taking the GRNN network training model as an internal efficiency prediction model of the water chilling unit.

8. The GRNN-based number of cold water units control method according to claim 7, wherein the specific process of predicting and selecting the working condition with the highest internal efficiency to start the corresponding cold water unit is as follows:

predicting the air conditioner load demand at the Ti +1 moment by using an energy consumption prediction method based on transfer learning;

using a screening functionWherein F is a historical operating condition set, F_iIs the feature vector of one of the operating conditions, f is the feature vector of the current operating condition, D (f)_iF) calculating the Euclidean distance of the two characteristic vectors, and calculating the internal efficiency of starting the water chilling unit by using the corresponding characteristic of the working condition as an input parameter;

and selecting the working condition parameter with the highest internal efficiency to control the starting of the water chilling unit.

9. A cold machine number control system based on GRNN is characterized by comprising:

the processing module is used for receiving and transmitting data, calculating, generating control logic and sending a control instruction to local BA equipment;

the acquisition module is used for acquiring the operation characteristics of the water chilling unit;

the training library module is used for storing and updating the operation characteristics of the water chilling unit;

the weather data module is used for acquiring weather information;

the training module is used for establishing a GRNN network training model and training by utilizing the running characteristics of the water chilling unit;

and the control module is used for executing the control command sent by the processing module and adjusting the operation of the water chilling unit.

Background

The building energy consumption accounts for about 1/4-1/3 of the total social energy consumption, and is still increasing along with the development of urbanization process. The significance of energy conservation and emission reduction in the construction industry is great.

During the whole life cycle of the building, the energy consumption in the operation stage is the most. In the energy consumption of building operation, the energy consumption of air conditioners and illumination accounts for the main part. Especially, the central air conditioning units of large public buildings have very high energy consumption, so that a plurality of water cooling units of the central air conditioners need to be monitored, and the energy conservation of the air conditioning system is well realized.

Disclosure of Invention

The invention aims to provide a cold machine number control method based on GRNN and a control system thereof, which solve the problems that in the prior art, after a central air conditioner is started, a plurality of cold water machine sets are started simultaneously, so that the energy consumption is high, and the number of the cold water machine sets cannot be started intelligently according to the operation working condition.

In order to solve the above technical problem, an embodiment of the present invention provides a method for controlling the number of cold machines based on GRNN, including:

acquiring the running characteristics of a water chilling unit according to a preset rule, wherein the running characteristics of the water chilling unit comprise at least one of the current time, the number of the current working coolers, the load rate and the flow rate of the interior of the cooler, the return water temperature of chilled water, the outlet water temperature of the chilled water, the condensation temperature, the evaporation temperature, the compressor frequency, the refrigerant tank filling amount, the refrigeration amount and the power consumption;

screening relevant characteristics influencing the energy efficiency ratio of the water chilling unit at the Ti moment to form a first characteristic set, and calculating the energy efficiency ratio of the water chilling unit at the Ti moment;

screening out relevant characteristics influencing the external efficiency and the internal efficiency of the water chilling unit Ti moment to form a second characteristic set, and calculating the external efficiency and the internal efficiency of the water chilling unit Ti moment;

establishing an internal efficiency prediction model of the water chilling unit; the internal efficiency prediction model is obtained by calculation through a GRNN network training model;

and according to the operation characteristics of the water chilling unit, in combination with an internal efficiency prediction model of the water chilling unit, predicting and selecting the working condition with the highest internal efficiency to start the corresponding water chilling unit according to the air conditioner load demand at the moment of Ti + 1.

Has the advantages that: the method comprises the steps of collecting the operation characteristics of a water chilling unit of the central air conditioner, and obtaining an internal efficiency prediction model of the water chilling unit by combining a GRNN network training model; and summarizing the existing data by adopting an artificial intelligence machine learning algorithm, thereby calculating the optimal internal efficiency at each working condition moment by adopting the GRNN network training model, and calculating the optimal internal efficiency under the working condition according to the running load of the water chilling unit at the next moment to control the running of the water chilling unit. Compared with the prior art, the method and the device have the advantages that the historical records are integrated, the operation load at the next moment is predicted according to the operation condition at the current moment, the optimal internal efficiency is calculated to control the operation of the water chilling unit, and the energy consumption is reduced as far as possible under the condition that the starting operation of the water chilling unit meets the load requirement.

