Iot-based refrigeration equipment defrosting and refrigeration collaborative control method and system

Abstract

The present application relates to the technical field of refrigeration equipment control, and particularly relates to a refrigeration equipment defrosting and refrigeration collaborative control method and system based on the Internet of Things. The present application collects the temperature and environmental parameters of multiple parts of the refrigeration equipment in real time through the Internet of Things sensor, constructs a multi-dimensional data structure describing the evolution of the thermal state of the equipment, and dynamically quantifies the degree of deviation of the equipment from the steady-state thermal balance; performs refrigeration pre-strengthening before defrosting to offset the subsequent thermal shock, and after defrosting, uses a multi-actuator staged and time-gradual progressive recovery strategy to suppress the temperature overshoot caused by thermal hysteresis; at the same time, the historical operation data is used to perform cross-cycle self-optimization on the control model, so that the control precision continuously improves with the operation time. The present application controls the temperature fluctuation in the freezer during the defrosting process within a very narrow range, significantly reduces the defrosting energy consumption, prolongs the service life of the equipment, and is suitable for application scenarios such as pharmaceutical cold chain, vaccine storage, blood product refrigeration and ultra-low temperature freezing.

Description

Technical Field

[0001] This invention relates to the field of refrigeration equipment control technology, and in particular to a method and system for coordinated control of defrosting and refrigeration of refrigeration equipment based on the Internet of Things. Background Technology

[0002] During operation, frost buildup on the evaporator surface of refrigeration equipment gradually reduces heat exchange efficiency, leading to decreased refrigeration performance and increased energy consumption. Therefore, regular defrosting is essential. Currently, mainstream defrosting control strategies fall into two categories: one is timed defrosting, which initiates defrosting at fixed time intervals. This method completely disregards the actual frost conditions, resulting in energy waste when frost is light and incomplete defrosting when frost is heavy. The other is threshold-triggered defrosting based on a single temperature parameter, initiating defrosting when the evaporator surface temperature and the pressure difference across the evaporator reach a preset value. However, both of these existing technologies have significant drawbacks: firstly, a single trigger parameter cannot fully reflect the true frost state on the evaporator surface; the density, thickness, and morphology of the frost layer vary under different operating conditions. The defrosting process itself is a strong thermal disturbance. When switching back to refrigeration mode after defrosting, the accumulated heat does not disappear immediately but is released slowly, causing the temperature inside the storage room to continue to rise after defrosting, resulting in significant overshoot. For scenarios such as pharmaceutical cold chain, vaccine storage, and blood product refrigeration, which are extremely sensitive to temperature fluctuations, this overshoot is enough to cause drug failure and vaccine spoilage. Thirdly, the existing system lacks the ability to adapt to the long-term performance degradation of the equipment. The fixed parameters set at the factory are no longer applicable after the equipment has been in service for several months or years, and the control quality declines year by year. Therefore, there is an urgent need to develop a method and system for coordinated control of defrosting and refrigeration of refrigeration equipment based on the Internet of Things. Summary of the Invention

[0003] To overcome the shortcomings of single-temperature adjustment for defrosting and cooling, this invention provides a method and system for coordinated control of defrosting and cooling of refrigeration equipment based on the Internet of Things.

[0004] The technical solution is: an IoT-based method for coordinated control of defrosting and cooling in refrigeration equipment, comprising the following steps:

[0005] S1: Obtain multi-source data from the ultra-low temperature cold storage, construct original observation vectors at different times based on the multi-source data, and construct thermal inertia feature vectors at different times based on the original observation vectors;

[0006] S2: Preset thermal inertial nodes, obtain the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes; calculate the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes and construct the thermal inertial state matrix;

[0007] S3: Calculate the thermal inertia offset and offset change rate based on the thermal inertia state matrix, preset the defrost trigger threshold, and determine defrost based on the comparison between the thermal inertia offset, offset change rate and defrost trigger threshold;

[0008] S4: Calculate the pre-compensated cooling capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtain the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determine the cooling by setting the defrosting termination threshold.

[0009] S5: Obtain the recovery priority index of each execution unit in the refrigeration stage, obtain the recovery unit sequence based on the recovery priority index, and restore the refrigeration of the cold storage based on the recovery unit sequence.

[0010] Preferably, the step of acquiring multi-source data of the ultra-low temperature cold storage, constructing original observation vectors at different times based on the multi-source data, and constructing thermal inertia feature vectors at different times based on the original observation vectors includes: acquiring multi-source data of the ultra-low temperature cold storage and the evaporator inlet and outlet pressure difference through an IoT sensor group at a fixed sampling interval; the multi-source data includes the evaporator coil surface temperature, the average air temperature inside the cold storage, the average temperature of the inner wall of the cold storage, and the compressor suction temperature; constructing the original observation vector of the ultra-low temperature cold storage at the current time based on the multi-source data; and obtaining the thermal inertia feature vector of the ultra-low temperature cold storage at the current time by differentiating the original observation vector at a fixed sampling interval. The thermal inertia feature vector is a vector representing the ability of the ultra-low temperature cold storage to represent the trend of thermal inertia changes. The thermal inertia feature vector includes the original observation vector at the current time, the first derivative of the original observation vector at the current time, and the second derivative of the original observation vector at the current time.

[0011] Preferably, the preset thermal inertial nodes, based on the thermal inertial feature vector and the thermal inertial nodes, obtain the feature input quantities of the thermal inertial nodes, including: preset thermal inertial nodes based on multi-source data sources of the ultra-low temperature cold storage, wherein the thermal inertial nodes include evaporator nodes, air nodes inside the cold storage, wall nodes of the cold storage, and compression equivalent nodes; preset the heat accumulation integration window by sampling time; and obtain the feature input quantities of different thermal inertial nodes by mapping the thermal inertial feature vector and the heat accumulation integration window. The feature input quantities are the input quantities of different thermal inertial nodes based on different defrosting and cooling forces, wherein the calculation formula of the feature input quantities is:

[0012] ;

[0013] ;

[0014] ;

[0015] ;

[0016] ;

[0017] ;

[0018] ;

[0019] ;

[0020] in, For the first Nodes in the first dimension, in The feature input at time step This represents the sequence number of the sampling time. For a fixed sampling interval, For the first Node at The basic quantity at any given time. For the first Nodes in the second dimension, in The feature input at time step For the first Nodes in the third dimension, in The feature input at time step For the first Nodes in the fourth dimension, in The feature input at time step For the heat accumulation integral window, This refers to the time sequence number within the heat accumulation integration window. In order to be in The surface temperature of the evaporator coil at any given time. In order to be in The average temperature of the air inside the warehouse at any given time. In order to be in The average temperature of the inner wall of the tank at any given time. In order to be in The compressor intake temperature at any given time.

