Temperature prediction method and device and server

Abstract

The invention provides a temperature prediction method and device and a server, and relates to a temperature control technology, and the method comprises the steps: obtaining to-be-predicted data and historical data at a current moment according to a preset time period; wherein the historical data comprises to-be-predicted data at the moment before the current moment, and the to-be-predicted data comprises the actual temperature value of the monitoring space at each moment. According to a preset temperature prediction model, performing temperature prediction on the to-be-predicted data at the current moment and the historical data to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training a plurality of historical to-be-predicted data based on a wavelet algorithm. According to the method, the technical problem that the temperature prediction accuracy of the prediction model is low is solved.

Description

Technical Field

[0001] The present application relates to temperature control technology, and in particular to a temperature prediction method, device and server. Background Art

[0002] At present, online temperature prediction is a common requirement of data centers and is often used in temperature control, energy-saving control, digital twins and other projects. Specifically, the common method of online temperature prediction is to train models such as neural networks and random forests based on the time domain characteristics of the data to be predicted, and predict the temperature based on the trained prediction model.

[0003] However, since inferior signals such as noise are mixed with the time domain features, the prediction process of the prediction model is more complicated if the model is trained only according to the time domain features of the data to be predicted, which may lead to lower prediction accuracy of the prediction model. Summary of the invention

[0004] The present application provides a temperature prediction method, device and server to solve the technical problem that the prediction model has low accuracy in predicting temperature.

[0005] In a first aspect, the present application provides a temperature prediction method, comprising:

[0006] Acquire the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at a moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment;

[0007] According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on a wavelet algorithm.

[0008] Further, the temperature prediction is performed on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model to obtain a predicted temperature value at the next moment, including:

[0009] According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data, and the temperature difference between the current moment and the next moment is predicted;

[0010] The predicted temperature difference is added to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

[0011] Furthermore, the data to be predicted also includes disturbance variable values ​​and / or control variable values ​​of the refrigeration equipment; wherein,

[0012] The disturbance variable value includes any one or all of the following:

[0013] Load power, outdoor temperature value outside the monitored space;

[0014] The control variable value includes any one or all of the following:

[0015] Air conditioning compressor speed, fan speed, water valve opening.

[0016] Furthermore, the method further comprises:

[0017] Performing increase and decrease processing on the control variable value in the to-be-predicted data at the current moment to generate a plurality of first variable values;

[0018] For each first variable value, according to a preset temperature prediction model, temperature prediction is performed on the first variable value at the current moment, the disturbance variable value and the historical data to generate a first predicted temperature value at the next moment;

[0019] According to the multiple first predicted temperature values ​​at the next moment, the first predicted temperature value with the smallest difference from the preset temperature value is determined as the target temperature value, and the first variable value corresponding to the smallest first predicted temperature value is determined as the target control variable and sent to the refrigeration equipment for execution.

[0020] Furthermore, the method further comprises:

[0021] Based on the preset time period, obtain the historical data to be predicted at each moment;

[0022] According to the historical data to be predicted at each moment, determine the state variable sequence and the disturbance variable sequence;

[0023] According to a preset wavelet algorithm, sample extraction processing is performed on the state variable sequence and the disturbance variable sequence respectively to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence;

[0024] Obtaining input variables according to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence, and the control variable value;

[0025] Determine the temperature difference between actual temperature values ​​at two adjacent moments, and use the temperature difference as output data;

[0026] The regression model is trained and updated according to the input variables and the output data to obtain a temperature prediction model.

[0027] Furthermore, the obtaining of historical data to be predicted at each moment based on a preset time period includes:

[0028] Based on a preset time period, the historical data to be predicted at each moment is obtained according to the temperature sensor of at least one temperature measurement point; wherein,

[0029] For each moment, if the number of the temperature measurement points is n, the actual temperature value in the historical data to be predicted is:

[0030]

[0031] n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t;

[0032] The disturbance variable value d t for:

[0033]

[0034] Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, wherein the disturbance variable value includes load power and outdoor temperature, Represents the wth disturbance variable value at the current time t.

[0035] Furthermore, the determining of the state variable sequence and the disturbance variable sequence according to the historical data to be predicted at each moment includes:

[0036] According to the historical data to be predicted at each moment, a state variable sequence about the actual temperature value and a disturbance variable sequence about the disturbance variable value are determined; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods.

[0037] s L =[s t-L+1 ,…,s t-1 ,s t ],

[0038] d L =[d t-L+1 ,…,d t-1 ,d t ],

[0039] Wherein, L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s tIndicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

[0040] Furthermore, the time-frequency features A of the state variable sequence include:

[0041] A=[a1,a2,…,a m ],

[0042] The time-frequency characteristics B of the disturbance variable sequence include:

[0043] B=[b1,b2,…,b m ],

[0044] m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m represents the mth time-frequency feature. Further, the input variable x t include:

[0045] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ].

[0046] Furthermore, the regression model is trained and updated according to the input variables and the output data to obtain a temperature prediction model, including:

[0047] The input variable x at time t t Input into the regression model for prediction to obtain the historical temperature difference at time t+1; wherein the temperature difference represents the output data at time t+1;

[0048] Add the historical temperature difference value to the actual temperature value at time t to obtain the predicted temperature value at time t+1;

[0049] The predicted temperature value at time t+1 is subtracted from the actual temperature value at time t+1 to obtain the prediction error value;

[0050] If it is determined that the prediction error value is greater than or equal to the preset critical value, the input variable x at time t is t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated according to the new samples to obtain a temperature prediction model;

[0051] If it is determined that the prediction error value is less than the preset critical value, the training and updating of the regression model is stopped until it is determined that the prediction error value is greater than or equal to the preset critical value, and then the training and updating of the regression model is continued.

[0052] In a second aspect, the present application provides a temperature prediction device, comprising:

[0053] A first acquisition unit is used to acquire the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment;

[0054] The first prediction unit is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on a wavelet algorithm.

