TWI795340B - Freezer control system and temperature control host - Google Patents

Abstract

A control system for a freezer is suitable for cooperating with an indoor temperature sensor and includes at least one freezer device and a temperature control host. The refrigerator equipment includes a refrigerator temperature sensor and a refrigeration unit. The temperature control host stores a set temperature data, a temperature forecast database, an indoor temperature database and a refrigerator temperature database, and includes an optimum temperature prediction module and a refrigerator temperature setting module. Among them, the optimal temperature prediction module can generate an estimation data by neural network algorithm according to the set temperature value, the forecasted temperature value, the indoor temperature value and the refrigerator temperature value. The freezer temperature setting module can select a set temperature value from the estimated data as a set temperature target value according to a predetermined rule. The temperature control host can transmit the set temperature target value to the refrigerator equipment.

Description

Freezer control system and temperature control host

The invention relates to a control system and a host, in particular to a refrigerator control system and a temperature control host.

The freezers of the merchants are generally set by the engineering personnel considering factors such as the weather and the type of food to be refrigerated, and setting the operating temperature of the freezer based on their own experience. However, this method of setting the operating temperature based on experience may fail to fully consider various factors, and set a temperature that is not optimal, making the refrigerator unable to operate stably within the predetermined temperature range and affecting the refrigeration effect. In addition, when the internal temperature of the freezer is too high or too low to remain within the predetermined range, the merchant will need to contact the engineering personnel to readjust the set temperature. Operating at high or low temperatures may cause food spoilage.

Therefore, one object of the present invention is to provide a refrigerator control system capable of solving the aforementioned problems.

Therefore, in some embodiments, the refrigerator control system of the present invention is adapted to cooperate with an indoor temperature sensor installed in an indoor space, and includes at least one refrigerator device and a temperature control host. The refrigerator equipment is arranged in the indoor space, and includes a refrigerator temperature sensor for detecting the internal temperature of the refrigerator equipment, and a refrigeration unit for adjusting the internal temperature of the refrigerator equipment. The temperature control host can communicate with the freezer equipment and can receive the indoor temperature value output by the indoor temperature sensor. The temperature control host internally stores a set temperature data, a temperature forecast database, an indoor temperature database and a freezer temperature database, and includes an optimum temperature prediction module and a freezer temperature setting module. The set temperature data has a plurality of different set temperature values for setting the operating temperature of the refrigerator equipment, the temperature forecast database has a plurality of forecasted temperature values received from the outside, and the indoor temperature database has a plurality of records from the indoor The indoor temperature value output by the temperature sensor, the refrigerator temperature database has a plurality of refrigerator temperature values output by the refrigerator temperature sensor. Wherein, the optimum temperature prediction module can generate an estimated data by a pre-trained neural network algorithm based on the set temperature values, the forecasted air temperature values, the indoor temperature values and the refrigerator temperature values, and the estimated data There are multiple estimated values corresponding to different set temperature values; the freezer temperature setting module can select the set temperature value corresponding to the most appropriate estimated value from the estimated data according to a predetermined rule to set the target value of the set temperature; The temperature control host can transmit the set temperature target value to the refrigerator equipment, so that the refrigeration unit can adjust the internal temperature of the refrigerator equipment with the set temperature target value.

In some implementations, the predetermined rule of the refrigerator temperature setting module is based on the largest of the estimated values, the smallest of the estimated values, or the closest to a predetermined value among the estimated values The set temperature value corresponding to one of them is used as the selected set temperature target value.

In some implementations, the neural network algorithm of the optimal temperature prediction module is trained by a learning data, the learning data has a plurality of different set temperature values, and a plurality of forecasted temperatures respectively corresponding to the set temperature values value, multiple indoor temperature values corresponding to the set temperature values, multiple freezer temperature values corresponding to the set temperature values, and multiple evaluation mark values corresponding to the set temperature values, the neural network algorithm The training is based on the estimated value generated based on each of the set temperature values and the forecasted air temperature value, the indoor temperature value, and the refrigerator temperature value corresponding to each of the set temperature values, compared with the estimated values corresponding to each of the set temperature values Evaluate the relevance of the tag values to train and adjust the weights and/or biases of the neural network algorithm.