[0021] Preferably, the step of calculating the real-time thermal inertia matching degree and constructing the thermal inertia state matrix based on the characteristic input quantities of the thermal inertia nodes includes: obtaining the baseline center value and data standard deviation of the thermal inertia nodes in the current dimension under standard operating conditions for the ultra-low temperature cold storage, wherein the baseline center value is the standard quantity of the ultra-low temperature cold storage at the current position under standard operating conditions; calculating the real-time thermal inertia matching degree of the thermal inertia nodes at the current moment based on the baseline center value, data standard deviation, and characteristic input quantities of the thermal inertia nodes, wherein the real-time thermal inertia matching degree is the response intensity of the thermal inertia nodes to thermal inertia; and constructing the thermal inertia state matrix of the ultra-low temperature cold storage at the current moment based on the real-time thermal inertia matching degrees of different thermal inertia nodes, wherein the calculation formula for the real-time thermal inertia matching degree is:

[0022] ;

[0023] in, For the first Node at the Dimension, in Real-time thermal inertia matching degree at any given moment. For the first Node at the Dimensional thermal inertia weighting coefficient For the first Node at the Dimension, in The feature input at time step For the first Node at the The baseline center value of the dimension, For the first Node at the Standard deviation of data in a given dimension.

[0024] Preferably, the step of calculating the thermal inertia offset and offset change rate based on the thermal inertia state matrix, preset a defrost trigger threshold, and determine defrost based on a comparison of the thermal inertia offset, offset change rate, and defrost trigger threshold includes: obtaining a reference thermal inertia state matrix under standard operating conditions for the ultra-low temperature cold storage; calculating the thermal inertia offset of the ultra-low temperature cold storage based on the thermal inertia state matrix and the reference thermal inertia state matrix, wherein the thermal inertia offset is the offset of thermal energy relative to standard operating conditions during cold storage refrigeration; calculating the offset change rate based on the thermal inertia offset, wherein the offset change rate is the offset change rate of thermal energy under standard operating conditions; preset a defrost trigger threshold at the current moment based on the evaporator inlet and outlet pressure difference; and determining the defrost stage of the current ultra-low temperature cold storage by comparing the thermal inertia offset, offset change rate, and defrost trigger threshold. The determination condition is: if the thermal inertia offset is greater than or equal to the defrost trigger threshold at the current moment and the offset change rate is greater than zero, then the current ultra-low temperature cold storage enters the defrost stage; wherein the calculation formulas for the thermal inertia offset and offset change rate are:

[0025] ;

[0026] ;

[0027] in, For at any time Thermal inertia offset, For at any time The thermal inertial state matrix, This is the reference thermal inertia state matrix under standard operating conditions. Let be the square of the Euclidean norm. The thermal inertial response coefficient, For a fixed sampling interval, For at any time The rate of change of offset.

[0028] Preferably, the step of presetting the defrost trigger threshold at the current moment based on the pressure difference between the evaporator inlet and outlet includes: obtaining a basic defrost trigger threshold based on the ultra-low temperature cold storage under standard operating conditions, wherein the basic defrost trigger threshold is the minimum thermal inertia offset for defrosting the ultra-low temperature cold storage under standard operating conditions; obtaining the reference pressure difference of the ultra-low temperature cold storage in a completely frost-free state based on the ultra-low temperature cold storage; obtaining the safety inflection point value of the fan through the fan operation manual of the ultra-low temperature cold storage, wherein the safety inflection point value is the maximum allowable pressure difference of the fan; and calculating the defrost trigger threshold of the current ultra-low temperature cold storage at different times based on the basic defrost trigger threshold and the safety inflection point value, wherein the calculation formula for the defrost trigger threshold is:

[0029] ;

[0030] in, For ultra-low temperature cold storage at all times The defrost trigger threshold, Based on the basic defrost trigger threshold The attenuation coefficient is affected by the pressure difference. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine.

[0031] Preferably, the step of calculating the pre-compensated cooling capacity and compressor operating frequency for the defrosting stage based on the thermal inertia offset, obtaining the thermal deviation of the defrosting process based on the current thermal inertia state matrix, and determining refrigeration based on a preset defrosting termination threshold includes: before the ultra-low temperature cold storage enters the defrosting stage, a thermal inertia pre-compensation stage is first performed; the estimated defrosting duration is calculated based on the evaporator inlet and outlet pressure difference, the reference pressure difference, and the safety inflection point value; the pre-compensated cooling capacity for the defrosting stage is calculated based on the estimated defrosting duration and the thermal inertia offset, wherein the pre-compensated cooling capacity is the compensation amount of cooling capacity required during the defrosting of the cold storage; the compressor operating frequency of the ultra-low temperature cold storage is calculated based on the pre-compensated cooling capacity; and the cold storage operates according to the compressor operating frequency during the thermal inertia pre-compensation stage; after the thermal inertia pre-compensation stage ends, the defrosting stage is performed, and the current... The thermal inertia state matrix at any given time is used as the thermal inertia prediction matrix. This prediction matrix is ​​the predicted thermal inertia state matrix after defrosting. The thermal deviation of the defrosting process is calculated based on the difference between the thermal inertia prediction matrix and the thermal inertia state matrix. The defrosting change rate is calculated based on the thermal deviation. A preset defrosting termination threshold is set. The defrosting stage ends and the cooling stage begins by comparing the defrosting change rate with the defrosting termination threshold. The determination conditions are: if the defrosting stage continues while issuing a warning, and if the defrosting change rate is less than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage ends and the cooling stage begins; if the defrosting change rate is greater than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage continues while issuing a warning. The calculation formulas for defrosting duration, pre-compensated cooling capacity, compressor operating frequency, and defrosting change rate are as follows:

[0032] ;

[0033] ;

[0034] ;

[0035] ;

[0036] ;

[0037] in, For at any time The estimated defrosting time, Based on the defrosting time, This is the duration conversion factor. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine. For at any time Pre-compensation cooling capacity, This refers to the thermal inertia node number. For the first The current temperature of each thermal inertial node. For the first The target set temperature for each thermal inertial node. This is the offset compensation coefficient. For at any time Thermal inertia offset, For the compressor operating frequency, For the maximum permissible frequency, The current operating frequency, This is the conversion factor for cooling capacity. For the first The equivalent heat capacity of each thermal inertial node. For at any time defrosting change rate For at any time Thermal deviation in the defrosting process For a fixed sampling interval, For at any time The thermal inertial state matrix, To predict the thermal inertial state matrix, It is the square of the norm.