[0055] Furthermore, the first prediction unit includes:

[0056] A first prediction module is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model, and predict the temperature difference between the current moment and the next moment;

[0057] The first adding module is used to add the predicted temperature difference value to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

[0058] Furthermore, the data to be predicted also includes disturbance variable values ​​and / or control variable values ​​of the refrigeration equipment; wherein,

[0059] The disturbance variable value includes any one or all of the following:

[0060] Load power, outdoor temperature value outside the monitored space;

[0061] The control variable value includes any one or all of the following:

[0062] Air conditioning compressor speed, fan speed, water valve opening.

[0063] Furthermore, the device also includes:

[0064] An increasing and decreasing unit, used for increasing and decreasing the control variable value in the data to be predicted at the current moment to generate a plurality of first variable values;

[0065] a second prediction unit, configured to perform temperature prediction on the first variable value at the current moment, the disturbance variable value, and the historical data according to a preset temperature prediction model for each first variable value, and generate a first predicted temperature value at the next moment;

[0066] The first determination unit is used to determine, based on the multiple first predicted temperature values ​​at the next moment, the first predicted temperature value with the smallest difference from the preset temperature value as the target temperature value, and to determine the first variable value corresponding to the smallest first predicted temperature value as the target control variable, and send it to the refrigeration equipment for execution.

[0067] Furthermore, the device also includes:

[0068] A second acquisition unit is used to acquire the historical data to be predicted at each moment based on a preset time period;

[0069] A second determination unit is used to determine a state variable sequence and a disturbance variable sequence according to the historical data to be predicted at each moment;

[0070] An extraction unit, used to perform sample extraction processing on the state variable sequence and the disturbance variable sequence respectively according to a preset wavelet algorithm to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence;

[0071] A third determining unit, configured to obtain an input variable according to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence, and a control variable value;

[0072] A fourth determining unit, used to determine a temperature difference between actual temperature values ​​at two adjacent moments, and use the temperature difference as output data;

[0073] The training and updating unit is used to train and update the regression model according to the input variables and the output data to obtain a temperature prediction model.

[0074] Furthermore, the second acquisition unit is specifically used to:

[0075] Based on a preset time period, the historical data to be predicted at each moment is obtained according to the temperature sensor of at least one temperature measurement point; wherein,

[0076] For each moment, if the number of the temperature measurement points is n, the actual temperature value in the historical data to be predicted is:

[0077]

[0078] n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t;

[0079] The disturbance variable value d t for:

[0080]

[0081] Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, wherein the disturbance variable value includes load power and outdoor temperature, Represents the wth disturbance variable value at the current time t.

[0082] Furthermore, the first determining unit is specifically configured to:

[0083] According to the historical data to be predicted at each moment, a state variable sequence about the actual temperature value and a disturbance variable sequence about the disturbance variable value are determined; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods.

[0084] s L =[s t-L+1 ,…,s t-1 ,s t ],

[0085] d L =[d t-L+1 ,…,d t-1 ,d t ],

[0086] Wherein, L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

[0087] Furthermore, the time-frequency features A of the state variable sequence include:

[0088] A=[a1,a2,…,a m ],

[0089] The time-frequency characteristics B of the disturbance variable sequence include:

[0090] B=[b1,b2,…,b m ],

[0091] m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m Represents the mth time-frequency feature.

[0092] Furthermore, the input variable x t include:

[0093] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ].

[0094] Furthermore, the training update unit includes:

[0095] The second prediction module is used to transform the input variable x at time t t Input into the regression model for prediction to obtain the historical temperature difference at time t+1; wherein the temperature difference represents the output data at time t+1;

[0096] A second adding module, used for adding the historical temperature difference value to the actual temperature value at time t to obtain a predicted temperature value at time t+1;

[0097] A difference module is used to make a difference between the predicted temperature value at time t+1 and the actual temperature value at time t+1 to obtain a prediction error value;

[0098] The training update module is used to update the input variable x at time t if it is determined that the prediction error value is greater than or equal to a preset critical value. t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated according to the new samples to obtain a temperature prediction model;

[0099] The stopping module is used to stop training and updating the regression model if it is determined that the prediction error value is less than the preset critical value, and continue training and updating the regression model until it is determined that the prediction error value is greater than or equal to the preset critical value.

[0100] In a third aspect, the present application provides a server, comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, and when the processor executes the computer program, the method described in the first aspect is implemented.

[0101] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the method described in the first aspect.

[0102] In a fifth aspect, the present application provides a computer program product, including a computer program, which implements the method described in the first aspect when executed by a processor.

[0103] The present application provides a temperature prediction method, device and server, which obtain the data to be predicted and historical data at the current moment according to a preset time period; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment. According to the preset temperature prediction model, the data to be predicted and the historical data at the current moment are temperature predicted to obtain the predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on the wavelet algorithm. In this scheme, the data to be predicted and the historical data at the current moment are obtained according to the preset time period, and the temperature prediction is performed on the data to be predicted and the historical data at the current moment according to the preset temperature prediction model to obtain the predicted temperature value at the next moment. Therefore, the temperature prediction model is obtained based on the wavelet algorithm training, and the temperature is predicted online according to the temperature prediction model, so that the wavelet algorithm is used to analyze and predict the data to be predicted in the prediction process, which is conducive to giving play to the role of high-value frequency signals in the data to be predicted, and is conducive to eliminating the interference of low-value frequency signals, simplifying the prediction process, improving the prediction accuracy of the model, and solving the technical problem of low accuracy of temperature prediction by the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0104] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0105] Figure 1 A schematic diagram of a temperature prediction method provided in an embodiment of the present application;

[0106] Figure 2 A schematic diagram of another temperature prediction method provided in an embodiment of the present application;

[0107] Figure 3 A schematic diagram of a flow chart of another temperature prediction method provided in an embodiment of the present application;

[0108] Figure 4 A schematic diagram of the structure of a temperature prediction device provided in an embodiment of the present application;

[0109] Figure 5A schematic diagram of the structure of another temperature prediction device provided in an embodiment of the present application;

[0110] Figure 6 A schematic diagram of the structure of a server provided in an embodiment of the present application.

[0111] The above drawings show clear embodiments of the present disclosure, which will be described in more detail below. These drawings and text descriptions are not intended to limit the scope of the present disclosure in any way, but to illustrate the concepts of the present disclosure to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0112] Here, exemplary embodiments will be described in detail, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present disclosure.