In some implementations, the temperature control host will regularly receive the new forecast air temperature value, the internal temperature value and the refrigerator temperature value respectively, and update the air temperature forecast database, the indoor temperature database and the refrigerator temperature data library; the temperature control host can judge whether the latest received freezer temperature value in the freezer temperature database is within a preset temperature range; if the judgment result is no, the temperature control host will use the updated temperature forecast data Database, the indoor temperature database and the refrigerator temperature database, re-execute the estimation data generated by the optimum temperature prediction module, the set temperature target value selected by the freezer temperature setting module, and the set temperature target value transmission to the program for the freezer unit.

In some implementations, the temperature control host receives the indoor temperature value and the refrigerator temperature value every minute to update the indoor temperature database and the refrigerator temperature database.

In some implementation aspects, the neural network algorithm uses a linear rectification function as the activation function, and the hidden layer of the neural network algorithm has 3 to 5 neural layers, and each neural layer has 64 to 128 neurons.

In some implementation aspects, the neural network algorithm uses a linear rectification function as the activation function, the hidden layer of the neural network algorithm has 3 to 5 neural layers, and each neural layer has 64 to 128 neurons; The neural network algorithm is trained with a batch size of 1 to 32 and an iteration number of 100 to 200.

Another object of the present invention is to provide a temperature control host.

Therefore, in some embodiments, the temperature control host of the present invention is suitable for cooperating with an indoor temperature sensor and a refrigerator device installed in an indoor space. The refrigerator equipment includes a refrigerator temperature sensor for detecting the internal temperature of the refrigerator equipment, and a refrigeration unit for adjusting the internal temperature of the refrigerator equipment. The temperature control host includes a set temperature data, a temperature forecast database, an indoor temperature database, a freezer temperature database, an optimum temperature prediction module and a freezer temperature setting module. The set temperature data has a plurality of different set temperature values for setting the operating temperature of the refrigerator equipment. The temperature forecast database has a plurality of forecasted temperature values received from outside. The indoor temperature database has a plurality of indoor temperature values output by the indoor temperature sensor. The refrigerator temperature database has a plurality of refrigerator temperature values output by the refrigerator temperature sensor. The optimal temperature prediction module can be controlled to execute a pre-trained neural network algorithm. The refrigerator temperature setting module can receive the output data of the optimum temperature prediction module to set the temperature of the refrigerator equipment. Wherein, the optimal temperature forecasting module can generate an estimated data with the neural network algorithm based on the set temperature values, the forecasted air temperature values, the indoor temperature values and the refrigerator temperature values, and the estimated data has multiple The pens correspond to the estimated values of different set temperature values respectively. The freezer temperature setting module can select the set temperature value corresponding to the most suitable estimated value from the estimated data according to a predetermined rule to determine the set temperature target value. The temperature control host can transmit the set temperature target value to the refrigerator equipment, so that the refrigeration unit can adjust the internal temperature of the refrigerator equipment according to the set temperature target value.

The present invention has at least the following effects: the control system of the refrigerator can operate according to the continuously updated forecasted air temperature values and the indoor temperatures through the cooperation of the optimum temperature prediction module of the temperature control host and the refrigerator temperature setting module. value, the temperature value of the freezer, efficiently and reliably select the most suitable one among multiple set temperature values to set the operating temperature of the freezer equipment, so that the freezer equipment can be stably within the predetermined temperature range It operates in the interval to ensure the refrigeration effect of food. In addition, the temperature control host can also monitor the operating temperature of the refrigerators, and reset the temperature of the refrigerators in real time when the temperature is abnormal, so as to ensure the continuity of the refrigeration effect.