[0038] Preferably, the step of obtaining the recovery priority index of each execution unit during the refrigeration phase, obtaining the recovery unit sequence based on the recovery priority index, and restoring cold storage refrigeration based on the recovery unit sequence includes: performing the refrigeration phase after the defrosting phase; obtaining control units by querying the control objects where the thermal inertia nodes in the ultra-low temperature cold storage are located; calculating the recovery priority index of each control unit, where the recovery priority index is the priority index for the control unit to recover from the defrosting state to the normal refrigeration state; sorting the control units in descending order based on the recovery priority index to obtain a collaborative scheduling queue; adjusting the recovery of different control units to the refrigeration phase based on the collaborative scheduling queue; and correcting the defrosting period by updating and iterating the thermal inertia weight matrix of the ultra-low temperature cold storage; wherein the formula for calculating the recovery priority index is:

[0039] ;

[0040] in, For the first Each control unit at time Recovery priority index, The serial number of the control unit is 1. For the first Each control unit at time Temperature deviation, This represents the maximum temperature deviation among all units. For the first Thermal inertia relaxation time constant of each control unit The maximum relaxation time constant among all elements. For the first Each control unit at time Thermal inertia offset, This represents the maximum thermal inertia offset among all elements. , and This is the priority weighting coefficient.

[0041] Preferably, the step of correcting the defrosting period by updating and iterating the thermal inertia weight matrix of the ultra-low temperature cold storage includes: flattening the thermal inertia weight matrix to obtain a thermal inertia weight vector; obtaining sensitivity vectors of different nodes and dimensions at different snapshot sampling times according to the snapshot sampling times within the optimization period; the optimization period being the total time from the start of data acquisition to the resumption of refrigeration; the snapshot sampling time being the interval for saving the thermal inertia weight matrix once; the sensitivity vector being the vector representing the change in the cold storage's sensitivity to heat with the number of iterations; calculating the Kalman gain vector at different snapshot sampling times based on the sensitivity vectors at different snapshot sampling times; the Kalman gain vector being the change in weight coefficients in each iteration; and calculating the thermal inertia weight vector at the current snapshot sampling time based on the Kalman gain vector, the sensitivity vector, and the thermal inertia weight vector at the previous snapshot sampling time, wherein the formula for calculating the thermal inertia weight matrix at the current snapshot sampling time is:

[0042] ;

[0043] ;

[0044] ;

[0045] ;

[0046] in, For the current optimization cycle, the first Thermal inertia weight vector at the next snapshot sampling time For the current optimization cycle, the first The average temperature of the air inside the storage room at the time of the next snapshot sampling. Set the target temperature for the air inside the storage facility. For the current optimization cycle, the first The first snapshot sampling time Node at the Dimensional thermal inertia weighting coefficient For the current optimization cycle, the first The covariance matrix of the thermal inertia weight matrix at the next snapshot sampling time. It is the identity matrix. For the current optimization cycle, the first Kalman gain vector at the next snapshot sampling time For the current optimization cycle, the first The sensitivity vector at the next snapshot sampling time. For transpose, For the current optimization cycle, the first The first snapshot sampling time Node at the Sensitivity to dimensions.

[0047] Preferably, the IoT-based defrosting and refrigeration coordinated control system for refrigeration equipment further includes:

[0048] The data acquisition module acquires multi-source data from the ultra-low temperature cold storage, constructs original observation vectors at different times based on the multi-source data, and constructs thermal inertia feature vectors at different times based on the original observation vectors.

[0049] The feature input module presets thermal inertial nodes, obtains the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes, calculates the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes, and constructs the thermal inertial state matrix.

[0050] The defrosting module calculates the thermal inertia offset and offset change rate based on the thermal inertia state matrix, presets a defrosting trigger threshold, and determines defrosting based on a comparison between the thermal inertia offset, offset change rate, and defrosting trigger threshold.

[0051] The refrigeration module is determined by calculating the pre-compensated refrigeration capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtaining the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determining the refrigeration by setting a defrosting termination threshold.

[0052] The refrigeration module is restored, the recovery priority index of each execution unit in the refrigeration stage is obtained, the recovery unit sequence is obtained based on the recovery priority index, and the refrigeration of the cold storage is restored based on the recovery unit sequence.

[0053] The beneficial effects are:

[0054] 1. This invention accurately senses the thermal state of the equipment and performs pre-cooling enhancement in advance. At the same time, it adopts a multi-component progressive recovery strategy when defrosting is completed, which narrows the temperature fluctuation in the warehouse during the entire defrosting cycle from the traditional solution to an extremely narrow range, especially meeting the stringent requirements for temperature stability in pharmaceutical cold chain and vaccine storage.

[0055] 2. This invention reduces the ineffective heating and excessive defrosting in light frost conditions by initiating defrosting only when the equipment's thermal state deviates from the actual requirement. This effectively reduces the electrical energy consumed in the defrosting process and related refrigeration compensation, and significantly reduces long-term electricity costs.

[0056] 3. This invention continuously optimizes internal control parameters by utilizing historical data generated during the equipment's operation, enabling the system to maintain factory-grade high-precision control quality even under conditions such as compressor aging, minor refrigerant leakage, and heat exchanger blockage. This eliminates the need for periodic manual recalibration, significantly reducing maintenance and management costs and technical barriers. Attached Figure Description

[0057] Figure 1 This is a schematic flowchart of the IoT-based refrigeration equipment defrosting and refrigeration coordinated control method of the present invention.

[0058] Figure 2 This is a schematic diagram of the structure of the IoT-based defrosting and refrigeration coordinated control method and system for refrigeration equipment according to the present invention. Detailed Implementation

[0059] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0060] Example 1: A method for coordinated control of defrosting and cooling in refrigeration equipment based on the Internet of Things, such as Figure 1 As shown, it includes the following steps:

[0061] S1: Obtain multi-source data from the ultra-low temperature cold storage, construct original observation vectors at different times based on the multi-source data, and construct thermal inertia feature vectors at different times based on the original observation vectors;

[0062] S2: Preset thermal inertial nodes, obtain the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes; calculate the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes and construct the thermal inertial state matrix;

[0063] S3: Calculate the thermal inertia offset and offset change rate based on the thermal inertia state matrix, preset the defrost trigger threshold, and determine defrost based on the comparison between the thermal inertia offset, offset change rate and defrost trigger threshold;

[0064] S4: Calculate the pre-compensated cooling capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtain the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determine the cooling by setting the defrosting termination threshold.

[0065] S5: Obtain the recovery priority index of each execution unit in the refrigeration stage, obtain the recovery unit sequence based on the recovery priority index, and restore the refrigeration of the cold storage based on the recovery unit sequence.

[0066] Multi-source data and evaporator inlet and outlet pressure differences of the ultra-low temperature cold storage are acquired through an IoT sensor array at fixed sampling intervals. The multi-source data includes the surface temperature of the evaporator coil, the average temperature of the air inside the cold storage, the average temperature of the inner wall of the cold storage, and the compressor suction temperature. Based on the multi-source data, the original observation vector of the ultra-low temperature cold storage at the current moment is constructed. The derivative of the original observation vector is calculated at a fixed sampling interval to obtain the thermal inertia feature vector of the ultra-low temperature cold storage at the current moment. The thermal inertia feature vector is the vector representing the thermal inertia change trend of the ultra-low temperature cold storage. The thermal inertia feature vector includes the original observation vector at the current moment, the first derivative of the original observation vector at the current moment, and the second derivative of the original observation vector at the current moment.

