A cold storage sub-region temperature control system and method

Abstract

The application provides a cold storage sub-region temperature control system and method, the system is provided with a collection module, a data processing module, a comparison module, a first calculation module, a generation module, a second calculation module, an adjustment module, a communication module and a storage module; the implementation method of the system is to divide the cold storage into multiple independent temperature control regions, collect the actual temperature data of multiple points in the cold storage through the collection module to generate an original temperature data set; then generate a standardized temperature data sequence through the data processing module; then calculate the real-time temperature deviation sequence of each region and the initial control amount of the temperature adjustment device through the first calculation module; generate a predicted temperature sequence based on historical temperature data and environmental parameter data through the generation module; calculate the initial control amount through feedforward compensation calculation through the second calculation module to obtain the final control amount of the temperature adjustment device of each region; and adjust the operation parameters of the temperature adjustment device of each region according to the corresponding control instruction generated by the adjustment module to realize sub-region temperature control.

Description

Technical Field

[0001] This invention relates to the field of cold storage zone temperature control technology, specifically a cold storage zone temperature control system and method. Background Technology

[0002] As a core facility for low-temperature storage, the accuracy of temperature control in cold storage directly affects the quality of stored goods and energy consumption. Traditional cold storage temperature control typically employs a "global temperature control" model, which uses a single temperature sensor to collect the overall temperature within the cold storage and combines it with a PID (proportional-integral-derivative) control algorithm to adjust the operating parameters of the refrigeration system, such as compressors and fans. However, due to the large internal space of cold storage, different areas (such as areas near the door, corner areas, and areas with densely stacked goods) are affected by external thermal disturbances (such as the number of times the door is opened and ambient temperature), heat release from goods (such as respiratory heat), and uneven airflow distribution, resulting in significant spatial differences in the temperature field. The traditional global temperature control model is prone to causing localized temperature variations. Temperature deviations from target values ​​(e.g., higher temperatures near the door and lower temperatures in interior areas) fail to meet the differentiated temperature requirements of different zones. While existing technologies address these issues, their effectiveness is less than ideal. For example, patent CN202011587984.X, "A Local Temperature Control Method and Cold Storage for Cold Storage," primarily identifies the storage types in each zone and controls the refrigeration units within each zone to operate at preset temperatures corresponding to the types of items stored there. It emphasizes ensuring a clear correspondence between the types of items and their preset temperatures. Another example is patent CN202411930596.5, "An Intelligent Variable Frequency Refrigeration System with Multi-Zone Independent Temperature Control." The preservation method and system achieves refined temperature control management of different areas inside the refrigeration equipment through the collaborative work of a zone division module, a temperature control module, an environmental monitoring module, a data analysis module, an intelligent frequency conversion control module, and a user interface module. The system uses 3D laser scanning and multispectral image acquisition technology to automatically divide temperature zones, and a quantum-level precision thermistor array and an intelligent fuzzy PID controller ensure accurate temperature control. Simultaneously, it monitors humidity and airflow to optimize the storage environment. Patent CN202411450050.X, "A Local Temperature Control Method and System for an Energy-Saving Cold Storage," divides the interior of the cold storage into several zones according to the temperature control distribution of the temperature regulation equipment; and determines the temperature control of each zone. The system prioritizes storing initial indicators; obtains the current storage status of each area; analyzes the predicted future storage volume of the cold storage; determines the areas that the cold storage will need to store in the future, and marks them as target temperature control areas; sends control commands to the temperature control equipment in the target temperature control areas to adjust the temperature of the target temperature control areas to the set target temperature of the cold storage; sends control commands to the temperature control equipment in areas other than the target temperature control areas to adjust the temperature of the areas other than the target temperature control areas to the standby temperature; the above patents all have certain limitations, cannot perform real-time temperature control adjustment, and have limited application scenarios. Therefore, based on the needs of large cold storage, a cold storage zoned temperature control system and method is proposed, which divides the internal space of the cold storage into N independent temperature control areas.By using a pre-built heat exchange correlation model, the correlation effects of airflow exchange and heat transfer between different areas are quantified, such as the diffusion of heat disturbance caused by opening doors in adjacent areas. This avoids control conflicts caused by ignoring heat interaction between areas in traditional independent temperature control, such as the passive temperature rise of adjacent areas when one area is refrigerated. At the same time, based on historical temperature data and environmental parameters, such as door opening frequency, ambient temperature, and heat release from goods, a pre-trained machine learning model is used to predict the temperature change trend of each area in the future. The prediction results are combined to perform feedforward compensation on the initial control quantity, upgrading the control strategy from "passively responding to current deviations" to "actively predicting future trends". This achieves high-precision and stable temperature control of each area, which is suitable for mixed storage scenarios of items with large differences in temperature sensitivity. Summary of the Invention