Referring to FIG. 1 and FIG. 2 , the refrigerator control system 100 of the present invention is suitable for cooperating with an indoor temperature sensor 201 disposed in an indoor space 200 . The indoor space 200 is, for example, a storefront interior space of a merchant, and the indoor temperature value of the indoor space 200 can be detected regularly by the indoor temperature sensor 201 . In this embodiment, the freezer control system 100 includes at least one freezer device 1 and a temperature control host 2. In the figure, two freezer devices 1 are taken as an example. However, according to actual needs, these freezer devices 1 can be used in any number implementation, and is not limited to only being set in the same indoor space 200, so it is not limited by the content disclosed in this embodiment.

The refrigerator equipment 1 is arranged in the indoor space, and each includes a refrigerator temperature sensor 11 for detecting the internal temperature of the refrigerator equipment 1, and a refrigeration unit for adjusting the internal temperature of the refrigerator equipment 1 12, and the cabinet body, cabinet door, communication unit and other components not shown in the figure. The refrigeration unit 12 may include, for example, components not specifically shown, such as a compressor, a condenser, a capillary tube, an evaporator, and a blower fan.

The temperature control host 2 can be installed in the indoor space 200, and can also be installed in an area outside the indoor space 200. It can be realized by various server systems, and can communicate with the refrigerator equipment 1 via a line. And the indoor temperature value output by the indoor temperature sensor 201 can be received. The temperature control host 2 includes, for example, hardware components such as processors, memory, storage units, communication units, and input/output interfaces not shown in the figure, and has an optimum temperature prediction module 21 and a software module running as a software module. The freezer temperature setting module 22 stores a set temperature data 23 , a temperature forecast database 24 , an indoor temperature database 25 and a freezer temperature database 26 .

The optimum temperature prediction module 21 can be controlled to execute a pre-trained neural network algorithm, so as to run the main steps of the temperature setting control of the refrigerators 1, which will be described later. The refrigerator temperature setting module 22 can receive the output data of the optimum temperature prediction module 21 to set the operating temperature of the refrigerator device 1 .

The set temperature data 23 has a plurality of different set temperature values for setting the operating temperature of the refrigerator device 1. When the temperature control host 2 is running, the set temperature values of the set temperature data 23 remain unchanged in principle. And as one of the input information of the input layer of the neural network algorithm of the optimal temperature prediction module 21. For example, the set temperature values may be 0° C., any value higher or lower than 0° C., and the value selection may depend on the type of food to be refrigerated.

The temperature forecast database 24 has a plurality of forecasted temperature values received from the outside, and the forecasted temperature values are one of the input information of the input layer of the neural network algorithm of the optimum temperature prediction module 21 . The forecast temperature values may be, for example (but not limited to), the predicted temperature values of the location of the indoor space 200 in the next week, and will be updated regularly, and the information sources thereof may be selected from government meteorological agencies or private meteorological units.

The indoor temperature database 25 has a plurality of indoor temperature values output by the indoor temperature sensor 201, and these indoor temperature values are one of the input information of the input layer of the neural network algorithm of the optimum temperature prediction module 21 , and will be updated periodically. In a preferred implementation, the temperature control host 2 receives an indoor temperature value every minute and updates the indoor temperature database 25. Such an update frequency and data quantity can ensure the safety of these refrigerators 1. The temperature setting provides a more immediate and accurate control basis.

The refrigerator temperature database 26 has a plurality of refrigerator temperature values output by the refrigerator temperature sensor 11, and these refrigerator temperature values are also one of the input information of the neural network algorithm input layer of the optimal temperature prediction module 21. One, and will be updated regularly. In a preferred implementation, the temperature control host 2 receives a freezer temperature value every minute to update the freezer temperature database 26, such an update frequency allows the temperature control host 2 to realize the control of the freezer equipment 1 Real-time monitoring of the operating temperature, and can detect abnormalities as early as possible and make adjustments when the operating temperature is too high or too low to maintain the predetermined temperature range. In addition, the above-mentioned update frequency enables the refrigerator temperature database 26 to have sufficient data volume to ensure a real-time and accurate control mechanism for the temperature setting of the refrigerator devices 1 .