Further, the preset rule is data of operation characteristics of the water chilling unit in not less than six months, and the sampling interval of the operation characteristics is five minutes.

Further, after the operation features of the water chilling unit are set according to the preset rule, before the first feature set and the second feature set are screened from the operation features, the method further comprises the following steps:

and processing the missing value and the abnormal value in the operation characteristic, and updating the operation characteristic of the water chilling unit.

Further, calculating the energy efficiency ratio of the water chilling unit Ti at the moment according to the first feature set, and the method comprises the following steps:

screening relevant operation characteristics from the operation characteristics of the water chilling unit to form a first characteristic set, wherein the first characteristic set comprises refrigerating capacity and power consumption;

calculating the energy efficiency ratio at the moment Ti by using the following formula:

wherein: COP_tiRepresents the energy efficiency ratio at the time of Ti, Q_tiShows the cooling capacity at the moment of Ti, W_tiRepresents the power consumption at time Ti.

Further, calculating the external efficiency and the internal efficiency of the water chilling unit Ti at the moment according to the second feature set, specifically as follows:

screening relevant operation characteristics from the operation characteristics of the water chilling unit to form a second characteristic set, wherein the second characteristic set comprises an evaporation temperature and a condensation temperature;

the external efficiency was calculated using the following formula:

wherein: ICOP_TiRepresents the external efficiency at the moment of Ti, Ti_evDenotes the evaporation temperature at the time of Ti, Ti_edRepresents the condensation temperature at the time of Ti;

calculating internal efficiency according to the energy efficiency ratio and the external efficiency:

wherein: DCOP_TiIndicating the internal efficiency at time Ti.

Further, the GRNN network training model establishing process is as follows:

screening out operation characteristics influencing internal efficiency to form a training set and a testing set, wherein the training set and the testing set respectively comprise at least one of flow, return water temperature of chilled water, outlet water temperature of chilled water, the number of cold machines to be started, condensing temperature, evaporating temperature and compressor frequency;

inputting the training set as a neuron into a GRNN network training model;

calculating the distance between the training set and the test set, and outputting a training value by using a radial-based activation function;

wherein: p is a radical of_iFor activating the output of the functional neuron i, σ is the smoothing parameter, n is the number of training sets, d_iThe distance between the training set and the test set;

wherein: x ═ X₁,x₂,…,x_m) For input vectors of the test set, x_jIs the value of the jth element therein, X_ij＝(x_i1，x_i2，…，x_im) As input vector, x, of the ith training set_ijIs the jth element value, m is the number of elements of the input vector, w_jIs the weight of the jth element;

calculating the sum of products of the training output data and the weight values of the activation function:

S＝{Y}^T·p

wherein Y ═ { Y ═ Y₁,Y₂，…，Y_nThe matrix is an output vector matrix of the training set, and p is the output of the activation function;

calculate the sum of all weight values:

where n is the number of neurons in the model layer, p_iThe output of the ith neuron of the mode layer;

obtaining an output value according to the sum of products of the training output data and the weight values of the activation function and the sum of the ownership weight values

Wherein: y represents the estimated DCOP value.

Further, the training process of the GRNN network training model specifically includes:

further, continuously updating the characteristics of an input training set and a test set of the GRNN network training model, and performing cyclic training iteration until the smoothing parameter sigma meets a set threshold value;

storing the GRNN network training model meeting the smoothing parameter sigma;

testing the model by using the test set, calculating the error of the test set, and storing the GRNN network training model when the error meets a threshold range;

and taking the GRNN network training model as an internal efficiency prediction model of the water chilling unit.