[0067] It should be explained that the evaporator coil surface temperature, average air temperature inside the cold storage, average internal wall temperature, and compressor suction temperature are obtained as follows: The IoT sensor group consists of sensor nodes deployed in key parts of the cold storage. The edge computing controller is installed in the cold storage equipment room and connects to each sensor node via industrial Ethernet. At the same time, it interacts with the remote monitoring platform through a communication module to obtain multi-source data of the ultra-low temperature cold storage and the pressure difference between the inlet and outlet of the evaporator. The fixed sampling interval is the minimum sampling interval of all sensors, with the dimension in seconds. The dimension of the thermal inertia feature vector is the number of multi-source data types multiplied by the differential quantity type, which includes the current time, the first derivative of the current time, and the second derivative of the current time. The dimensions of the evaporator coil surface temperature, average air temperature inside the cold storage, average internal wall temperature, and compressor suction temperature are in Kelvin. The unit of the first derivative is Kelvin per second, and the unit of the second derivative is Kelvin per second squared.

[0068] Based on multi-source data from the ultra-low temperature cold storage, pre-set thermal inertial nodes are established. These nodes include evaporator nodes, internal air nodes, cold storage wall nodes, and compression equivalent nodes. A thermal accumulation integration window is pre-set at each sampling time. The feature input quantities of different thermal inertial nodes are obtained by mapping the thermal inertial feature vector and the thermal accumulation integration window. These feature input quantities are the input quantities of different thermal inertial nodes based on different defrosting and cooling capacities. The calculation formula for the feature input quantities is as follows:

[0069] ;

[0070] ;

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] in, For the first Nodes in the first dimension, in The feature input at time step This represents the sequence number of the sampling time. For a fixed sampling interval, For the first Node at The basic quantity at any given time. For the first Nodes in the second dimension, in The feature input at time step For the first Nodes in the third dimension, in The feature input at time step For the first Nodes in the fourth dimension, in The feature input at time step For the heat accumulation integral window, This refers to the time sequence number within the heat accumulation integration window. In order to be in The surface temperature of the evaporator coil at any given time. In order to be in The average temperature of the air inside the warehouse at any given time. In order to be in The average temperature of the inner wall of the tank at any given time. In order to be in The compressor intake temperature at any given time.

[0078] It should be explained that the four thermal inertia nodes are the evaporator node, the internal air node, the cold storage wall node, and the compression equivalent node. The evaporator node corresponds to the evaporator coil and the frost layer attached to it; the internal air node corresponds to the overall heat capacity of the circulating air inside the cold storage; the cold storage wall node corresponds to the overall thermal inertia of the six-sided enclosure structure of the cold storage; and the compression equivalent node corresponds to the equivalent lumped heat capacity of the vaccines and blood products stored in the cold storage. The heat accumulation integral window is a fixed-length sliding time window in seconds, obtained by: the controller performing an offline thermodynamic identification experiment during the system initialization phase. The experiment involves applying a step-like thermal power disturbance at the evaporator inlet and recording the time required for the evaporator coil surface temperature to reach 60% of the new steady-state value from the start of the disturbance. This time is the thermal inertia-dominant time constant. After repeated experiments and data processing, the upper and lower limits of the thermal inertia-dominant time constant in this experiment are obtained, and the average of the upper and lower limits is taken. The value serves as the thermal accumulation integration window; the base quantity is the fundamental value for calculating the feature input quantities of different dimensions at the current node, in Kelvin. The calculation of each set of feature input quantities starts from the trigger interrupt at the current sampling time and is calculated based on the base quantity of the current thermal inertial node at the current sampling time. The feature input quantities are organized into a four-row, four-column feature input quantity matrix according to the node index and dimension index, where the row index represents different thermal inertial nodes and the column index represents different thermal state dimensions. In the entire feature input quantity mapping process, the original data buffer adopts a first-in-first-out queue structure, and each enqueued data has a hardware timestamp to ensure that when calculating the thermal accumulation integration window, the time interval of each sampling point in the window is strictly equal to the sampling interval, and there is no time drift or cumulative error. At the same time, the controller executes a complete feature input quantity mapping calculation in each sampling interrupt service routine to ensure that a complete four-by-four feature input quantity matrix can be obtained for each sampling interval.

[0079] Based on the ultra-low temperature cold storage under standard operating conditions, the baseline center value and data standard deviation of the thermal inertia node in the current dimension are obtained. The baseline center value is the standard quantity of the ultra-low temperature cold storage at the current position under standard operating conditions. The real-time thermal inertia matching degree of the thermal inertia node at the current moment is calculated based on the baseline center value, data standard deviation, and characteristic input quantities of the thermal inertia node. The real-time thermal inertia matching degree is the response intensity of the thermal inertia node to thermal inertia. The thermal inertia state matrix of the ultra-low temperature cold storage at the current moment is constructed based on the real-time thermal inertia matching degrees of different thermal inertia nodes. The formula for calculating the real-time thermal inertia matching degree is:

[0080] ;

[0081] in, For the first Node at the Dimension, in Real-time thermal inertia matching degree at any given moment. For the first Node at the Dimensional thermal inertia weighting coefficient For the first Node at the Dimension, in The feature input at time step For the first Node at the The baseline center value of the dimension, For the first Node at the Standard deviation of data in a given dimension.

[0082] It should be explained that the standard operating condition refers to the cold storage under no-load conditions, with the ambient temperature and humidity specified by national standards, the door kept closed, no defrosting operation, and the refrigeration system running continuously at rated capacity until the internal temperature reaches the set value. The data acquisition duration for the standard operating condition is recorded, and the baseline center values ​​for the four dimensions are calculated based on this duration. The baseline center values ​​are obtained by continuously acquiring and storing all original observation vector data at fixed sampling intervals under the standard operating condition. For each thermal inertial node, all data points are obtained in each dimension, and the arithmetic mean of these data points is calculated as the baseline center value for that node in that dimension. The baseline center values ​​have the same physical dimensions as the corresponding feature input quantities. The baseline center values ​​for the first dimension are in Kelvin, the second dimension in Kelvin per second, the third dimension in Kelvin per second, and the fourth dimension in Kelvin per second. The unit is squared per second; the standard deviation of the data is obtained as follows: under the same standard operating conditions, for the feature input data of each dimension, the standard deviation of each data point and the baseline center value is calculated as the standard deviation of the data at that node for that dimension. The standard deviation of the data has the same physical dimension as the corresponding feature input. The judgment interval for normal fluctuation is the upper limit of the baseline center value plus one standard deviation and the lower limit of the baseline center value minus one standard deviation; the real-time thermal inertia matching degree is a dimensionless positive real number, and the unit of the thermal inertia weight coefficient is dimensionless. Specifically, it includes sixteen weight coefficients for different nodes and different dimensions. The thermal inertia weight coefficient is obtained as follows: in the first defrosting cycle, all weight coefficients are the same and equal to one. The thermal inertia weight of each node in different dimensions is the reciprocal of the number of all weight coefficients; the thermal inertia state matrix is ​​a square matrix. The row index of the matrix corresponds to the number of thermal inertia nodes, and the column index of the matrix corresponds to the number of dimensions.