[0113] At present, online temperature prediction is a common requirement of data centers and is often used in temperature control, energy-saving control, digital twins and other projects. Specifically, the common method of online temperature prediction is to train models such as neural networks and random forests based on the time domain characteristics of the data to be predicted, and predict the temperature based on the trained prediction model. However, since inferior signals such as noise are mixed with time domain characteristics, the prediction process of the prediction model is more complicated if the model is trained only based on the time domain characteristics of the data to be predicted, which may result in low accuracy of the prediction model.

[0114] The present application provides a temperature prediction method, device and server, which are intended to solve the above technical problems in the prior art.

[0115] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0116] Figure 1 A flow chart of a temperature prediction method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes:

[0117] Step 101, obtaining the data to be predicted at the current moment and historical data according to a preset time period; wherein the historical data includes the data to be predicted at a moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment.

[0118] Exemplarily, the executor of this embodiment can be a server. First, the server monitors and adjusts the temperature in the monitoring space. The monitoring space includes cabinets, loads, air conditioners, temperature sensors, etc. The cabinet refers to the bracket used to place the load; the load refers to the server, computer and other equipment running in the monitoring space, for example, the load refers to multiple servers in the computer room; the air conditioner is used to dissipate heat to maintain the temperature in the monitoring space at a suitable temperature; the temperature sensor can be installed on the front and back doors of each cabinet, and the number of installations is not limited. The preset time period is a pre-set time interval for collecting the data to be predicted. The time interval can be selected as needed. For example, for long-term temperature prediction tasks, 5 minutes can be taken; for device-level control tasks (air conditioning control tasks usually require temperature prediction to select a better control configuration or action), in order to ensure the timeliness of control, 10 seconds can be taken, and there is no limit on this.

[0119] In this step, the server obtains the data to be predicted at the current moment and the historical data according to the preset time period, and the historical data includes the data to be predicted at the moment before the current moment. The data to be predicted includes the actual temperature value of the monitoring space at each moment. Specifically, the data to be predicted includes the disturbance variable value and / or the control variable value u of the refrigeration equipment. t The disturbance variable value includes any one or all of the following: load power, the outdoor temperature value outside the monitored space; the control variable value u t Includes any or all of the following: air conditioning compressor speed, fan speed, water valve opening.

[0120] Step 102: According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training a plurality of historical data to be predicted based on a wavelet algorithm.

[0121] Exemplarily, the preset temperature prediction model is obtained by training a plurality of historical data to be predicted based on the wavelet algorithm. According to the preset temperature prediction model, the server performs temperature prediction on the data to be predicted at the current moment and the historical data to obtain the temperature difference between the current moment and the next moment. The temperature difference is added to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

[0122] In the embodiment of the present application, the data to be predicted and the historical data at the current moment are obtained according to the preset time period; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment. According to the preset temperature prediction model, the data to be predicted and the historical data at the current moment are temperature predicted to obtain the predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on the wavelet algorithm. In this scheme, the data to be predicted and the historical data at the current moment are obtained according to the preset time period, and the temperature prediction is performed on the data to be predicted and the historical data at the current moment according to the preset temperature prediction model to obtain the predicted temperature value at the next moment. Therefore, the temperature prediction model is obtained based on the wavelet algorithm training, and the temperature is predicted online according to the temperature prediction model, so that the wavelet algorithm is used to analyze and predict the data to be predicted in the prediction process, which is conducive to giving play to the role of high-value frequency signals in the data to be predicted, and is conducive to eliminating the interference of low-value frequency signals, simplifying the prediction process, improving the prediction accuracy of the model, and solving the technical problem of low accuracy of temperature prediction by the prediction model.

[0123] Figure 2 A flow chart of another temperature prediction method provided in an embodiment of the present application is shown as follows: Figure 2 As shown, the method includes:

[0124] Step 201: Based on a preset time period, obtain historical data to be predicted at each moment.

[0125] In one example, the data to be predicted also includes disturbance variable values ​​and / or control variable values ​​of the refrigeration equipment; wherein the disturbance variable values ​​include any one or all of the following: load power, outdoor temperature value outside the monitored space; the control variable values ​​include any one or all of the following: air conditioning compressor speed, fan speed, water valve opening.

[0126] In one example, step 201 includes: based on a preset time period, obtaining historical data to be predicted at each moment according to a temperature sensor at at least one temperature measurement point; wherein,

[0127] For each moment, if the number of temperature measurement points is n, the actual temperature value in the historical data to be predicted is:

[0128]

[0129] n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t;

[0130] Disturbance variable value d t for:

[0131]

[0132] Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, which includes load power and outdoor temperature. Represents the wth disturbance variable value at the current time t.

[0133] For example, based on a preset time period, the server obtains the historical data to be predicted at each moment according to multiple temperature sensors. Specifically, the historical data to be predicted includes the actual temperature value in the monitoring space, the disturbance variable value, and the control variable value u of the refrigeration equipment. t , control variable value u t Including any of the following: air conditioning compressor speed, fan speed, water valve opening; disturbance variable value includes any or all of the following: load power, outdoor temperature value outside the monitored space.

[0134] For each moment, if the number of temperature measurement points is n, the server obtains n actual temperature values ​​at that moment. The actual temperature value in the historical data to be predicted is:

[0135]

[0136] n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Indicates the first actual temperature value at the current time t, and so on. Represents the nth actual temperature value at the current time t.

[0137] Disturbance variable value d t for:

[0138]

[0139] Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, which includes load power and outdoor temperature. represents the value of the first disturbance variable at the current time t, and so on. Represents the wth disturbance variable value at the current time t.

[0140] Step 202: Determine the state variable sequence and the disturbance variable sequence according to the historical data to be predicted at each moment.

[0141] In one example, step 202 includes: determining a state variable sequence about the actual temperature value and a disturbance variable sequence about the disturbance variable value according to the historical data to be predicted at each moment; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods.