[0003] To address the aforementioned technical issues, this invention proposes a zoned temperature control system and method for cold storage. By dividing the internal space of the cold storage into N independent temperature control zones, a pre-constructed heat exchange correlation model quantifies the correlation effects between zones, such as airflow exchange and heat transfer, including the diffusion of heat disturbance caused by opening doors in adjacent zones. This avoids control conflicts caused by neglecting inter-zone heat interaction in traditional independent temperature control, such as the passive temperature rise of adjacent zones when one zone is refrigerated. Simultaneously, based on historical temperature data and environmental parameters, such as door opening frequency, ambient temperature, and heat release from goods, a pre-trained machine learning model predicts the future temperature change trends of each zone. The prediction results are then used to feedforward compensate the initial control input, upgrading the control strategy from "passively responding to current deviations" to "actively predicting future trends." This achieves high-precision and stable temperature control in each zone, suitable for mixed storage scenarios involving items with significant temperature sensitivity differences.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A cold storage zoned temperature control system and method, characterized in that: the cold storage zoned temperature control system includes a data acquisition module, a data processing module, a comparison module, a first calculation module, a generation module, a second calculation module, an adjustment module, a communication module, and a storage module; the data acquisition module is installed in N independent temperature control zones dividing the internal space of the cold storage, with a temperature detection device installed in each zone, and continuously acquires actual temperature data from multiple points within each zone at preset time intervals to generate an original temperature dataset; the data processing module cleans the generated original temperature dataset, removes outliers, and completes missing data using an interpolation algorithm to generate an effective temperature dataset, and preprocesses the effective temperature dataset, including data smoothing and standardization, to generate a standardized temperature data sequence; the comparison module compares the standardized temperature data sequences of each zone... The system compares the temperature with the corresponding preset target temperature value to calculate the real-time temperature deviation sequence for each region. The calculation module one, based on the real-time temperature deviation sequence for each region, calculates the initial control quantity of the temperature regulating device for each region using a pre-built heat exchange correlation model. The generation module, based on historical temperature data and environmental parameter data, predicts the temperature change trend of each region in the future using a pre-trained machine learning model, generating a predicted temperature sequence. The calculation module two performs feedforward compensation calculation on the initial control quantity according to the predicted temperature sequence to obtain the final control quantity of the temperature regulating device for each region. The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulating device for each region, achieving regional temperature control. The communication module is responsible for communication between the various system modules. The storage module stores data during system operation.

[0006] Furthermore, the implementation steps of the cold storage zoned temperature control system are as follows:

[0007] Step 1: Divide the internal space of the cold storage into N independent temperature control zones. Install a temperature detection device in each zone and connect it to a data acquisition module. The data acquisition module continuously collects the actual temperature data of multiple points inside each zone at preset time intervals to generate the original temperature dataset.

[0008] Step 2: The data processing module cleans the generated raw temperature dataset, removes outliers, and fills in missing data using an interpolation algorithm to generate a valid temperature dataset. The valid temperature dataset is then preprocessed, and a standardized temperature data sequence is generated through data smoothing and standardization.

[0009] Step 3: The comparison module compares the standardized temperature data sequence of each region with its corresponding preset target temperature value to calculate the real-time temperature deviation sequence of each region.

[0010] Step 4: Calculation Module 1 calculates the initial control quantity of the temperature regulation device in each region based on the real-time temperature deviation sequence of each region and through a pre-built heat exchange correlation model.

[0011] Step 5: The generation module uses historical temperature data and environmental parameter data to predict the temperature change trend of each region in the future through a pre-trained machine learning model, and generates a predicted temperature sequence.

[0012] Step 6: Calculation Module 2 performs feedforward compensation calculation on the initial control quantity based on the predicted temperature sequence to obtain the final control quantity of the temperature regulation device in each region.

[0013] Step 7: The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulation devices in each area, thereby achieving zoned temperature control.

[0014] Furthermore, the specific process of data cleaning of the original temperature dataset by the data processing module in step two of the implementation method of the cold storage zone temperature control system is as follows:

[0015] 1) Use the moving average method or median filtering method to smooth the original temperature dataset, and identify and remove outlier data points that exceed the preset confidence interval;

[0016] 2) Missing data points are filled in using linear interpolation based on temperature values ​​at adjacent time points or correlation interpolation based on temperature values ​​in adjacent regions;

[0017] 3) The following formula is used to identify outlier data points:

[0018]

[0019] in: This is the raw temperature data at the current time t;

[0020] The average temperature within a sliding window centered at t;

[0021] This represents the standard deviation of the temperature within that window.

[0022] The preset confidence coefficient has a value range of [2.5, 3.5].