Hereinafter, the execution flow of setting the operating temperature of the refrigerators 1 by the temperature control host 2 will be described.

In step S01, the optimum temperature prediction module 21 will access the latest update information in the set temperature data 23, the air temperature forecast database 24, the indoor temperature database 25, and the refrigerator temperature database 26, according to the settings The temperature value, the forecast air temperature value, the indoor temperature value, and the refrigerator temperature value, use the pre-trained neural network algorithm to generate an estimated data, and the estimated data has a plurality of estimated values corresponding to different set temperature values . In this step, the estimated values are used as the output data of the output layer of the neural network algorithm, which is the evaluation value calculated by the hidden layer of the neural network algorithm for each set temperature value based on the information received by the input layer , which means that the temperature control host 2 can select which of the set temperature values according to environmental factors (these forecast air temperature values, these indoor temperature values) and equipment factors (these freezer temperature values) in the follow-up procedure It is most suitable as the setting parameter that should be selected at the moment.

In step S02, after the optimum temperature prediction module 21 completes the calculation of the estimation data, the freezer temperature setting module 22 can select the temperature corresponding to the most appropriate estimation value from the estimation data according to a predetermined rule. The set temperature value is set temperature target value. For example, the predetermined rule may be to select the maximum of the estimated values, the smallest of the estimated values, or the set temperature corresponding to the one of the estimated values that is closest to a predetermined value value, as the selected target value of the set temperature, the selection of the mechanism of the above-mentioned predetermined rule depends on the training content of the neural network algorithm. For example, there are 10 such set temperature values in the set temperature data 23, and an estimated value is calculated respectively for these 10 set temperature values in step S01, if the predetermined rule is based on the largest of these estimated values as the selection If the basis is determined, then in step S02, the set temperature value corresponding to the maximum estimated value will be used as the set temperature target value. In this way, the temperature control host 2 can obtain the corresponding estimated values after calculating and analyzing sufficient information based on the latest updated forecasted air temperature values, the indoor temperature values, and the refrigerator temperature values, and The set temperature target value that enables the refrigerators 1 to operate stably within a predetermined temperature range is further selected. Finally, in step S03, the temperature control host 2 can transmit the set temperature target value to the freezer equipment 1, so that the refrigeration unit 12 can adjust the internal temperature of the freezer equipment 1 with the set temperature target value, and complete the control of the set temperature. The regular automatic temperature setting process of the refrigerator equipment 1.

Further speaking, regarding the operation temperature monitoring and abnormal temperature management of the refrigerator equipment 1, since the temperature control host 2 will receive the new forecasted air temperature values, the internal temperature values and the refrigerator temperature values respectively regularly, and Update the air temperature forecast database 24, the indoor temperature database 25 and the refrigerator temperature database 26, so the temperature control host 2 can update the refrigerator equipment 1 in real time according to the data update of the refrigerator temperature database 26. operating temperature monitoring. Therefore, the temperature control host 2 can continuously judge whether the latest refrigerator temperature value received in the refrigerator temperature database is within a preset temperature range interval, and if the judgment result is yes, it represents the last set temperature target The value can meet the current environmental factors and equipment factors, so that the refrigerator equipment 1 can provide a predetermined type of food refrigeration function at a relatively stable operating temperature. On the contrary, if the above judgment result is negative, the temperature control host 2 will re-execute the optimal temperature prediction module based on the updated air temperature forecast database 24, the indoor temperature database 25 and the refrigerator temperature database 26. 21. The process of generating the estimated data, selecting the set temperature target value by the freezer temperature setting module 22, and transmitting the set temperature target value to the freezer equipment 1, replacing the previous set temperature with a new set temperature target value The target value allows the operating temperature of the refrigerator equipment 1 to gradually return to the normal temperature range.

On the other hand, in order to allow the neural network algorithm of the optimum temperature prediction module 21 to analyze and calculate reliable estimated values, this embodiment carries out specific configurations for the model setting and training of the neural network algorithm, described as follows.