[0142] s L =[s t-L+1 ,…,s t-1 ,s t ],

[0143] d L =[d t-L+1 ,…,d t-1 ,d t ],

[0144] Where L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

[0145] For example, the server may summarize the historical data to be predicted at each moment to determine the state variable sequence about the actual temperature value and the disturbance variable sequence about the disturbance variable value. L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods.

[0146] s L =[s t-L+1 ,…,s t-1 ,s t ],

[0147] d L =[d t-L+1 ,…,d t-1 ,d t ],

[0148] Where L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, and so on, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1represents the disturbance variable value at time t-L+1, and so on, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

[0149] Step 203: According to a preset wavelet algorithm, sample extraction processing is performed on the state variable sequence and the disturbance variable sequence respectively to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence.

[0150] In one example, the time-frequency features A of the state variable sequence include:

[0151] A=[a1,a2,…,a m ],

[0152] The time-frequency characteristics B of the disturbance variable sequence include:

[0153] B=[b1,b2,…,b m ],

[0154] m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m Represents the mth time-frequency feature.

[0155] Exemplarily, the server can use a wavelet algorithm for feature extraction. There are obvious technical advantages in analyzing and decomposing sequence data into time-frequency signals through a wavelet algorithm. First, noise is usually a relatively high-frequency or relatively low-frequency signal, which is easy to separate in the frequency domain and thus easy to remove; secondly, after the signals of different frequencies are separated, it is beneficial for machine learning to learn higher-value frequency signals. Therefore, the server can perform sample extraction processing on the state variable sequence and the disturbance variable sequence according to a preset wavelet algorithm to obtain the time-frequency feature A of the state variable sequence and the time-frequency feature B of the disturbance variable sequence. The time-frequency feature A and the time-frequency feature B can both be low-frequency, or both can be high-frequency, or one can be low-frequency and the other high-frequency, without limitation. Specifically,

[0156] The time-frequency characteristics A of the state variable sequence include:

[0157] A=[a1,a2,…,a m ],

[0158] The time-frequency characteristics B of the disturbance variable sequence include:

[0159] B=[b1,b2,…,b m ],

[0160] Among them, m is a positive integer greater than 1, a1 represents the first time-frequency feature, and so on, a mrepresents the mth time-frequency feature; b1 represents the 1st time-frequency feature, and so on. m Represents the mth time-frequency feature.

[0161] Step 204: According to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence and the control variable value, the input variable is obtained. In one example, the input variable x t include:

[0162] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ].

[0163] For example, the server can store the time-frequency features A of the state variable sequence, the time-frequency features B of the disturbance variable sequence, and the control variable value u t Together as input, we get the input variable x t ,

[0164] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ].

[0165] Step 205: determine the temperature difference between the actual temperature values ​​at two adjacent moments, and use the temperature difference as output data.

[0166] For example, the server calculates the temperature difference between the actual temperature values ​​at two adjacent moments and stores the temperature difference y t As output data,

[0167] y t =Δ t =s t+1 -s t

[0168] Step 206: Train and update the regression model based on the input variables and output data to obtain a temperature prediction model.

[0169] In one example, step 206 includes: transforming the input variable x at time t into t Input the regression model for prediction and obtain the historical temperature difference at time t+1; the temperature difference represents the output data at time t+1; add the historical temperature difference to the actual temperature value at time t to obtain the predicted temperature value at time t+1; make a difference between the predicted temperature value at time t+1 and the actual temperature value at time t+1 to obtain the prediction error value; if it is determined that the prediction error value is greater than or equal to the preset critical value, then the input variable x at time t is tAs input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated according to the new samples to obtain a temperature prediction model; if it is determined that the prediction error value is less than the preset critical value, the training and updating of the regression model is stopped until it is determined that the prediction error value is greater than or equal to the preset critical value, and then the training and updating of the regression model continues.

[0170] For example, the server sends the input variable x at time t t Input the regression model for prediction and obtain the historical temperature difference at time t+1. The temperature difference represents the output data at time t+1. Add the historical temperature difference to the actual temperature value at time t to obtain the predicted temperature value at time t+1. Difference the predicted temperature value at time t+1 from the actual temperature value at time t+1 to obtain the prediction error value, and determine whether the prediction error value is greater than or equal to the preset critical value, where the preset critical value can be pre-defined by the user. If it is determined that the prediction error value is greater than or equal to the preset critical value, it means that the current sample is a valid sample, and the input variable x at time t is changed to t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of valid new samples, and the regression model is trained and updated based on the new samples to obtain a temperature prediction model; if it is determined that the prediction error value is less than the preset critical value, it means that the sample is invalid, and the training and updating of the regression model is stopped until it is determined that the prediction error value is greater than or equal to the preset critical value, and the training and updating of the regression model is continued. By analogy, in the subsequent prediction process of the temperature prediction model, the temperature prediction model is still continuously corrected according to the prediction error value greater than or equal to the critical value to ensure the prediction accuracy of the temperature prediction model.

[0171] Therefore, the machine learning regression model is used to periodically train and learn at certain time intervals, and the parameters of the temperature prediction model are continuously corrected to obtain the temperature prediction model, and the temperature at the next moment t+1 is predicted based on the states at L moments [t-L+1, ..., t-1, t]. It should be noted that the machine learning regression model can be a linear model, or an SVM regression model, etc., and there is no limitation on this.

[0172] Step 207, obtaining the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at a moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment.

[0173] For example, this step can be referred to Figure 1 Step 101 in the above is not described in detail.

[0174] Step 208: According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data, and the temperature difference between the current moment and the next moment is predicted.

[0175] Exemplarily, the preset temperature prediction model is obtained by training multiple historical data to be predicted based on the wavelet algorithm. According to the preset temperature prediction model, the server performs temperature prediction on the data to be predicted at the current moment and the historical data to obtain the temperature difference between the current moment and the next moment.

[0176] Step 209: Add the predicted temperature difference to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

[0177] Exemplarily, the server may add the temperature difference to the actual temperature value at the current moment to obtain a predicted temperature value at the next moment.

[0178] Step 210: Increase or decrease the control variable value in the data to be predicted at the current moment to generate a plurality of first variable values.