[0023] Furthermore, the specific process of building the pre-constructed heat exchange correlation model used in the calculation module one of the implementation method of the cold storage zone temperature control system is as follows:

[0024] 1) During the operation of the cold storage system, continuously collect the operating parameters of the temperature control devices in each area, the real-time temperature data of each area, and the ambient temperature data to build a training dataset;

[0025] 2) Perform feature engineering on the training dataset to extract feature variables including temperature change rate, cumulative running time of the regulating device, and ambient temperature difference;

[0026] 3) Using a regression analysis algorithm, based on characteristic variables and temperature change data of each region, a heat exchange coefficient matrix is ​​trained to describe the influence relationship of heat exchange between regions;

[0027] 4) Based on the heat exchange coefficient matrix, a heat exchange correlation model is constructed to calculate the temperature impact of any temperature change in one region on adjacent regions;

[0028] 5) The heat exchange correlation model is specifically represented as follows:

[0029]

[0030] in: This represents the temperature change in the i-th region caused by the operation of other regions.

[0031] Let be the heat exchange coefficient from region j to region i;

[0032] Let be the operating power of the temperature control device in region j.

[0033] Furthermore, in step four of the implementation method of the cold storage zone temperature control system, the process of calculating the initial control quantity using the pre-built heat exchange correlation model in calculation module one is as follows:

[0034] 1) For each region, the basic adjustment amount for that region is calculated using either a PID control algorithm or a fuzzy control algorithm based on its temperature deviation sequence;

[0035] 2) Input the basic adjustment amount into the heat exchange correlation model to calculate the predicted temperature impact on all adjacent areas when the temperature is adjusted in this area;

[0036] 3) Based on the predicted temperature impact, calculate the compensation adjustment required for all adjacent areas to offset the impact;

[0037] 4) The basic adjustment amount and the corresponding compensation adjustment amount are weighted and summed to obtain the initial control amount of the temperature control device in each zone;

[0038] 5) The basic adjustment calculation formula for the PID control algorithm is:

[0039]

[0040] in: Basic adjustment amount;

[0041] This represents the real-time temperature deviation.

[0042] , , These are the proportional, integral, and differential coefficients, respectively.

[0043] Furthermore, the pre-trained machine learning model processing procedure used in step five of the implementation method of the cold storage zone temperature control system is as follows:

[0044] 1) Obtain historical operating data of the cold storage, including historical temperature data, historical environmental parameter data, historical operating parameter data of the regulating device, and the corresponding timestamps;

[0045] 2) Label the historical operating data, divide the data into time windows, and label the temperature value at the end of each time window as the prediction target;

[0046] 3) Divide the processed dataset into training set, validation set, and test set;

[0047] 4) Select the LSTM model as the initial prediction model, train the initial prediction model using the training set, and perform hyperparameter tuning using the validation set to obtain the pre-trained machine learning model.

[0048] 5) Use the test set to evaluate the prediction accuracy of the pre-trained machine learning model. If the accuracy meets the preset requirements, the model training is complete.

[0049] 6) The cell state update formula for the LSTM model is:

[0050]

[0051]

[0052]

[0053]

[0054] in: Output for the forget gate;

[0055] For input gate output;

[0056] Candidate cell state;

[0057] This represents the current state of the cell.

[0058] The hidden state from the previous moment;

[0059] For the current input;

[0060] For the sigmoid function;

[0061] It is the hyperbolic tangent function;

[0062] These are the weight parameters of the model;

[0063] This is the bias parameter.

[0064] Furthermore, the online update process of the pre-trained machine learning model used in step five of the implementation method of the cold storage zone temperature control system is as follows:

[0065] 1) Regularly add newly generated temperature data and environmental parameter data as new samples to the model's training dataset;

[0066] 2) When the number of new samples reaches a preset threshold, the model retraining process is triggered, and the pre-trained machine learning model is incrementally trained or retrained using the updated training dataset to update the model parameters.

[0067] 3) The loss function for model retraining is the root mean square error, and its calculation formula is as follows:

[0068]

[0069] in: The number of samples;

[0070] This represents the actual temperature value of the i-th sample.

[0071] This represents the predicted temperature value corresponding to the model.

[0072] Furthermore, in step six of the implementation method of the cold storage zone temperature control system, the specific process of calculation module two performing feedforward compensation calculation on the initial control quantity based on the predicted temperature sequence is as follows:

[0073] 1) Compare the predicted temperature sequence with the corresponding preset target temperature value to generate a predicted temperature deviation sequence;

[0074] 2) Based on the predicted temperature deviation sequence, the feedforward compensation amount is calculated through the feedforward control algorithm. The feedforward compensation amount is used to offset the predicted temperature deviation in advance.