Further, the specific process of predicting and selecting the working condition with the highest internal efficiency to start the corresponding water chilling unit is as follows:

predicting the air conditioner load demand at the Ti +1 moment by using an energy consumption prediction method based on transfer learning;

using a screening functionWherein F is a historical operating condition set, F_iIs the feature vector of one of the operating conditions, f is the feature vector of the current operating condition, D (f)_iF) calculating the Euclidean distance of the two characteristic vectors, and calculating the internal efficiency of starting the water chilling unit by using the corresponding characteristic of the working condition as an input parameter;

and selecting the working condition parameter with the highest internal efficiency to control the starting of the water chilling unit.

An embodiment of the present invention further provides a system for controlling the number of cold machines based on GRNN, including:

the processing module is used for receiving and transmitting data, calculating, generating control logic and sending a control instruction to local BA equipment;

the acquisition module is used for acquiring the operation characteristics of the water chilling unit;

the training library module is used for storing and updating the operation characteristics of the water chilling unit;

the weather data module is used for acquiring weather information;

the training module is used for establishing a GRNN network training model and training by utilizing the running characteristics of the water chilling unit;

and the building automatic control module is used for executing the control command sent by the processing module and adjusting the operation of the water chilling unit.

Advantageous effects

According to the collected running characteristics of the water chilling unit of the central air conditioner, an internal efficiency prediction model of the water chilling unit is obtained by combining a GRNN network training model; and summarizing the existing data by adopting an artificial intelligence machine learning algorithm, thereby calculating the optimal internal efficiency at each working condition moment by adopting the GRNN network training model, and calculating the optimal internal efficiency under the working condition according to the running load of the water chilling unit at the next moment to control the running of the water chilling unit. Compared with the prior art, the method and the device have the advantages that the historical records are integrated, the operation load at the next moment is predicted according to the operation condition at the current moment, the optimal internal efficiency is calculated to control the operation of the water chilling unit, and the energy consumption is reduced as far as possible under the condition that the starting operation of the water chilling unit meets the load requirement.

Drawings

One or more embodiments are illustrated by the corresponding figures in the drawings, which are not meant to be limiting.

Fig. 1 is a flowchart of a method for controlling the number of cooling units based on GRNN according to a first embodiment of the present invention.

Fig. 2 is a schematic block diagram of a system for controlling the number of cold machine based on GRNN according to a second embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention more apparent, embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, it will be appreciated by those of ordinary skill in the art that numerous technical details are set forth in order to provide a better understanding of the present application in various embodiments of the present invention. However, the technical solution claimed in the present application can be implemented without these technical details and various changes and modifications based on the following embodiments.

The first embodiment:

a first embodiment of the present invention relates to a method and a system for controlling the number of cold machines based on GRNN, including: s1, acquiring the operation characteristics of the water chilling unit; s2, calculating the energy efficiency ratio of the water chilling unit Ti at the moment according to the first feature set; s3, calculating the external efficiency and the internal efficiency of the water chilling unit Ti at the moment according to the second feature set; s4, establishing an internal efficiency prediction model of the water chilling unit; and S5, predicting and selecting the working condition with the highest internal efficiency at the moment of Ti +1 according to the internal efficiency prediction model to start the corresponding water chilling unit.

In S1, the operation characteristics of the chiller include at least one of the current time, the number of chillers currently operating, the load factor, the flow rate, the return water temperature of chilled water, the outlet water temperature of chilled water, the condensing temperature, the evaporating temperature, the frequency of the compressor, the amount of refrigerant filled in the tank, the refrigerating capacity, and the power consumption. In S2, the first feature set includes cooling capacity and power consumption. In S3, the second set of characteristics includes an evaporation temperature and a condensation temperature. In S4, the internal efficiency prediction model is derived from a GRNN network training model, which is trained using a training set and a test set, which are selected from the operating characteristics of the chiller.

According to the collected running characteristics of the water chilling unit of the central air conditioner, an internal efficiency prediction model of the water chilling unit is obtained by combining a GRNN network training model; and summarizing the existing data by adopting an artificial intelligence machine learning algorithm, thereby calculating the optimal internal efficiency at each working condition moment by adopting the GRNN network training model, and calculating the optimal internal efficiency under the working condition according to the running load of the water chilling unit at the next moment to control the running of the water chilling unit. Compared with the prior art, the method and the device have the advantages that the historical records are integrated, the operation load at the next moment is predicted according to the operation condition at the current moment, the optimal internal efficiency is calculated to control the operation of the water chilling unit, and the energy consumption is reduced as far as possible under the condition that the starting operation of the water chilling unit meets the load requirement.