[0179] Exemplarily, the server performs increase and decrease processing on the control variable value in the data to be predicted at the current moment to generate multiple first variable values. For example, the control variable value is subtracted from a preset value to obtain the first first variable value, the control variable value is added to the preset value to obtain the second first variable value; the first first variable value is subtracted from the preset value again to obtain the third first variable value, the second first variable value is added to the preset value again to obtain the fourth first variable value; and so on, multiple first variable values ​​can be obtained.

[0180] Step 211: for each first variable value, perform temperature prediction on the first variable value at the current moment, the disturbance variable value and historical data according to a preset temperature prediction model to generate a first predicted temperature value at the next moment.

[0181] Exemplarily, for each first variable value, the server performs temperature prediction on the first variable value, the disturbance variable value, and the historical data at the current moment according to a preset temperature prediction model, and generates a first predicted temperature value at the next moment. Therefore, multiple first predicted temperature values ​​at the next moment can be obtained under multiple first variable values.

[0182] Step 212: determine the first predicted temperature value with the smallest difference from the preset temperature value based on multiple first predicted temperature values ​​at the next moment, and determine the first variable value corresponding to the smallest first predicted temperature value as the target control variable, and send it to the refrigeration equipment for execution.

[0183] Exemplarily, the preset temperature value refers to the temperature of the refrigeration equipment that is preset. The server can determine the difference between each first predicted temperature value and the preset temperature value among multiple first predicted temperature values ​​at the next moment, determine the first predicted temperature value with the smallest difference with the preset temperature value, and determine the first variable value corresponding to the smallest first predicted temperature value as the target control variable, and send it to the refrigeration equipment for execution. After receiving the target control variable, the refrigeration equipment adjusts its own target control variable by changing the speed of the air-conditioning compressor and other operations, so that the refrigeration equipment dissipates heat according to the adjusted control variable value, and makes the temperature value of the monitored space at the next moment closest to the actual temperature value at the current moment. By analogy, the target control variable at the nth moment can be obtained.

[0184] Alternatively, when the server determines the predicted temperature value at the next moment (t+1) and the target control variable corresponding to the predicted temperature value, it can further use the temperature data at the next moment (t+1) as the current moment data, increase or decrease the first variable value, generate multiple second variable values, perform temperature prediction on the second variable value at the next moment (t+1), the disturbance variable value and the historical data, predict and determine the predicted temperature value at the next moment (t+2), and so on, to obtain n predicted temperature values. Then, based on the predicted temperature value at the next moment (t+1), the predicted temperature value at the next moment (t+2)...n predicted temperature values, the predicted temperature value at the next moment (t+1) is comprehensively determined. Among them, the predicted temperature value at the next moment (t+1), the predicted temperature value at the next moment (t+2),..., the comprehensive deviation of the nth predicted temperature value from the preset temperature value is calculated, and the calculation method includes: the weighted sum of the absolute deviations of each predicted temperature value from the preset temperature value, or the weighted sum of the squares of the deviations. According to the minimum value of the comprehensive deviation, the corresponding first variable value is selected as the control variable and sent to the refrigeration equipment for execution.

[0185] In the embodiment of the present application, based on a preset time period, historical data to be predicted at each moment is obtained. According to the historical data to be predicted at each moment, a state variable sequence and a disturbance variable sequence are determined. According to a preset wavelet algorithm, sample extraction processing is performed on the state variable sequence and the disturbance variable sequence respectively to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence. According to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence and the control variable value, an input variable is obtained. The temperature difference between the actual temperature values ​​at two adjacent moments is determined, and the temperature difference is used as output data. According to the input variables and the output data, the regression model is trained and updated to obtain a temperature prediction model. According to a preset time period, the data to be predicted and the historical data at the current moment are obtained; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment. According to the preset temperature prediction model, the data to be predicted and the historical data at the current moment are temperature predicted, and the temperature difference between the current moment and the next moment is predicted. The predicted temperature difference is added to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment. The control variable value in the data to be predicted at the current moment is increased or decreased to generate multiple first variable values. For each first variable value, the temperature is predicted for the first variable value, the disturbance variable value and the historical data at the current moment according to the preset temperature prediction model to generate the first predicted temperature value at the next moment. According to the multiple first predicted temperature values ​​at the next moment, the first predicted temperature value with the smallest difference with the preset temperature value is determined as the target temperature value, and the first variable value corresponding to the smallest first predicted temperature value is determined as the target control variable and sent to the refrigeration equipment for execution. Therefore, a temperature prediction model is obtained based on wavelet algorithm training, and temperature is predicted online according to the temperature prediction model, so that the wavelet algorithm is used to analyze and predict the data to be predicted during the prediction process, which is conducive to giving play to the role of high-value frequency signals in the data to be predicted, and is conducive to eliminating the interference of low-value frequency signals, simplifying the prediction process, improving the prediction accuracy of the model, and solving the technical problem of low accuracy of temperature prediction by the prediction model. In addition, the wavelet algorithm is combined with the machine learning regression model to train a temperature prediction model. The temperature prediction model is used to perform online prediction of the temperature under different control variables in the data center based on the time-frequency domain characteristics. The wavelet algorithm is used to decompose the time-frequency of the sequence data and extract the time-frequency domain features, which is beneficial to giving full play to the role of high-value frequency signals and eliminating the interference of low-value frequency signals, further improving the accuracy of the model in predicting temperature.

[0186] Figure 3 A flow chart of another temperature prediction method provided in an embodiment of the present application is shown as follows: Figure 3As shown, the method includes: collecting data to be predicted; extracting time-frequency features; extracting samples; regression model; judging the validity of samples; if the sample is determined to be valid, training the regression model; if the sample is determined to be invalid, determining it as an invalid sample; and performing temperature prediction through the trained temperature prediction model.

[0187] In one example, taking an air-cooled data center as an example, the data to be predicted of the air-cooled data center is collected, and the temperature at the next moment is predicted. Specifically,

[0188] 1) Collect the data center temperature, load power, compressor capacity, and fan speed at a certain period. Among them, temperature is a state variable, denoted by s t ; Load power is the disturbance variable value, denoted as d t ; The fan speed is the control variable value, denoted as u t .