[0075] 3) The feedforward compensation amount is superimposed with the initial control amount to obtain the final control amount of the temperature regulation device in each zone;

[0076] 4) The formula for calculating the final control quantity is:

[0077]

[0078] in: For the final control quantity;

[0079] This is the initial control variable;

[0080] This is the feedforward compensation gain coefficient;

[0081] For the future Predicted temperature deviation at any given time.

[0082] Furthermore, the specific process for handling abnormalities encountered in the implementation method of the cold storage zone temperature control system is as follows:

[0083] 1) Monitor the actual temperature data of each area. When the actual temperature value of any area exceeds the preset safety threshold range and the duration of this state exceeds the preset duration, it is determined to be a temperature abnormality event.

[0084] 2) In response to an abnormal temperature event, an alarm signal is triggered, and the emergency control mode is activated;

[0085] 3) In emergency control mode, the prediction and feedforward compensation calculation of the machine learning model are suspended, and the final control quantity is directly generated by the maximum output power control algorithm based on the temperature deviation sequence.

[0086] 4) The formula for generating the control quantity in the maximum output power control algorithm is:

[0087]

[0088] in: For emergency control purposes;

[0089] This is the maximum output power of the temperature control device;

[0090] The safe temperature deviation threshold;

[0091] This is a normal PID control value.

[0092] The benefits of this application are:

[0093] 1. The cold storage zone temperature control system and method quantifies the correlation effects of airflow exchange and heat transfer between zones through a pre-constructed heat exchange correlation model. This effectively avoids the spread of heat disturbance caused by opening doors in adjacent zones, effectively avoids control conflicts caused by ignoring heat interaction between zones in traditional independent temperature control, and effectively avoids the passive rise in temperature of adjacent zones when zones are refrigerated.

[0094] 2. The cold storage zone temperature control system and method are based on historical temperature data and environmental parameters. It makes full use of parameters such as door opening frequency, ambient temperature, and heat release of goods. It uses a pre-trained machine learning model to predict the temperature change trend of each zone in the future. The prediction results are combined to feedforward compensation of the initial control quantity, so that the control strategy is upgraded from "passively responding to the current deviation" to "actively predicting the future trend", which significantly shortens the response time of temperature regulation.

[0095] 3. The cold storage zone temperature control system and method can increase the refrigeration power in advance after the door is opened to offset the thermal disturbance, reduce control overshoot, avoid the temperature from being too low due to lag adjustment, improve the dynamic stability of the system, and achieve high-precision and stable temperature control of each zone. It is suitable for mixed storage scenarios of items with large differences in temperature sensitivity and has significant practical value and promotion significance. Attached Figure Description

[0096] Figure 1 This is a schematic diagram of the system implementation method of the present invention;

[0097] Figure 2 This is a schematic diagram illustrating the implementation process of the pre-constructed heat exchange correlation model of the present invention;

[0098] Figure 3 This is a schematic diagram illustrating the implementation process of the pre-trained machine learning model of the present invention;

[0099] Figure 4 This is a schematic diagram of the process for calculating the initial control quantity using the pre-constructed heat exchange correlation model of this invention;

[0100] Figure 5 This is a schematic diagram of the feedforward compensation calculation process for the initial control quantity based on the predicted temperature sequence of the present invention. Detailed Implementation

[0101] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0102] like Figure 1-5The diagram illustrates a zoned temperature control system and method for a cold storage facility. The system comprises a data acquisition module, a data processing module, a comparison module, a first calculation module, a generation module, a second calculation module, an adjustment module, a communication module, and a storage module. The data acquisition module is installed within N independent temperature-controlled zones (N≥2). Each zone is equipped with a temperature detection device. For each zone, the module continuously acquires actual temperature data from multiple points within it at preset time intervals, generating a raw temperature dataset. The data processing module cleans the generated raw temperature dataset, removing outliers and using interpolation algorithms to fill in missing data, generating a valid temperature dataset. This valid temperature dataset undergoes preprocessing, including data smoothing and standardization, to generate a standardized temperature data sequence. The comparison module compares the standardized temperatures of each zone... The data sequence is compared with its corresponding preset target temperature value to calculate the real-time temperature deviation sequence of each region. Calculation module one, based on the real-time temperature deviation sequence of each region, calculates the initial control quantity of the temperature regulating device in each region through a pre-built heat exchange correlation model. The generation module, based on historical temperature data and environmental parameter data, predicts the temperature change trend of each region in the future through a pre-trained machine learning model, generating a predicted temperature sequence. Calculation module two, based on the predicted temperature sequence, performs feedforward compensation calculation on the initial control quantity to obtain the final control quantity of the temperature regulating device in each region. The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulating device in each region, achieving regional temperature control. The communication module is responsible for communication between the various system modules. The storage module stores data during system operation.