Table 1: Model settings for the neural network algorithm model structure project illustrate input layer Set temperature value The input information of the input layer Forecast temperature value The input information of the input layer Indoor temperature value The input information of the input layer Freezer temperature value The input information of the input layer hidden layer The number of layers of the neural layer 3 to 5 floors are better Number of neurons per neural layer 64 to 128 is better output layer estimated value Output information of the output layer (one for each set temperature value)

Referring to Table 1 above, in this embodiment, the input layer of the neural network algorithm is configured to receive four types of information, including the set temperature values, the forecasted temperature values, the indoor temperature values, and the refrigerator temperature values. It is to allow the neural network algorithm to accurately analyze and calculate the current time based on the continuously updated environmental data (these forecast air temperature values, the these indoor temperature values) and equipment data (these freezer temperature values) with sufficient data volume. Whichever is the best set temperature value. In the configuration of the hidden layer, in this embodiment, the hidden layer preferably has 3 to 5 neural layers, and the number of neurons in each neural layer is preferably 64 to 128. If the number of neural layers and the number of neurons are too small, it is difficult to obtain accurate and reliable estimation results. Although increasing the number of neural layers and the number of neurons can improve the accuracy of the estimation results, it may be due to the overly complex model. The overfitting factor enables the neural network algorithm to obtain good accuracy when analyzing the learning data during the training process, but the accuracy may decrease when the new data is analyzed later. Therefore, this embodiment helps to obtain accurate and effective estimation results when performing the aforementioned steps S01 and S02 by properly configuring the hidden layer.

On the other hand, in this embodiment, the training of the neural network algorithm of the optimum temperature prediction module 21 is performed by a learning material. The learning materials have multiple different set temperature values, multiple forecast air temperature values corresponding to the set temperature values, multiple indoor temperature values corresponding to the set temperature values, and multiple indoor temperature values corresponding to the set temperature values. The temperature value of the refrigerator, and a plurality of artificially given evaluation label values (labels) respectively corresponding to the set temperature values. The training of the neural network algorithm is based on the estimated value generated based on each of the set temperature values and the forecasted air temperature value corresponding to each of the set temperature values, the indoor temperature value and the refrigerator temperature value, compared with the estimated value corresponding to each of the set temperature values The correlation of the evaluation mark value of the temperature value is set to train and adjust the weight value and/or bias value of the neural network algorithm.

Specifically, in step S01, the neural network algorithm calculates different set temperature values based on the same (updated in the latest period) values of the forecasted air temperature, the indoor temperature, and the temperature of the refrigerator the estimated value of . During the training process of the neural network algorithm, the forecasted air temperature values, the indoor temperature values, and the refrigerator temperature values used to calculate the estimated values of different set temperature values in the learning materials are the past Therefore, the training engineers will set the freezer equipment 1 based on the data combination of various forecast air temperature values, indoor temperature values, and freezer temperature values according to the set temperature values corresponding to each data combination. The stability of the displayed operating temperature is artificially given different evaluation mark values. For example, the evaluation mark value can be an integer from 0 to 9. The larger the number, the better the temperature stability (it is not easy to exceed the predetermined temperature range range) . Therefore, the neural network algorithm can adjust the weight value and/or bias value during the training process using the learning data, so that the estimated value calculated by each set temperature value can be gradually compared with the predetermined value. Corresponding to the value of the evaluation mark, in order to achieve the purpose of training. In the above example, since the neural network algorithm is trained with a larger value of the evaluation flag, the predetermined rule used in step S02 will select the set temperature target based on the condition of having the largest estimated value. value. If the evaluation flag value set during training has different evaluation criteria, the predetermined rule used in step S02 can be adjusted accordingly.