[0189] 2) Extract features from the collected data and use the SVM regression model to predict the temperature at time t+1 based on the state of the data center at L times [t-L+1,…,t-1,t].

[0190] 3) Use the Haar wavelet algorithm to extract time-frequency features and extract samples:

[0191] i. Extract state variable sequence [s t-L+1 ,…,s t-1 ,s t ]’s time-frequency features A=[a1,a2,…,a m ],

[0192] ii. Extract the disturbance variable sequence d t-L+1 ,…,d t-1 ,d t ]’s time-frequency features B=[b1,b2,…,b m ],

[0193] iii. Combine the time-frequency features A and B with the control variable value u t Together as input variables:

[0194] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ],

[0195] iv. Calculate the temperature change between time t and time t+1 and output it as:

[0196] y t =Δ t =s t+1 -st ,

[0197] Among them, x t ,y t A pair of samples is formed, and new samples can be continuously generated as time t increases.

[0198] 4) Use SVM regression model for temperature fitting and prediction:

[0199] i. Training: For each time step t, the regression model is trained based on the input x t Predict the temperature difference at the next moment t+1, and record the predicted temperature difference as y t ′, then the predicted temperature value at the next moment t+1 Based on the actual temperature value corresponding to t+1 in the sample generated in the previous step, the prediction error value between the predicted temperature value at the next moment t+1 and the actual temperature value at the next moment t+1 is calculated. When the prediction error value bias≥δ, the regression model is trained and updated, where δ is the preset critical value. When the prediction error value bias<δ, no training update is performed to prevent the steady-state data from destroying the model. The prediction error uses the absolute error sum of different temperature measurement points at the same time. The prediction error can also use the square sum of the errors of different temperature measurement points at the same time, and there is no limitation on this.

[0200] ii. Prediction: For each time t, the model is classified according to the input x t Predict the future temperature, and the predicted temperature difference is recorded as y t ′, then the predicted temperature value is

[0201] 5) The load power in this article can be replaced by the load factor percentage.

[0202] 6) When performing wavelet decomposition and feature extraction according to the wavelet algorithm, specific frequencies can be selected for use. For example, removing the high-frequency part is helpful to completely eliminate the influence of noise, and model training is performed based on the low-frequency part, thereby simplifying the prediction process of the trained model.

[0203] Figure 4 A schematic diagram of the structure of a temperature prediction device provided in an embodiment of the present application is shown in FIG. Figure 4 As shown, the device comprises:

[0204] The first acquisition unit 31 is used for the first acquisition unit, and is used to obtain the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment.

[0205] The first prediction unit 32 is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training a plurality of historical data to be predicted based on a wavelet algorithm.

[0206] The device of this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principles are the same and will not be repeated here.

[0207] Figure 5 A schematic diagram of the structure of another temperature prediction device provided in an embodiment of the present application, Figure 4 Based on the embodiment shown, Figure 5 As shown, the first prediction unit 32 includes:

[0208] The first prediction module 321 is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model, and predict the temperature difference between the current moment and the next moment.

[0209] The first adding module 322 is used to add the predicted temperature difference value to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

[0210] In one example, the data to be predicted also includes disturbance variable values ​​and / or control variable values ​​of the refrigeration equipment; wherein,

[0211] The disturbance variable value includes any or all of the following:

[0212] Load power, outdoor temperature value outside the monitored space;

[0213] Control variable values ​​include any or all of the following:

[0214] Air conditioning compressor speed, fan speed, water valve opening.

[0215] In one example, the device further includes:

[0216] The increasing and decreasing unit 41 is used to increase and decrease the control variable value in the data to be predicted at the current moment to generate a plurality of first variable values.

[0217] The second prediction unit 42 is used to perform temperature prediction on the first variable value, the disturbance variable value and the historical data at the current moment for each first variable value according to a preset temperature prediction model, and generate a first predicted temperature value at the next moment.

[0218] The first determination unit 43 is used to determine the first predicted temperature value with the smallest difference from the preset temperature value based on multiple first predicted temperature values ​​at the next moment, and determine the first variable value corresponding to the smallest first predicted temperature value as the target control variable, and send it to the refrigeration equipment for execution.

[0219] In one example, the device further includes:

[0220] The second acquisition unit 44 is used to acquire the historical data to be predicted at each moment based on a preset time period.

[0221] The second determining unit 45 is used to determine the state variable sequence and the disturbance variable sequence according to the historical data to be predicted at each moment.

[0222] The extraction unit 46 is used to perform sample extraction processing on the state variable sequence and the disturbance variable sequence respectively according to a preset wavelet algorithm to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence.

[0223] The third determining unit 47 is used to obtain the input variable according to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence and the control variable value.

[0224] The fourth determining unit 48 is used to determine the temperature difference between the actual temperature values ​​at two adjacent moments, and use the temperature difference as output data.

[0225] The training and updating unit 49 is used to train and update the regression model according to the input variables and the output data to obtain a temperature prediction model.

[0226] In one example, the second obtaining unit 44 is specifically configured to:

[0227] Based on a preset time period, the historical data to be predicted at each moment is obtained according to the temperature sensor of at least one temperature measurement point; wherein,

[0228] For each moment, if the number of temperature measurement points is n, the actual temperature value in the historical data to be predicted is:

[0229]

[0230] n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t;

[0231] Disturbance variable value d t for:

[0232]

[0233] Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, which includes load power and outdoor temperature. Represents the wth disturbance variable value at the current time t.

[0234] In one example, the second determining unit 45 is specifically configured to:

[0235] According to the historical data to be predicted at each moment, the state variable sequence about the actual temperature value and the disturbance variable sequence about the disturbance variable value are determined; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods.

[0236] s L =[s t-L+1 ,…,s t-1 ,s t ],

[0237] d L =[d t-L+1 ,…,d t-1 ,d t ],

[0238] Where L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

[0239] In one example, the time-frequency features A of the state variable sequence include:

[0240] A=[a1,a2,…,a m ],

[0241] The time-frequency characteristics B of the disturbance variable sequence include:

[0242] B=[b1,b2,…,b m ],

[0243] m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m Represents the mth time-frequency feature.