The implementation details of the method for controlling the number of cold machines based on GRNN according to the present embodiment are specifically described below, and the following description is only provided for facilitating understanding of the implementation details, and is not necessary to implement the present embodiment, and a specific flow of the present embodiment is shown in fig. 1, and the present embodiment is applied to a server on a network side.

And S1, acquiring the operation characteristics of the water chilling unit.

Specifically, a communication link is established with the controller of the chiller, and the controller collects the operating parameters of the chiller and then receives the operating parameters from the present step S1. Meanwhile, communication connection is established with a network platform, weather information is obtained and updated in real time, the weather information comprises outdoor temperature and humidity, and the updating rate is one day. And combining the operation parameters and the outdoor temperature and humidity into operation characteristics. Therefore, after the operation parameters and the weather parameters are obtained in step S1, the operation parameters are marked according to the outdoor temperature and humidity as reference marks, so as to form operation characteristics of the water chilling unit under different outdoor temperatures and humidities. The method specifically comprises the following steps: s1-1, receiving the operation parameters of the water chilling unit; s1-2, receiving outdoor temperature and humidity parameters; s1-3, marking the operation parameters of the water chilling unit with different outdoor temperatures and humidities to form operation characteristics; s1-4, screening the operation characteristics of the water chilling unit according to a preset rule; .

The collected data usually covers the data of the external environment, control signals, various key components and the like of the running of the water chilling unit, the data objectively and truly records the history and the instant running condition of the water chilling unit and can be used for tracing the history, detecting abnormal conditions and the like, so the running characteristics of the treated water chilling unit comprise at least one of the current time, the number of the current working chillers, the load rate and the flow rate of the interior of the chiller, the return water temperature of chilled water, the outlet water temperature of the chilled water, the condensation temperature, the evaporation temperature, the compressor frequency, the refrigerant filling amount, the refrigerating amount and the power consumption. For detailed analysis, the operating characteristics of the chiller are divided into two categories: the system comprises current time information, A-type information related to the working of the water chilling unit and B-type information related to the operation parameters of the water chilling unit.

Information of type A, such as the number of current working coolers, power consumption and the like; the type B information comprises the load factor, the flow rate, the return water temperature of the chilled water, the outlet water temperature of the chilled water, the condensation temperature, the evaporation temperature, the frequency of the compressor, the tank filling amount of the refrigerant and the refrigerating capacity of the inside of the refrigerator.

S1-4: the preset rule is that the running characteristic of the water chilling unit is continuously operated for not less than six months, and the sampling interval of the running characteristic is five minutes.

In another example, after the operation parameters and the weather parameters are acquired in S1, the operation characteristics of the water chilling unit are formed, specifically: receiving operation parameters of a water chilling unit; s1-2, receiving outdoor temperature and humidity parameters; and S1-3, marking the operating parameters of the water chilling unit with different outdoor temperatures and humidities to form operating characteristics.

In one example, after S1-4 is completed, and before S2-1 is executed, the following operations are performed: and processing the missing value and the abnormal value in the operation characteristic, and updating the operation characteristic of the water chilling unit. And cleaning and sorting the data with the plurality of operation characteristics obtained in the step S1-4, processing missing value processing and abnormal value processing by using a conventional method, presenting the data in a section of complete operation process as much as possible, keeping the selected data to have a stable sampling rate, and improving the robustness of an analysis result.

S2 calculating the energy efficiency ratio of the water chilling unit Ti at the moment according to the first feature set.

Specifically, the energy efficiency ratio of the water chilling unit Ti at the moment is a reference index of the operation efficiency of the reaction cooler, and relevant characteristics are screened from the updated operation characteristics of the water chilling unit and calculated.