[0083] The baseline thermal inertia state matrix under standard operating conditions is obtained for the ultra-low temperature cold storage. The thermal inertia offset of the ultra-low temperature cold storage is calculated based on the thermal inertia state matrix and the baseline thermal inertia state matrix. This thermal inertia offset represents the shift of thermal energy relative to the standard operating conditions during refrigeration. The rate of change of offset is calculated based on this offset, which is the rate of change of thermal energy under standard operating conditions. A defrost trigger threshold is preset based on the evaporator inlet and outlet pressure difference. The defrost stage of the ultra-low temperature cold storage is determined by comparing the thermal inertia offset, the rate of change of offset, and the defrost trigger threshold. The determination condition is: if the thermal inertia offset is greater than or equal to the defrost trigger threshold and the rate of change of offset is greater than zero, then the ultra-low temperature cold storage enters the defrost stage. The formulas for calculating the thermal inertia offset and the rate of change of offset are as follows:

[0084] ;

[0085] ;

[0086] in, For at any time Thermal inertia offset, For at any time The thermal inertial state matrix, This is the reference thermal inertia state matrix under standard operating conditions. Let be the square of the Euclidean norm. The thermal inertial response coefficient, For a fixed sampling interval, For at any time The rate of change of offset.

[0087] It should be explained that the other judgment conditions for the defrosting stage are as follows: if the thermal inertia offset is greater than or equal to the defrosting trigger threshold at the current moment and the offset change rate is less than zero, then the current ultra-low temperature cold storage is operating normally; if the thermal inertia offset is less than the defrosting trigger threshold at the current moment and the offset change rate is greater than zero, then the current ultra-low temperature cold storage is operating normally; if the thermal inertia offset is less than the defrosting trigger threshold at the current moment and the offset change rate is less than zero, then the current ultra-low temperature cold storage is operating normally; the dimension of the thermal inertia offset is a dimensionless non-negative scalar, and the dimension of the reference thermal inertia state matrix is ​​the same as that of the thermal inertia state matrix. The reference thermal inertia state matrix is ​​obtained by statistically averaging all thermal inertia state matrices during steady-state operation to obtain a four-row, four-column matrix. The quasi-thermal inertia state matrix; the thermal inertia response coefficient is the time weight of the contribution of the rate of change of thermal inertia offset to the total offset, with the dimension of seconds. It is obtained by applying a standardized thermal disturbance on the basis of steady-state operation under standard conditions, recording the evolution data of the thermal inertia state matrix and the air temperature response data in the storage room during the entire process of disturbance generation and dissipation, and scanning the thermal inertia response coefficient in the range of zero to one with a step size of 0.01. For each value, the theoretical thermal inertia offset sequence during the disturbance period is calculated, and the thermal inertia response coefficient that makes the cross-correlation coefficient between the theoretical thermal inertia offset and the actual temperature response reach the maximum value is selected as the thermal inertia response coefficient of the equipment; the unit of the defrost trigger threshold is the same as that of the thermal inertia offset, which is a dimensionless parameter.

[0088] The basic defrost trigger threshold is obtained based on the ultra-low temperature cold storage under standard operating conditions. This basic defrost trigger threshold is the minimum thermal inertia offset required for defrosting under standard operating conditions. The reference pressure difference of the ultra-low temperature cold storage in a completely frost-free state is obtained based on the cold storage's operating instructions. The safety inflection point value of the fan is obtained from the fan's user manual. This safety inflection point value is the maximum allowable pressure difference for the fan. Based on the basic defrost trigger threshold and the safety inflection point value, the defrost trigger threshold of the current ultra-low temperature cold storage at different times is calculated. The formula for calculating the defrost trigger threshold is:

[0089] ;

[0090] in, For ultra-low temperature cold storage at all times The defrost trigger threshold, Based on the basic defrost trigger threshold The attenuation coefficient is affected by the pressure difference. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine.

[0091] It should be explained that the basic defrost trigger threshold is the minimum thermal inertia offset required for defrosting in an ultra-low temperature cold storage under standard operating conditions. The unit is a dimensionless, non-negative real number. It is obtained by: allowing the ultra-low temperature cold storage to operate stably under standard operating conditions, turning off the defrost function and allowing frost to form naturally on the evaporator surface, while continuously recording the thermal inertia state matrix and calculating the thermal inertia offset at fixed sampling intervals. During the frosting process, the pressure difference between the evaporator inlet and outlet is monitored simultaneously. When the pressure difference reaches 70% of the fan's safety inflection point, the thermal inertia offset value is recorded. The arithmetic mean of the critical thermal inertia offsets recorded in three experiments is taken as the basic defrost trigger threshold. The reference pressure difference is the pressure difference between the evaporator inlet and outlet of the ultra-low temperature cold storage when the evaporator surface is completely free of frost and the fins are clean. The unit is Pascal. It is obtained by: confirming that the evaporator surface is completely frost-free during steady-state operation under standard operating conditions, closing the cold storage door, and continuously recording the pressure difference between the evaporator inlet and outlet. The baseline pressure difference is obtained by taking the arithmetic mean of all measured values. The safety inflection point value is obtained by referring to the fan performance curve manual provided by the fan manufacturer, and the unit is Pascal. The pressure difference influence attenuation coefficient is a dimensionless positive real number. It is obtained by taking the natural logarithm of the ratio of the optimal trigger threshold obtained for each pressure difference interval to the baseline defrost trigger threshold in the frosting acceleration experiment, and then dividing it by the normalized pressure difference ratio to obtain the estimated value of the attenuation coefficient corresponding to that pressure difference interval. The arithmetic mean of the estimated values ​​of the attenuation coefficient for all pressure difference intervals is taken to obtain the final value of the pressure difference influence attenuation coefficient. The optimal trigger threshold for different pressure difference values ​​is obtained as follows: 1.2 times, 1.5 times, 2.0 times, 2.5 times, and 2.8 times the baseline pressure difference. Under each pressure difference target level, the thermal balance offset that minimizes the weighted sum of the overshoot of the air temperature in the defrost chamber and the defrost duration is selected as the optimal trigger threshold for that pressure difference value.