In addition, further considering the operating performance of the temperature control host 2 and the execution efficiency of the neural network algorithm, this embodiment will input the input information of the input layer ( The set temperature values, the forecast temperature values, the indoor temperature values, and the refrigerator temperature values) are preprocessed, for example, the data values are scaled to a range of 0 to 1 to speed up the convergence speed of the neural network algorithm , and the null value (null) in the data can be removed first. In addition, the processing of the input information can preferably use a batch size (batch size) of 1 to 32, an iteration number (epoch) of 100 to 200, and an activation function of Rectified Linear Unit (ReLU). In this way, the temperature control host 2 can allow the neural network algorithm to perform calculations efficiently under the state of fully utilizing the operating performance, so that the analysis and calculation steps of step S01 can be quickly completed and the subsequent temperature setting steps can be performed. In order to realize the real-time temperature setting and monitoring management of these refrigerators 1 .

Based on the foregoing description, the refrigerator control system 100 of the present invention can operate according to the continuously updated forecasted temperature values, the The indoor temperature value and the temperature value of the refrigerators are efficiently and reliably selected among multiple set temperature values, and the most suitable one is selected according to the operation temperature setting of the refrigerator equipment 1, so that the refrigerator equipment 1 can Continuously operate in the predetermined temperature range to ensure the refrigeration effect of food. In addition, the temperature control host 2 can also monitor the operating temperature of the refrigerators 1, and reset the temperature of the refrigerators 1 immediately when the temperature is abnormal, so as to ensure the continuity of the refrigeration effect. Therefore, the refrigerator control system 100 of the present invention can indeed achieve the purpose of the present invention.

But what is described above is only an embodiment of the present invention, and should not limit the scope of the present invention. All simple equivalent changes and modifications made according to the patent scope of the present invention and the content of the patent specification are still within the scope of the present invention. Within the scope covered by the patent of the present invention.

100: Freezer control system 200: Interior space 201: Indoor temperature sensor 1: Freezer equipment 11: Freezer temperature sensor 12: Refrigeration unit 2: Temperature control host 21:Optimum temperature prediction module 22: Freezer temperature setting module 23: Set temperature data 24: Temperature forecast database 25: Indoor temperature database 26: Freezer temperature database S01~S03: Steps

Other features and effects of the present invention will be clearly presented in the implementation manner with reference to the drawings, wherein: Fig. 1 is a system block diagram illustrating an embodiment of the refrigerator control system of the present invention; and Figure 2 is a flow chart of the operation of the refrigerator control system.

100: Freezer control system

200: Interior space

201: Indoor temperature sensor

1: Freezer equipment

11: Freezer temperature sensor

12: Refrigeration unit

2: Temperature control host

21:Optimum temperature prediction module

22: Freezer temperature setting module

23: Set temperature data

24: Temperature forecast database

25: Indoor temperature database

26: Freezer temperature database

Claims (8)