S2-1, screening relevant operation characteristics influencing the energy efficiency ratio of the water chilling unit at the moment Ti from the operation characteristics of the water chilling unit to form a first characteristic set, wherein the first characteristic set comprises refrigerating capacity and power consumption;

s2-2, calculating the energy efficiency ratio at the moment Ti by using the following formula:

wherein: COP_tiRepresents the energy efficiency ratio at the time of Ti, Q_tiShows the cooling capacity at the moment of Ti, W_tiRepresents the power consumption at time Ti.

And S3, calculating the external efficiency and the internal efficiency of the water chilling unit Ti at the moment according to the second feature set.

Specifically, the influence factors of the energy efficiency ratio of the water chilling unit Ti at the moment include two major factors of the external operating environment and the performance of the water chilling unit, two reference indexes of the external efficiency and the internal efficiency of the chiller are introduced according to the two major factors, wherein the external efficiency actually reflects the ideal energy efficiency ratio of the water chilling unit, the internal efficiency reflects the deviation between the energy efficiency ratio and the external efficiency, and the influence factors inside and outside the water chilling unit are comprehensively analyzed according to three indexes of COP, ICOP and DCOP.

S3-1, screening out operation characteristics related to external efficiency and internal efficiency at the moment of influencing the water chilling unit Ti from the operation characteristics of the water chilling unit to form a second characteristic set, wherein the second characteristic set comprises an evaporation temperature and a condensation temperature;

the external efficiency was calculated using the following formula:

wherein: ICOP_TiRepresents the external efficiency at the moment of Ti, Ti_evDenotes the evaporation temperature at the time of Ti, Ti_edRepresents the condensation temperature at the time of Ti;

calculating internal efficiency according to the energy efficiency ratio and the external efficiency:

wherein: DCOP_TiIndicating the internal efficiency at time Ti.

And S4, establishing an internal efficiency prediction model of the water chilling unit.

Specifically, the internal efficiency prediction model of the water chilling unit is obtained through calculation of a GRNN network training model, and the GRNN network training model is trained by combining a training set and a test set which are screened from operation characteristics.

The specific process of training the GRNN network training model by combining a training set and a test set which are screened from the operation characteristics comprises the following steps:

s4-1, screening out operation characteristics influencing internal efficiency to form a training set and a testing set, wherein the training set and the testing set respectively comprise at least one of flow, return water temperature of chilled water, outlet water temperature of chilled water, number of cold machines, condensing temperature, evaporating temperature and compressor frequency;

s4-2, inputting the training set as a neuron into a GRNN network training model;

calculating the distance between the training set and the test set, and outputting a training value by using a radial-based activation function;

wherein: p is a radical of_iFor activating the output of the functional neuron i, σ is the smoothing parameter, n is the number of training sets, d_iThe distance between the training set and the test set;

wherein: x ═ X₁，x₂，…，x_m) For input vectors of the test set, x_jIs the value of the jth element therein, X_ij＝(x_i1，x_i2，…，x_im) As input vector, x, of the ith training set_ijIs the jth element value, m is the number of elements of the input vector, w_jIs the weight of the jth element.

S4-3, calculating the sum of products of the training output data and the weight values of the activation function:

S＝{Y}^T·p

wherein Y ═ { Y ═ Y₁,Y₂,…,Y_nIs the output vector matrix of the training set, p is shockThe output of the live function;

calculate the sum of all weight values:

where n is the number of neurons in the model layer, p_iThe output of the ith neuron of the mode layer;

s4-4, obtaining an output value according to the sum of the products of the training output data and the weight values of the activation function and the sum of the ownership weight values

Wherein: y represents the estimated DCOP value.

S4-5, continuously updating the characteristics of an input training set and a test set of the GRNN network training model, circularly training and iterating until the smoothing parameter sigma meets a set threshold, and storing the GRNN network training model meeting the smoothing parameter sigma;

s4-6, testing the model by using the test set, calculating the error of the test set, and storing the GRNN network training model when the error meets the threshold range; and taking the GRNN network training model as an internal efficiency prediction model of the water chilling unit.

And S5, according to the operation characteristics of the water chilling unit, combining with the internal efficiency prediction model of the water chilling unit, and according to the air conditioning load demand at the moment of Ti +1, predicting and selecting the working condition with the highest internal efficiency to start the corresponding water chilling unit.