[0103] like Figure 1 As shown, the implementation steps of the zoned temperature control system for the cold storage are as follows:

[0104] Step 1: Divide the internal space of the cold storage into N independent temperature control zones. Install a temperature detection device in each zone and connect it to a data acquisition module. The data acquisition module continuously collects the actual temperature data of multiple points inside each zone at preset time intervals to generate the raw temperature dataset. The data acquisition module covers the temperature distribution characteristics of different locations within the zone, such as goods stacking areas, airflow dead zones, and areas near the storage door, avoiding the problem of insufficient local temperature representativeness caused by traditional single-point or small-point data acquisition.

[0105] Step 2: The data processing module cleans the generated raw temperature dataset, removes outliers, and fills in missing data using an interpolation algorithm to generate a valid temperature dataset. The valid temperature dataset is then preprocessed, and a standardized temperature data sequence is generated through data smoothing and standardization.

[0106] Step 3: The comparison module compares the standardized temperature data sequence of each region with its corresponding preset target temperature value to calculate the real-time temperature deviation sequence of each region. Data cleaning removes outliers such as instantaneous sensor fault jump values, interpolation algorithms fill in missing data, and data holes caused by sensor obstruction or communication interruption are removed. Combined with data smoothing (such as moving average filtering) and standardization operations (such as normalization to a unified dimension), noise data in the high humidity and electromagnetic interference environment of cold storage is effectively filtered out, improving the integrity and reliability of temperature data. This provides a high-quality input basis for subsequent control quantity calculations and solves the problem of control decision deviation caused by data errors in existing technologies.

[0107] Step 4: Calculation Module 1 calculates the initial control quantity of the temperature regulation device in each region based on the real-time temperature deviation sequence of each region and through the pre-built heat exchange correlation model. Calculation Module 1 quantifies the correlation effects of airflow exchange and heat transfer between regions, such as the heat disturbance diffusion caused by the opening of doors in adjacent regions, avoiding the control conflict caused by ignoring the heat interaction between regions in traditional independent temperature control, such as the passive temperature rise of adjacent regions when a certain region is cooled.

[0108] Step 5: The generation module, based on historical temperature data and environmental parameter data, uses a pre-trained machine learning model to predict the temperature change trend of each region in the future, generating a predicted temperature sequence. The generation module, based on historical temperature data and environmental parameters such as door opening frequency, ambient temperature, and cargo heat release, uses a pre-trained machine learning model to predict the temperature change trend of each region in the future, and combines the prediction results to perform feedforward compensation on the initial control quantity. This upgrades the control strategy from "passively responding to current deviations" to "actively predicting future trends," significantly shortening the response time of temperature regulation. For example, after a door is opened, cooling power can be increased in advance to offset thermal disturbances, reducing control overshoot. It also avoids excessively low temperatures due to delayed regulation, improving the dynamic stability of the system.

[0109] Step 6: Calculation Module 2 performs feedforward compensation calculation on the initial control quantity based on the predicted temperature sequence to obtain the final control quantity of the temperature regulation device in each region.

[0110] Step 7: The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulation devices in each area, thereby achieving zoned temperature control.

[0111] The specific process of data cleaning of the original temperature dataset in step two of the implementation method of the cold storage zone temperature control system shown is as follows:

[0112] 1) Use the moving average method or median filtering method to smooth the original temperature dataset, and identify and remove outlier data points that exceed the preset confidence interval;

[0113] 2) Missing data points are filled in using linear interpolation based on temperature values ​​at adjacent time points or correlation interpolation based on temperature values ​​in adjacent regions;

[0114] 3) The following formula is used to identify outlier data points:

[0115]

[0116] in: This is the raw temperature data at the current time t;

[0117] The average temperature within a sliding window centered at t;

[0118] This represents the standard deviation of the temperature within that window.

[0119] The preset confidence coefficient has a value range of [2.5, 3.5].

[0120] like Figure 2 As shown, the specific process of building the pre-constructed heat exchange correlation model used in step four of the implementation method of the cold storage zone temperature control system is as follows:

[0121] 1) During the operation of the cold storage system, continuously collect the operating parameters of the temperature control devices in each area, the real-time temperature data of each area, and the ambient temperature data to build a training dataset;

[0122] 2) Perform feature engineering on the training dataset to extract feature variables including temperature change rate, cumulative running time of the regulating device, and ambient temperature difference;

[0123] 3) Using a regression analysis algorithm, based on characteristic variables and temperature change data of each region, a heat exchange coefficient matrix is ​​trained to describe the influence relationship of heat exchange between regions;

[0124] 4) Based on the heat exchange coefficient matrix, a heat exchange correlation model is constructed to calculate the temperature impact of any temperature change in one region on adjacent regions;

[0125] 5) The heat exchange correlation model is specifically represented as follows:

[0126]

[0127] in: This represents the temperature change in the i-th region caused by the operation of other regions.