[0244] In an example, the input variable x t include:

[0245] x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ].

[0246] In one example, the training update unit 49 includes:

[0247] The second prediction module 491 is used to convert the input variable x at time t into t The input is made into the regression model for prediction, and the historical temperature difference at time t+1 is obtained; wherein the temperature difference represents the output data at time t+1.

[0248] The second adding module 492 is used to add the historical temperature difference value to the actual temperature value at time t to obtain the predicted temperature value at time t+1.

[0249] The difference module 493 is used to make a difference between the predicted temperature value at time t+1 and the actual temperature value at time t+1 to obtain a prediction error value.

[0250] The training update module 494 is used to update the input variable x at time t if it is determined that the prediction error value is greater than or equal to a preset critical value. t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated based on the new samples to obtain a temperature prediction model.

[0251] The stop module 495 is used to stop training and updating the regression model if it is determined that the prediction error value is less than a preset critical value, and continue training and updating the regression model until it is determined that the prediction error value is greater than or equal to the preset critical value.

[0252] The device of this embodiment can execute the technical solution in the above method. Its specific implementation process and technical principles are the same and will not be repeated here.

[0253] Figure 6 A schematic diagram of the structure of a server provided in an embodiment of the present application, such as Figure 6 As shown, the server includes: a memory 51 and a processor 52.

[0254] The memory 51 stores a computer program that can be executed on the processor 52 .

[0255] The processor 52 is configured to execute the method provided in the above embodiment.

[0256] The server further includes a receiver 53 and a transmitter 54. The receiver 53 is used to receive instructions and data sent by an external device, and the transmitter 54 is used to send instructions and data to the external device.

[0257] The embodiment of the present application also provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the server, the server can execute the method provided by the above embodiment.

[0258] An embodiment of the present application also provides a computer program product, which includes: a computer program, which is stored in a readable storage medium, and at least one processor of the server can read the computer program from the readable storage medium, and at least one processor executes the computer program so that the server executes the solution provided by any of the above embodiments.

[0259] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0260] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A temperature prediction method, characterized in that: include: Acquire the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at a moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment; According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on a wavelet algorithm.

2. The method according to claim 1, characterized in that The step of performing temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model to obtain a predicted temperature value at the next moment includes: According to a preset temperature prediction model, temperature prediction is performed on the data to be predicted at the current moment and the historical data, and the temperature difference between the current moment and the next moment is predicted; The predicted temperature difference is added to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

3. The method according to claim 1, characterized in that The data to be predicted also includes disturbance variable values, and / or control variable values ​​of the refrigeration equipment; wherein, The disturbance variable value includes any one or all of the following: Load power, outdoor temperature value outside the monitored space; The control variable value includes any one or all of the following: Air conditioning compressor speed, fan speed, water valve opening.

4. The method according to claim 3, characterized in that The method further comprises: Performing increase and decrease processing on the control variable value in the to-be-predicted data at the current moment to generate a plurality of first variable values; For each first variable value, according to a preset temperature prediction model, temperature prediction is performed on the first variable value at the current moment, the disturbance variable value and the historical data to generate a first predicted temperature value at the next moment; According to the multiple first predicted temperature values ​​at the next moment, the first predicted temperature value with the smallest difference from the preset temperature value is determined, and the first variable value corresponding to the smallest first predicted temperature value is determined as the target control variable and sent to the refrigeration equipment for execution.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Based on the preset time period, obtain the historical data to be predicted at each moment; According to the historical data to be predicted at each moment, determine the state variable sequence and the disturbance variable sequence; According to a preset wavelet algorithm, sample extraction processing is performed on the state variable sequence and the disturbance variable sequence respectively to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence; Obtaining input variables according to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence, and the control variable value; Determine the temperature difference between actual temperature values ​​at two adjacent moments, and use the temperature difference as output data; The regression model is trained and updated according to the input variables and the output data to obtain a temperature prediction model.

6. The method according to claim 5, characterized in that The step of obtaining the historical data to be predicted at each moment based on a preset time period includes: Based on a preset time period, the historical data to be predicted at each moment is obtained according to the temperature sensor of at least one temperature measurement point; wherein, For each moment, if the number of the temperature measurement points is n, the actual temperature value in the historical data to be predicted is: n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t; The disturbance variable value d t for: Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, wherein the disturbance variable value includes load power and outdoor temperature, Represents the wth disturbance variable value at the current time t.

7. The method according to claim 6, characterized in that The step of determining the state variable sequence and the disturbance variable sequence according to the historical data to be predicted at each moment includes: According to the historical data to be predicted at each moment, a state variable sequence about the actual temperature value and a disturbance variable sequence about the disturbance variable value are determined; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods. s L =[s t-L+1 ,…,s t-1 ,s t ], d L =[d t-L+1 ,…,d t-1 ,d t ], Wherein, L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

8. The method according to claim 7, characterized in that The time-frequency features A of the state variable sequence include: <h2 style=";text-align:left;direction:ltr">A=[a1,a2,…,a<h2 style=";text-align:left;direction:ltr"> m <h2 style=";text-align:left;direction:ltr"> ], The time-frequency characteristics B of the disturbance variable sequence include: B=[b1,b2,…,b m ], m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m Represents the mth time-frequency feature.

9. The method according to claim 8, characterized in that The input variable x t include: x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ]。 10. The method according to claim 9, characterized in that The step of training and updating the regression model according to the input variables and the output data to obtain the temperature prediction model includes: The input variable x at time t t Input into the regression model for prediction to obtain the historical temperature difference at time t+1; wherein the temperature difference represents the output data at time t+1; Add the historical temperature difference value to the actual temperature value at time t to obtain the predicted temperature value at time t+1; The predicted temperature value at time t+1 is subtracted from the actual temperature value at time t+1 to obtain the prediction error value; If it is determined that the prediction error value is greater than or equal to the preset critical value, the input variable x at time t is t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated according to the new samples to obtain a temperature prediction model; If it is determined that the prediction error value is less than the preset critical value, the training and updating of the regression model is stopped until it is determined that the prediction error value is greater than or equal to the preset critical value, and then the training and updating of the regression model is continued.