[0092] Before the defrosting stage, the ultra-low temperature cold storage first undergoes a thermal inertia pre-compensation stage. The estimated defrosting time is calculated based on the evaporator inlet and outlet pressure difference, the reference pressure difference, and the safety inflection point value. The pre-compensation cooling capacity for the defrosting stage is then calculated based on the estimated defrosting time and the thermal inertia offset. This pre-compensation cooling capacity is the amount of cooling capacity compensation required during defrosting. The compressor operating frequency of the ultra-low temperature cold storage is calculated based on the pre-compensation cooling capacity, and the cold storage operates according to this compressor operating frequency during the thermal inertia pre-compensation stage. After the thermal inertia pre-compensation stage, the defrosting stage begins, and the current thermal inertia state matrix is ​​obtained as the thermal inertia prediction matrix. This thermal inertia prediction matrix is ​​used for the defrosting process. The predicted thermal inertia state matrix is ​​used to calculate the defrosting process thermal deviation based on the difference between the predicted thermal inertia matrix and the thermal inertia state matrix. The defrosting rate of change is then calculated based on this thermal deviation. A preset defrosting termination threshold is set. The defrosting stage ends and the cooling stage begins by comparing the defrosting rate of change with the defrosting termination threshold. The determination criteria are: if the defrosting rate of change is less than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage ends and the cooling stage begins; if the defrosting rate of change is greater than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage continues while a warning is issued. The formulas for calculating the defrosting duration, pre-compensated cooling capacity, compressor operating frequency, and defrosting rate of change are as follows:

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] in, For at any time The estimated defrosting time, Based on the defrosting time, This is the duration conversion factor. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine. For at any time Pre-compensation cooling capacity, This refers to the thermal inertia node number. For the first The current temperature of each thermal inertial node. For the first The target set temperature for each thermal inertial node. This is the offset compensation coefficient. For at any time Thermal inertia offset, For the compressor operating frequency, For the maximum permissible frequency, The current operating frequency, This is the conversion factor for cooling capacity. For the first The equivalent heat capacity of each thermal inertial node. For at any time defrosting change rate For at any time Thermal deviation in the defrosting process For a fixed sampling interval, For at any time The thermal inertial state matrix, To predict the thermal inertial state matrix, It is the square of the norm.

[0099] It should be explained that the thermal deviation of the defrosting process is dimensionless, and the unit of the defrosting change rate is per second; the duration of the pre-compensation stage is obtained as follows: the ultra-low temperature cold storage is stably operated under standard conditions, and then a standard defrosting process is simulated. The change curve of the cold storage wall temperature from the start of defrosting over time is recorded, and the time required for the wall temperature to rise from the initial value to the peak value is extracted as the duration of the pre-compensation stage; the thermal inertia prediction matrix is ​​obtained by calculating the thermal inertia state matrix at the end of the pre-compensation stage using the formula for calculating the real-time thermal inertia matching degree, and using the thermal inertia state matrix at the end of the pre-compensation stage as the thermal inertia prediction matrix; the estimated defrosting time is a pre-estimate of the time required for the defrosting process from the opening of the hot gas bypass to the complete melting of the frost layer. The unit is seconds. The baseline defrost time is the time required to complete one standard defrost cycle under baseline pressure difference conditions. It is obtained by performing a standard-thickness frost layer on the evaporator surface during calibration tests before the equipment leaves the factory. After this frost layer forms, hot gas bypass defrost is initiated, and the total time required from the start of defrost to the evaporator surface temperature rising above zero degrees Celsius and the pressure difference returning to the baseline pressure difference is recorded. This standard defrost test is repeated five times, and the arithmetic mean of the five recorded defrost times is taken as the baseline defrost time. The time conversion factor is a dimensionless positive real number. It is obtained by performing three defrost tests at different multiples of the baseline pressure difference at the evaporator inlet and outlet, recording the actual defrost time required under each condition, and calculating the ratio of the actual defrost time to the baseline defrost time. Subtract one and divide by the normalized differential pressure ratio of the group to obtain the estimated time conversion coefficient for that group. Take the arithmetic mean of all the estimated time conversion coefficients to obtain the final time conversion coefficient. The pre-compensation cooling capacity is the amount of cooling capacity compensation required during defrosting of the cold storage, in watts. The equivalent heat capacity is in joules per Kelvin and is obtained by: obtaining the mass of the object at the node using a weighing instrument, obtaining the specific heat capacity of the object from a table, multiplying the mass by the specific heat capacity, and adding the mass of ice multiplied by the specific heat capacity of ice to obtain the equivalent heat capacity of this node. The offset compensation coefficient is a positive real number in watts and is obtained by: stabilizing the ultra-low temperature cold storage and inducing frost; when the thermal balance offset reaches the defrosting trigger threshold, set the following parameters respectively. Different offset compensation coefficient values ​​are used, and a complete thermal inertia pre-compensation and defrosting process is performed after each value is selected. The overshoot of the air temperature in the air chamber after defrosting is recorded, and the offset compensation coefficient that minimizes the overshoot is selected. The compressor operating frequency is in Hertz, representing the speed command value that the variable frequency compressor should operate at during the thermal inertia pre-compensation stage. The maximum allowable frequency is determined by the nameplate parameters provided by the compressor manufacturer. The cooling capacity conversion coefficient is the conversion ratio between power and frequency. It is obtained by measuring the cooling capacity output at different frequencies on a compressor performance test bench, fitting the relationship between cooling capacity and frequency into a linear function curve, and calculating the slope of the linear function curve as the cooling capacity conversion coefficient, in Hertz per watt.The defrost termination threshold is obtained as follows: A complete defrost process is performed at the standard frost thickness, and the complete change curve of the thermal deviation during the defrost process is recorded. When the evaporator surface temperature rises above zero degrees Celsius and the pressure difference drops to the reference pressure difference, the frost layer is considered completely melted. The value of the thermal deviation during the defrost process at this point is recorded as the defrost termination threshold.

[0100] After the defrosting stage, the refrigeration stage begins. Control units are obtained by querying the control objects located at the thermal inertia nodes in the ultra-low temperature cold storage. A recovery priority index is calculated for each control unit; this index represents the priority of the control unit recovering from the defrosting state to the normal refrigeration state. The control units are then sorted in descending order based on their recovery priority indices to obtain a collaborative scheduling queue. Different control units are adjusted to return to the refrigeration stage based on the collaborative scheduling queue. The defrosting period is corrected by updating and iterating the thermal inertia weight matrix of the ultra-low temperature cold storage. The formula for calculating the recovery priority index is as follows:

[0101] ;

[0102] in, For the first Each control unit at time Recovery priority index, The serial number of the control unit is 1. For the first Each control unit at time Temperature deviation, This represents the maximum temperature deviation among all units. For the first Thermal inertia relaxation time constant of each control unit The maximum relaxation time constant among all elements. For the first Each control unit at time Thermal inertia offset, This represents the maximum thermal inertia offset among all elements. , and This is the priority weighting coefficient.