[0128] Let be the heat exchange coefficient from region j to region i;

[0129] Let be the operating power of the temperature control device in region j.

[0130] like Figure 4 As shown, the process of calculating the initial control quantity using the pre-built heat exchange correlation model in step four of the implementation method of the cold storage zone temperature control system is as follows:

[0131] 1) For each region, the basic adjustment amount for that region is calculated using either a PID control algorithm or a fuzzy control algorithm based on its temperature deviation sequence;

[0132] 2) Input the basic adjustment amount into the heat exchange correlation model to calculate the predicted temperature impact on all adjacent areas when the temperature is adjusted in this area;

[0133] 3) Based on the predicted temperature impact, calculate the compensation adjustment required for all adjacent areas to offset the impact;

[0134] 4) The basic adjustment amount and the corresponding compensation adjustment amount are weighted and summed to obtain the initial control amount of the temperature control device in each zone;

[0135] 5) The basic adjustment calculation formula for the PID control algorithm is:

[0136]

[0137] in: Basic adjustment amount;

[0138] This represents the real-time temperature deviation.

[0139] , , These are the proportional, integral, and differential coefficients, respectively.

[0140] like Figure 3 As shown, the specific process of the pre-trained machine learning model used in step five of the implementation method of the cold storage zone temperature control system is as follows:

[0141] 1) Obtain historical operating data of the cold storage, including historical temperature data, historical environmental parameter data, historical operating parameter data of the regulating device, and the corresponding timestamps;

[0142] 2) Label the historical operating data, divide the data into time windows, and label the temperature value at the end of each time window as the prediction target;

[0143] 3) Divide the processed dataset into training set, validation set, and test set;

[0144] 4) Select the LSTM model as the initial prediction model, train the initial prediction model using the training set, and perform hyperparameter tuning using the validation set to obtain the pre-trained machine learning model.

[0145] 5) Use the test set to evaluate the prediction accuracy of the pre-trained machine learning model. If the accuracy meets the preset requirements, the model training is complete.

[0146] 6) The cell state update formula for the LSTM model is:

[0147]

[0148]

[0149]

[0150]

[0151] in: Output for the forget gate;

[0152] For input gate output;

[0153] Candidate cell state;

[0154] This represents the current state of the cell.

[0155] The hidden state from the previous moment;

[0156] For the current input;

[0157] For the sigmoid function;

[0158] It is the hyperbolic tangent function;

[0159] These are the weight parameters of the model;

[0160] This is the bias parameter.

[0161] The online update process of the pre-trained machine learning model used in step five of the implementation method of the cold storage zone temperature control system is as follows:

[0162] 1) Regularly add newly generated temperature data and environmental parameter data as new samples to the model's training dataset;

[0163] 2) When the number of new samples reaches a preset threshold, the model retraining process is triggered. The pre-trained machine learning model is incrementally trained or retrained using the updated training dataset to update the model parameters.

[0164] 3) The loss function for model retraining is the root mean square error, and its calculation formula is as follows:

[0165]

[0166] in: The number of samples;

[0167] This represents the actual temperature value of the i-th sample.

[0168] This represents the predicted temperature value corresponding to the model.

[0169] like Figure 5 As shown, the specific process of calculation module two performing feedforward compensation calculation on the initial control quantity based on the predicted temperature sequence in step six of the implementation method of the cold storage zone temperature control system is as follows:

[0170] 1) Compare the predicted temperature sequence with the corresponding preset target temperature value to generate a predicted temperature deviation sequence;

[0171] 2) Based on the predicted temperature deviation sequence, the feedforward compensation amount is calculated through the feedforward control algorithm. The feedforward compensation amount is used to offset the predicted temperature deviation in advance.

[0172] 3) The feedforward compensation amount is superimposed with the initial control amount to obtain the final control amount of the temperature regulation device in each zone;

[0173] 4) The formula for calculating the final control quantity is:

[0174]

[0175] in: For the final control quantity;

[0176] This is the initial control variable;

[0177] This is the feedforward compensation gain coefficient;

[0178] For the future Predicted temperature deviation at any given time.

[0179] Furthermore, the specific handling process for abnormalities encountered in the implementation method of the illustrated cold storage zone temperature control system is as follows:

[0180] 1) Monitor the actual temperature data of each area. When the actual temperature value of any area exceeds the preset safety threshold range and the duration of this state exceeds the preset duration, it is determined to be a temperature abnormality event.