11. A temperature prediction device, characterized in that: include: A first acquisition unit is used to acquire the data to be predicted at the current moment and the historical data according to a preset time period; wherein the historical data includes the data to be predicted at the moment before the current moment, and the data to be predicted includes the actual temperature value of the monitoring space at each moment; The first prediction unit is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model to obtain a predicted temperature value at the next moment; wherein the preset temperature prediction model is obtained by training multiple historical data to be predicted based on a wavelet algorithm.

12. The device according to claim 11, characterized in that The first prediction unit comprises: A first prediction module is used to perform temperature prediction on the data to be predicted at the current moment and the historical data according to a preset temperature prediction model, and predict the temperature difference between the current moment and the next moment; The first adding module is used to add the predicted temperature difference value to the actual temperature value at the current moment to obtain the predicted temperature value at the next moment.

13. The device according to claim 11, characterized in that The data to be predicted also includes disturbance variable values, and / or control variable values ​​of the refrigeration equipment; wherein, The disturbance variable value includes any one or all of the following: Load power, outdoor temperature value outside the monitored space; The control variable value includes any one or all of the following: Air conditioning compressor speed, fan speed, water valve opening.

14. The device according to claim 13, characterized in that The device also includes: An increasing and decreasing unit, used for increasing and decreasing the control variable value in the data to be predicted at the current moment to generate a plurality of first variable values; a second prediction unit, configured to perform temperature prediction on the first variable value at the current moment, the disturbance variable value and the historical data according to a preset temperature prediction model for each first variable value, and generate a first predicted temperature value at the next moment; The first determination unit is used to determine the first predicted temperature value with the smallest difference from the preset temperature value as the target temperature value based on the multiple first predicted temperature values ​​at the next moment, and determine the first variable value corresponding to the smallest first predicted temperature value as the target control variable, and send it to the refrigeration equipment for execution.

15. The device according to any one of claims 11 to 14, characterized in that: The device also includes: A second acquisition unit is used to acquire the historical data to be predicted at each moment based on a preset time period; A second determination unit is used to determine a state variable sequence and a disturbance variable sequence according to the historical data to be predicted at each moment; An extraction unit, used to perform sample extraction processing on the state variable sequence and the disturbance variable sequence respectively according to a preset wavelet algorithm to obtain the time-frequency characteristics of the state variable sequence and the time-frequency characteristics of the disturbance variable sequence; A third determining unit, configured to obtain an input variable according to the time-frequency characteristics of the state variable sequence, the time-frequency characteristics of the disturbance variable sequence, and a control variable value; A fourth determining unit, used to determine a temperature difference between actual temperature values ​​at two adjacent moments, and use the temperature difference as output data; The training and updating unit is used to train and update the regression model according to the input variables and the output data to obtain a temperature prediction model.

16. The device according to claim 15, characterized in that The second acquisition unit is specifically used to: Based on a preset time period, the historical data to be predicted at each moment is obtained according to the temperature sensor of at least one temperature measurement point; wherein, For each moment, if the number of the temperature measurement points is n, the actual temperature value in the historical data to be predicted is: n is a positive integer greater than or equal to 1, t represents the current time, s t represents the actual temperature value at the current time t, Represents the nth actual temperature value at the current time t; The disturbance variable value d t for: Where w is a positive integer greater than or equal to 1, t represents the current time, d t represents the disturbance variable value at the current time t, wherein the disturbance variable value includes load power and outdoor temperature, Represents the wth disturbance variable value at the current time t.

17. The device according to claim 16, characterized in that The first determining unit is specifically configured to: According to the historical data to be predicted at each moment, a state variable sequence about the actual temperature value and a disturbance variable sequence about the disturbance variable value are determined; the state variable sequence s L and the disturbance variable sequence d L They are all determined based on the data to be predicted in multiple time periods. s L =[s t-L+1 ,…,s t-1 ,s t ], d L =[d t-L+1 ,…,d t-1 ,d t ], Wherein, L represents the number of moments before the current moment, s t-L+1 Indicates the actual temperature value at time t-L+1, s t-1 Indicates the actual temperature value at time t-1, s t Indicates the actual temperature value at the current time t; d t-L+1 represents the disturbance variable value at time t-L+1, d t-1 represents the disturbance variable value at time t-1, d t Represents the value of the disturbance variable at the current time t.

18. The device according to claim 17, characterized in that The time-frequency features A of the state variable sequence include: <h2 style=";text-align:left;direction:ltr">A=[a1,a2,…,a<h2 style=";text-align:left;direction:ltr"> m <h2 style=";text-align:left;direction:ltr"> ], The time-frequency characteristics B of the disturbance variable sequence include: B=[b1,b2,…,b m ], m is a positive integer greater than 1, a m represents the mth time-frequency feature, b m Represents the mth time-frequency feature.

19. The device according to claim 18, characterized in that The input variable x t include: x t =[a1,a2,…,a m ,b1,b2,…,b m ,u t ]。 20. The device according to claim 19, characterized in that The training update unit comprises: The second prediction module is used to transform the input variable x at time t t Input into the regression model for prediction to obtain the historical temperature difference at time t+1; wherein the temperature difference represents the output data at time t+1; A second adding module, used for adding the historical temperature difference value to the actual temperature value at time t to obtain a predicted temperature value at time t+1; A difference module is used to make a difference between the predicted temperature value at time t+1 and the actual temperature value at time t+1 to obtain a prediction error value; The training update module is used to update the input variable x at time t if it is determined that the prediction error value is greater than or equal to a preset critical value. t As input, the actual temperature value at time t+1 and the historical temperature difference are used as output to form a pair of new samples; and the regression model is trained and updated according to the new samples to obtain a temperature prediction model; The stopping module is used to stop training and updating the regression model if it is determined that the prediction error value is less than the preset critical value, and continue training and updating the regression model until it is determined that the prediction error value is greater than or equal to the preset critical value.

21. A server, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor implements the method according to any one of claims 1 to 10 when executing the computer program.

22. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 10 when executed by a processor.

23. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 10 when being executed by a processor.

Images