[0103] It should be explained that the control unit obtains the control unit corresponding to different thermal inertia nodes by querying a preset device configuration mapping table; the recovery priority index is the priority index of the control unit to recover from the defrost state to the normal cooling state, which is a dimensionless positive real number; the temperature deviation is the absolute value of the difference between the actual temperature of the thermal inertia node corresponding to the control unit at the current moment and its target set temperature, in Kelvin; the maximum temperature deviation among all units is the maximum value of the temperature deviations of the four control units, in Kelvin; the thermal inertia relaxation time constant is the time required for the output of the physical system controlled by the control unit to change from the initial value to the final steady-state value, a fixed quantity in seconds, obtained by applying a step start signal to the current thermal inertia node in the step response test before the equipment leaves the factory. The time required for the surface temperature of the current thermal inertia node to rise from its initial value to its steady-state value is recorded as the thermal inertia relaxation time constant, with the largest relaxation time constant among all units being selected. The priority weighting coefficient is a dimensionless parameter, obtained by first setting the initial value to 0.1, then sequentially adjusting the temperature deviation factor, thermal inertia relaxation time factor, and thermal balance offset factor, changing only one factor at a time and performing a complete defrost recovery experiment, recording the overshoot of the air temperature inside the cold storage, and selecting the value that minimizes the temperature adjustment amplitude. The thermal inertia offset of the control unit is the thermal inertia offset of the corresponding thermal inertia node of the control unit. For cases where the cold storage is overcooled, the minimum suitable temperature for different goods is obtained based on the goods in the cold storage as the lower limit of the current cold storage's refrigeration temperature. In cases where priorities are tied, the one with the larger thermal inertia relaxation time constant is adjusted first.

[0104] The thermal inertia weight matrix is ​​unfolded to obtain the thermal inertia weight vector. Sensitivity vectors of different nodes and dimensions at different snapshot sampling times are obtained based on the snapshot sampling times within the optimization cycle. The optimization cycle is the total time from the start of data acquisition to the resumption of refrigeration. The snapshot sampling time is the interval at which the thermal inertia weight matrix is ​​saved once. The sensitivity vector is the vector representing the change in the cold storage's sensitivity to heat with the number of iterations. The Kalman gain vector is calculated based on the sensitivity vectors at different snapshot sampling times, where the Kalman gain vector represents the change in weight coefficients in each iteration. The thermal inertia weight vector at the current snapshot sampling time is calculated based on the Kalman gain vector, the sensitivity vector, and the thermal inertia weight vector from the previous snapshot sampling time. The formula for calculating the thermal inertia weight matrix at the current snapshot sampling time is:

[0105] ;

[0106] ;

[0107] ;

[0108] ;

[0109] in, For the current optimization cycle, the first Thermal inertia weight vector at the next snapshot sampling time For the current optimization cycle, the first The average temperature of the air inside the storage room at the time of the next snapshot sampling. Set the target temperature for the air inside the storage facility. For the current optimization cycle, the first The first snapshot sampling time Node at the Dimensional thermal inertia weighting coefficient For the current optimization cycle, the first The covariance matrix of the thermal inertia weight matrix at the next snapshot sampling time. It is the identity matrix. For the current optimization cycle, the first Kalman gain vector at the next snapshot sampling time For the current optimization cycle, the first The sensitivity vector at the next snapshot sampling time. For transpose, For the current optimization cycle, the first The first snapshot sampling time Node at the Sensitivity to dimensions.

[0110] It needs to be explained that the covariance matrix of the thermal inertia weight matrix has the following characteristics: the initial covariance matrix is ​​the identity matrix, with dimensions equal to the square of the number of multi-source data types multiplied by the square of the number of thermal inertia nodes; the identity matrix has dimensions equal to the square of the number of multi-source data types multiplied by the square of the number of thermal inertia nodes; the Kalman gain vector has dimensions equal to the square of the number of thermal inertia nodes multiplied by one; the sensitivity vector has dimensions equal to the square of the number of thermal inertia nodes multiplied by one. Within the same defrost cycle, including the data acquisition to the final recovery cooling stage, the thermal inertia weight matrix remains unchanged. The thermal inertia weight matrix for the next snapshot sampling time in the current cycle is calculated using the thermal inertia weight matrix at the first snapshot sampling time in the previous cycle. The thermal inertia weight matrix is ​​updated at all snapshot sampling times within the current defrost cycle, and the last snapshot sampling time serves as the initial value for the first snapshot sampling time of the next defrost cycle. The initial value of the thermal inertia weight matrix in the first defrost cycle is one divided by the square of the number of thermal inertia nodes. The snapshot sampling time is an integer multiple of the fixed sampling time. The acquisition method is as follows: the number of multi-source data acquisitions in a complete defrost cycle is obtained based on the fixed number of sampling times in a complete defrost cycle. The snapshot sampling interval is an integer multiple of the fixed sampling interval. Each snapshot sampling interval is used as a snapshot sampling time to record multi-source data once. The thermal inertia weight matrix of the last snapshot sampling time in the current defrost cycle is updated and iteratively calculated based on the snapshot sampling time. Each iteration corresponds to a real snapshot sampling time, and the number of iterations is equal to the total number of snapshot sampling times in the optimization cycle. The updated weights are used to construct an updated weight matrix according to the nodes and dimensions. The real-time thermal inertia matching degree is calculated using the same nodes and dimensions in the next complete defrost cycle. The numerical stability of the covariance matrix is ​​handled by forcibly performing symmetry processing after each iteration update of the covariance matrix. The specific implementation method is as follows: after this iteration, the average of the covariance matrix and its transpose is forcibly used as the calculation result for the next iteration.

[0111] Example 2: Based on Example 1, the IoT-based defrosting and refrigeration coordinated control system for refrigeration equipment further includes:

[0112] The data acquisition module acquires multi-source data from the ultra-low temperature cold storage, constructs original observation vectors at different times based on the multi-source data, and constructs thermal inertia feature vectors at different times based on the original observation vectors.

[0113] The feature input module presets thermal inertial nodes, obtains the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes, calculates the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes, and constructs the thermal inertial state matrix.

[0114] The defrosting module calculates the thermal inertia offset and offset change rate based on the thermal inertia state matrix, presets a defrosting trigger threshold, and determines defrosting based on a comparison between the thermal inertia offset, offset change rate, and defrosting trigger threshold.

[0115] The refrigeration module is determined by calculating the pre-compensated refrigeration capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtaining the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determining the refrigeration by setting a defrosting termination threshold.