[0181] 2) In response to an abnormal temperature event, an alarm signal is triggered, and the emergency control mode is activated;

[0182] 3) In emergency control mode, the prediction and feedforward compensation calculation of the machine learning model are suspended, and the final control quantity is directly generated by the maximum output power control algorithm based on the temperature deviation sequence.

[0183] 4) The formula for generating the control quantity in the maximum output power control algorithm is:

[0184]

[0185] in: For emergency control purposes;

[0186] This is the maximum output power of the temperature control device;

[0187] The safe temperature deviation threshold;

[0188] This is a normal PID control value.

[0189] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any modifications or equivalent changes made based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A zoned temperature control system for cold storage, characterized in that: The cold storage zoned temperature control system includes a data acquisition module, a data processing module, a comparison module, a calculation module 1, a generation module, a calculation module 2, an adjustment module, a communication module, and a storage module. The data acquisition module is installed within N independent temperature-controlled zones that divide the cold storage space. Each zone is equipped with a temperature detection device. For each zone, it continuously acquires actual temperature data from multiple points within that zone at preset time intervals, generating a raw temperature dataset. The data processing module cleans the generated raw temperature dataset, removing outliers and using interpolation algorithms to fill in missing data, generating a valid temperature dataset. This valid temperature dataset is then preprocessed, including data smoothing and standardization, to generate a standardized temperature data sequence. The comparison module compares the standardized temperature data sequence of each zone with... The system compares the corresponding preset target temperature values ​​to calculate the real-time temperature deviation sequence for each region. The calculation module one, based on the real-time temperature deviation sequence for each region, calculates the initial control quantity of the temperature regulating device for each region using a pre-built heat exchange correlation model. The generation module, based on historical temperature data and environmental parameter data, predicts the temperature change trend of each region in the future using a pre-trained machine learning model, generating a predicted temperature sequence. The calculation module two performs feedforward compensation calculation on the initial control quantity according to the predicted temperature sequence to obtain the final control quantity of the temperature regulating device for each region. The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulating device for each region, achieving regional temperature control. The communication module is responsible for communication between the various system modules. The storage module stores data generated during system operation.

2. The method for a zoned temperature control system for cold storage according to claim 1, characterized in that: The implementation steps of the cold storage zoned temperature control system are as follows: Step 1: Divide the internal space of the cold storage into N independent temperature control zones. Install a temperature detection device in each zone and connect it to a data acquisition module. The data acquisition module continuously collects the actual temperature data of multiple points inside each zone at preset time intervals to generate the original temperature dataset. Step 2: The data processing module cleans the generated raw temperature dataset, removes outliers, and fills in missing data using an interpolation algorithm to generate a valid temperature dataset. The valid temperature dataset is then preprocessed, and a standardized temperature data sequence is generated through data smoothing and standardization. Step 3: The comparison module compares the standardized temperature data sequence of each region with its corresponding preset target temperature value to calculate the real-time temperature deviation sequence of each region. Step 4: Calculation Module 1 calculates the initial control quantity of the temperature regulation device in each region based on the real-time temperature deviation sequence of each region and through a pre-built heat exchange correlation model. Step 5: The generation module uses historical temperature data and environmental parameter data to predict the temperature change trend of each region in the future through a pre-trained machine learning model, and generates a predicted temperature sequence. Step 6: Calculation Module 2 performs feedforward compensation calculation on the initial control quantity based on the predicted temperature sequence to obtain the final control quantity of the temperature regulation device in each region. Step 7: The adjustment module generates corresponding control commands based on the final control quantity to adjust the operating parameters of the temperature regulation devices in each area, thereby achieving zoned temperature control.

3. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The specific process of data cleaning of the original temperature dataset by the data processing module in step two of the implementation method of the cold storage zone temperature control system is as follows: 1) Use the moving average method or median filtering method to smooth the original temperature dataset, and identify and remove outlier data points that exceed the preset confidence interval; 2) Missing data points are filled in using linear interpolation based on temperature values ​​at adjacent time points or correlation interpolation based on temperature values ​​in adjacent regions; 3) The following formula is used to identify outlier data points: ； in: This is the raw temperature data at the current time t; The average temperature within a sliding window centered at t; This represents the standard deviation of the temperature within that window. The preset confidence coefficient has a value range of [2.5, 3.5].

4. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The specific process of building the pre-constructed heat exchange correlation model used in the calculation module one of the implementation method of the cold storage zone temperature control system is as follows: 1) During the operation of the cold storage system, continuously collect the operating parameters of the temperature control devices in each area, the real-time temperature data of each area, and the ambient temperature data to build a training dataset; 2) Perform feature engineering on the training dataset to extract feature variables including temperature change rate, cumulative running time of the regulating device, and ambient temperature difference; 3) Using a regression analysis algorithm, based on characteristic variables and temperature change data of each region, a heat exchange coefficient matrix is ​​trained to describe the influence relationship of heat exchange between regions; 4) Based on the heat exchange coefficient matrix, a heat exchange correlation model is constructed to calculate the temperature impact of any temperature change in one region on adjacent regions; 5) The heat exchange correlation model is specifically represented as follows: ； in: This represents the temperature change in the i-th region caused by the operation of other regions. Let be the heat exchange coefficient from region j to region i; Let be the operating power of the temperature control device in region j.

5. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The process of calculating the initial control quantity using the pre-built heat exchange correlation model in step four of the implementation method of the cold storage zone temperature control system is as follows: 1) For each region, the basic adjustment amount for that region is calculated using either a PID control algorithm or a fuzzy control algorithm based on its temperature deviation sequence; 2) Input the basic adjustment amount into the heat exchange correlation model to calculate the predicted temperature impact on all adjacent areas when the temperature is adjusted in this area; 3) Based on the predicted temperature impact, calculate the compensation adjustment required for all adjacent areas to offset the impact; 4) The basic adjustment amount and the corresponding compensation adjustment amount are weighted and summed to obtain the initial control amount of the temperature control device in each zone; 5) The basic adjustment calculation formula for the PID control algorithm is: ； in: Basic adjustment amount; This represents the real-time temperature deviation. , , These are the proportional, integral, and differential coefficients, respectively.

6. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The specific process of the pre-trained machine learning model used in step five of the implementation method of the cold storage zone temperature control system is as follows: 1) Obtain historical operating data of the cold storage, including historical temperature data, historical environmental parameter data, historical operating parameter data of the regulating device, and the corresponding timestamps; 2) Label the historical operating data, divide the data into time windows, and label the temperature value at the end of each time window as the prediction target; 3) Divide the processed dataset into training set, validation set, and test set; 4) Select the LSTM model as the initial prediction model, train the initial prediction model using the training set, and perform hyperparameter tuning using the validation set to obtain the pre-trained machine learning model. 5) Use the test set to evaluate the prediction accuracy of the pre-trained machine learning model. If the accuracy meets the preset requirements, the model training is complete. 6) The cell state update formula for the LSTM model is: ； ； ； ； in: Output for the forget gate; For input gate output; Candidate cell state; This represents the current state of the cell. The hidden state from the previous moment; For the current input; For the sigmoid function; It is the hyperbolic tangent function; These are the weight parameters of the model; This is the bias parameter.

7. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The online update process of the pre-trained machine learning model used in step five of the implementation method of the cold storage zone temperature control system is as follows: 1) Regularly add newly generated temperature data and environmental parameter data as new samples to the model's training dataset; 2) When the number of new samples reaches a preset threshold, the model retraining process is triggered, and the pre-trained machine learning model is incrementally trained or retrained using the updated training dataset to update the model parameters. 3) The loss function for model retraining is the root mean square error, and its calculation formula is as follows: ； in: The number of samples; This represents the actual temperature value of the i-th sample. This represents the predicted temperature value corresponding to the model.

8. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The specific process of calculating the feedforward compensation of the initial control quantity based on the predicted temperature sequence in step six of the implementation method of the cold storage zone temperature control system is as follows: 1) Compare the predicted temperature sequence with the corresponding preset target temperature value to generate a predicted temperature deviation sequence; 2) Based on the predicted temperature deviation sequence, the feedforward compensation amount is calculated through the feedforward control algorithm. The feedforward compensation amount is used to offset the predicted temperature deviation in advance. 3) The feedforward compensation amount is superimposed with the initial control amount to obtain the final control amount of the temperature regulation device in each zone; 4) The formula for calculating the final control quantity is: ； in: For the final control quantity; This is the initial control variable; This is the feedforward compensation gain coefficient; For the future Predicted temperature deviation at any given time.

9. The method for a zoned temperature control system for cold storage according to claim 2, characterized in that: The specific procedure for handling abnormalities encountered in the implementation method of the cold storage zoned temperature control system is as follows: 1) Monitor the actual temperature data of each area. When the actual temperature value of any area exceeds the preset safety threshold range and the duration of this state exceeds the preset duration, it is determined to be a temperature abnormality event. 2) In response to an abnormal temperature event, an alarm signal is triggered, and the emergency control mode is activated; 3) In emergency control mode, the prediction and feedforward compensation calculation of the machine learning model are suspended, and the final control quantity is directly generated by the maximum output power control algorithm based on the temperature deviation sequence. 4) The formula for generating the control quantity in the maximum output power control algorithm is: ； in: For emergency control purposes; This is the maximum output power of the temperature control device; The safe temperature deviation threshold; This is a normal PID control value.

Images