A control system for a refrigerator is suitable for cooperating with an indoor temperature sensor installed in an indoor space, the control system for the refrigerator includes: At least one refrigerator device is installed in the indoor space, and includes a refrigerator temperature sensor for detecting the internal temperature of the refrigerator device, and a refrigeration unit for adjusting the internal temperature of the refrigerator device; and A temperature control host, which can communicate with the refrigerator equipment, and can receive the indoor temperature value output by the indoor temperature sensor. The temperature control host stores a set temperature data, an air temperature forecast database, and an indoor temperature database and a freezer temperature database, including an optimum temperature prediction module and a freezer temperature setting module, the set temperature data has multiple different set temperature values for setting the operating temperature of the freezer equipment, the air temperature forecast data The database has multiple forecast air temperature values received from the outside, the indoor temperature database has multiple indoor temperature values output by the indoor temperature sensor, and the refrigerator temperature database has multiple indoor temperature values output by the refrigerator temperature sensor. Freezer temperature value, Wherein, the optimum temperature prediction module can generate an estimated data by a pre-trained neural network algorithm based on the set temperature values, the forecasted air temperature values, the indoor temperature values and the refrigerator temperature values, and the estimated data There are multiple estimated values corresponding to different set temperature values; the freezer temperature setting module can select the set temperature value corresponding to the most appropriate estimated value from the estimated data according to a predetermined rule to set the target value of the set temperature; The temperature control host can transmit the set temperature target value to the refrigerator equipment, so that the refrigeration unit can adjust the internal temperature of the refrigerator equipment with the set temperature target value. The refrigerator control system as described in claim 1, wherein the predetermined rule of the refrigerator temperature setting module is the largest of the estimated values, the smallest of the estimated values, or the estimated value The set temperature value corresponding to the one closest to a predetermined value is used as the selected set temperature target value. The refrigerator control system as described in Claim 1, wherein, the neural network algorithm of the optimum temperature prediction module is trained by a learning data, and the learning data has a plurality of different set temperature values, and the plurality of records correspond to each of the The forecast air temperature value of the set temperature value, multiple indoor temperature values corresponding to each set temperature value, multiple freezer temperature values respectively corresponding to each set temperature value, and multiple evaluation mark values respectively corresponding to each set temperature value , the training of the neural network algorithm is based on the estimated value generated based on each of the set temperature values and the forecasted air temperature value, the indoor temperature value, and the refrigerator temperature value corresponding to each of the set temperature values. The correlation of the evaluation mark value of the set temperature value is used to train and adjust the weight value and/or bias value of the neural network algorithm. The freezer control system as described in claim 1, wherein the temperature control host will regularly receive the new forecasted air temperature value, the internal temperature value and the freezer temperature value, and update the air temperature forecast database and the indoor temperature data library and the freezer temperature database; the temperature control host can judge whether the latest received freezer temperature value in the freezer temperature database is within a preset temperature range; if the judgment result is no, the temperature control host will update the After the air temperature forecast database, the indoor temperature database and the refrigerator temperature database, re-execute the estimation data generated by the optimum temperature prediction module, the set temperature target value selected by the refrigerator temperature setting module, and the The set temperature target value is sent to the program of the refrigerator equipment. The refrigerator control system as described in Claim 4, wherein, the temperature control host receives the indoor temperature value and the refrigerator temperature value every minute to update the indoor temperature database and the refrigerator temperature database. The refrigerator control system as described in claim 1, wherein the neural network algorithm uses a linear rectification function as the activation function, and the hidden layer of the neural network algorithm has 3 to 5 neural layers, and each neural layer has 64 to 128 neurons. The refrigerator control system as described in claim 3, wherein the neural network algorithm uses a linear rectification function as the activation function, and the hidden layer of the neural network algorithm has 3 to 5 neural layers, and each neural layer has 64 to 128 neurons; the neural network algorithm is trained with batch sizes of 1 to 32 and iterations of 100 to 200. A temperature control host suitable for cooperating with an indoor temperature sensor and a refrigerator device installed in an indoor space, the refrigerator device including a refrigerator temperature sensor for detecting the internal temperature of the refrigerator device, and A refrigeration unit for adjusting the internal temperature of the refrigerator equipment, the temperature control host includes: A set temperature data, which has multiple different set temperature values for setting the operating temperature of the refrigerator equipment; A temperature forecast database, which has multiple forecasted temperature values received from the outside; An indoor temperature database with multiple indoor temperature values output by the indoor temperature sensor; A freezer temperature database with multiple freezer temperature values output by the freezer temperature sensor; An optimal temperature prediction module, which can be controlled to execute a pre-trained neural network algorithm; and A freezer temperature setting module capable of receiving the output data from the optimum temperature prediction module to set the temperature of the freezer equipment, Wherein, the optimal temperature prediction module can generate an estimated data by the neural network algorithm based on the set temperature values, the forecasted air temperature values, the indoor temperature values and the refrigerator temperature values, and the estimated data has multiple Each of the estimated values corresponding to different set temperature values; the freezer temperature setting module can select the set temperature value corresponding to the most appropriate estimated value from the estimated data according to a predetermined rule; the set temperature target value; the temperature The control host can transmit the set temperature target value to the refrigerator equipment, so that the refrigeration unit can adjust the internal temperature of the refrigerator equipment according to the set temperature target value.

Images