S5-1, predicting the air conditioning load demand at the Ti +1 moment by using an energy consumption prediction method based on transfer learning;

specifically, a prediction model is established, the final prediction value of the prediction model is determined by calculating the periodic component and the residual component of the energy consumption data of the water chilling unit and adjusting the periodic component based on a least square method, the water chilling unit similar to a target water chilling unit lacking historical energy consumption data and having abundant historical energy consumption data is used for training the prediction model, then the trained model is directly applied to the energy consumption prediction of the target water chilling unit, the parameter setting of the prediction model is finely adjusted according to the input and output data of the target water chilling unit, the time for retraining the prediction model to predict the energy consumption data is reduced, and the efficiency of predicting the energy consumption data of the target water chilling unit is greatly improved.

S5-2, using a screening functionWherein F is a historical operating condition set, F_iIs the feature vector of one of the operating conditions, f is the feature vector of the current operating condition, D (f)_iF) calculating the Euclidean distance of the two characteristic vectors, and calculating the internal efficiency of starting the water chilling unit by using the corresponding characteristic of the working condition as an input parameter;

and S5-3, selecting the working condition parameter with the highest internal efficiency to control the starting of the water chilling unit.

The steps of the above methods are divided for clarity, and the implementation may be combined into one step or split some steps, and the steps are divided into multiple steps, so long as the same logical relationship is included, which are all within the protection scope of the present patent; it is within the scope of the patent to add insignificant modifications to the algorithms or processes or to introduce insignificant design changes to the core design without changing the algorithms or processes.

Second embodiment:

a second embodiment of the present invention provides a system for controlling the number of cold machines based on GRNN, including:

the processing module 201 is configured to receive and transmit data, calculate, generate a control logic, and send a control instruction to the local BA device;

the acquisition module 202 is used for acquiring the operation characteristics of the water chilling unit;

the training library module 203 is used for storing and updating the running characteristics of the water chilling unit;

a weather data module 204 for acquiring weather information;

the training module 205 is configured to establish a GRNN network training model and train by using the operation characteristics of the chiller;

and the control module 206 is configured to execute the control instruction sent by the processing module, and adjust the operation of the water chilling unit.

The processing module 201 adopts an edge computing controller, the acquisition module 202 adopts a bottom-layer energy consumption monitoring platform or sensor, the training library module 203 adopts a memory, the meteorological data module 204 adopts a meteorological data API (application program interface) interface to acquire weather information of a network platform and store the weather information in the training library module 203, the training module 205 adopts a background server, the training process can be repeated periodically along with the increase of the running characteristic data quantity of the water chilling unit, and the model is issued to the edge computing controller after the training is finished; the control module 206 executes a control instruction issued by the edge computing controller by using a local BA device.

It should be understood that this embodiment is a system example corresponding to the first embodiment, and may be implemented in cooperation with the first embodiment. The related technical details mentioned in the first embodiment are still valid in this embodiment, and are not described herein again in order to reduce repetition. Accordingly, the related-art details mentioned in the present embodiment can also be applied to the first embodiment.

It should be noted that each module referred to in this embodiment is a logical module, and in practical applications, one logical unit may be one physical unit, may be a part of one physical unit, and may be implemented by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, elements that are not so closely related to solving the technical problems proposed by the present invention are not introduced in the present embodiment, but this does not indicate that other elements are not present in the present embodiment.

The foregoing is merely an example of the present invention, and common general knowledge in the field of known specific structures and characteristics is not described herein in any greater extent than that known in the art at the filing date or prior to the priority date of the application, so that those skilled in the art can now appreciate that all of the above-described techniques in this field and have the ability to apply routine experimentation before this date can be combined with one or more of the present teachings to complete and implement the present invention, and that certain typical known structures or known methods do not pose any impediments to the implementation of the present invention by those skilled in the art. It should be noted that, for those skilled in the art, without departing from the structure of the present invention, several changes and modifications can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicability of the patent. The scope of the claims of the present application shall be determined by the contents of the claims, and the description of the embodiments and the like in the specification shall be used to explain the contents of the claims.