[0116] The refrigeration module is restored, the recovery priority index of each execution unit in the refrigeration stage is obtained, the recovery unit sequence is obtained based on the recovery priority index, and the refrigeration of the cold storage is restored based on the recovery unit sequence.

[0117] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the essence and scope of the technical solutions of the present invention.

Claims

1. A method for coordinated control of defrosting and refrigeration in refrigeration equipment based on the Internet of Things, characterized in that: Includes the following steps: S1: Obtain multi-source data from the ultra-low temperature cold storage, construct original observation vectors at different times based on the multi-source data, and construct thermal inertia feature vectors at different times based on the original observation vectors; S2: Preset thermal inertial nodes, obtain the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes; calculate the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes and construct the thermal inertial state matrix; S3: Calculate the thermal inertia offset and offset change rate based on the thermal inertia state matrix, preset the defrost trigger threshold, and determine defrost based on the comparison between the thermal inertia offset, offset change rate and defrost trigger threshold; S4: Calculate the pre-compensated cooling capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtain the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determine the cooling by setting the defrosting termination threshold. S5: Obtain the recovery priority index of each execution unit in the refrigeration stage, obtain the recovery unit sequence based on the recovery priority index, and restore the refrigeration of the cold storage based on the recovery unit sequence.

2. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 1, characterized in that, The process of acquiring multi-source data from the ultra-low temperature cold storage, constructing original observation vectors at different times based on the multi-source data, and constructing thermal inertia feature vectors at different times based on the original observation vectors includes: acquiring multi-source data and evaporator inlet and outlet pressure differences of the ultra-low temperature cold storage through an IoT sensor group at a fixed sampling interval; the multi-source data includes evaporator coil surface temperature, average air temperature inside the cold storage, average internal wall temperature of the cold storage, and compressor suction temperature; constructing the original observation vector of the ultra-low temperature cold storage at the current time based on the multi-source data; and obtaining the thermal inertia feature vector of the ultra-low temperature cold storage at the current time by differentiating the original observation vector at a fixed sampling interval. The thermal inertia feature vector is a vector representing the ability of the ultra-low temperature cold storage to represent the trend of thermal inertia changes. The thermal inertia feature vector includes the original observation vector at the current time, the first derivative of the original observation vector at the current time, and the second derivative of the original observation vector at the current time.

3. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 1, characterized in that, The preset thermal inertial nodes, based on thermal inertial feature vectors and thermal inertial nodes, obtain feature input quantities for the thermal inertial nodes, including: preset thermal inertial nodes based on multi-source data sources of the ultra-low temperature cold storage, wherein the thermal inertial nodes include evaporator nodes, internal air nodes, cold storage wall nodes, and compression equivalent nodes; preset heat accumulation integration windows are obtained through sampling time; and feature input quantities for different thermal inertial nodes are obtained by mapping based on thermal inertial feature vectors and heat accumulation integration windows. The feature input quantities are the input quantities of different thermal inertial nodes based on different defrosting and cooling forces, wherein the calculation formula for the feature input quantities is: ； ； ； ； ； ； ； ； in, For the first Nodes in the first dimension, in The feature input at time step This represents the sequence number of the sampling time. For a fixed sampling interval, For the first Node at The basic quantity at any given time. For the first Nodes in the second dimension, in The feature input at time step For the first Nodes in the third dimension, in The feature input at time step For the first Nodes in the fourth dimension, in The feature input at time step For the heat accumulation integral window, This refers to the time sequence number within the heat accumulation integration window. In order to be in The surface temperature of the evaporator coil at any given time. In order to be in The average temperature of the air inside the warehouse at any given time. In order to be in The average temperature of the inner wall of the tank at any given time. In order to be in The compressor intake temperature at any given time.

4. The IoT-based method for coordinated defrosting and refrigeration control of refrigeration equipment according to claim 1, characterized in that, The step of calculating the real-time thermal inertia matching degree and constructing the thermal inertia state matrix based on the characteristic input quantities of the thermal inertia nodes includes: obtaining the baseline center value and data standard deviation of the thermal inertia nodes in the current dimension under standard operating conditions for the ultra-low temperature cold storage, wherein the baseline center value is the standard quantity of the ultra-low temperature cold storage at the current position under standard operating conditions; calculating the real-time thermal inertia matching degree of the thermal inertia nodes at the current moment based on the baseline center value, data standard deviation, and characteristic input quantities of the thermal inertia nodes, wherein the real-time thermal inertia matching degree is the response intensity of the thermal inertia nodes to thermal inertia; and constructing the thermal inertia state matrix of the ultra-low temperature cold storage at the current moment based on the real-time thermal inertia matching degrees of different thermal inertia nodes, wherein the calculation formula for the real-time thermal inertia matching degree is: ； in, For the first Node at the Dimension, in Real-time thermal inertia matching degree at any given moment. For the first Node at the Dimensional thermal inertia weighting coefficient For the first Node at the Dimension, in The feature input at time step For the first Node at the The baseline center value of the dimension, For the first Node at the Standard deviation of data in a given dimension.

5. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 1, characterized in that, The process of calculating the thermal inertia offset and offset change rate based on the thermal inertia state matrix, setting a defrost trigger threshold, and determining defrost based on a comparison of the thermal inertia offset, offset change rate, and defrost trigger threshold includes: obtaining a reference thermal inertia state matrix under standard operating conditions for the ultra-low temperature cold storage; calculating the thermal inertia offset of the ultra-low temperature cold storage based on the thermal inertia state matrix and the reference thermal inertia state matrix, wherein the thermal inertia offset is the offset of thermal energy relative to standard operating conditions during cold storage refrigeration; calculating the offset change rate based on the thermal inertia offset, wherein the offset change rate is the offset change rate of thermal energy under standard operating conditions; setting a defrost trigger threshold at the current moment based on the evaporator inlet and outlet pressure difference; and determining the current defrost stage of the ultra-low temperature cold storage by comparing the thermal inertia offset, offset change rate, and defrost trigger threshold. The determination condition is: if the thermal inertia offset is greater than or equal to the defrost trigger threshold at the current moment and the offset change rate is greater than zero, then the current ultra-low temperature cold storage enters the defrost stage; wherein the calculation formulas for the thermal inertia offset and offset change rate are: ； ； in, For at any time Thermal inertia offset, For at any time The thermal inertial state matrix, This is the reference thermal inertia state matrix under standard operating conditions. Let be the square of the Euclidean norm. The thermal inertial response coefficient, For a fixed sampling interval, For at any time The rate of change of offset.

6. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 5, characterized in that, The step of presetting the defrost trigger threshold based on the pressure difference between the evaporator inlet and outlet includes: obtaining a basic defrost trigger threshold based on the ultra-low temperature cold storage under standard operating conditions, wherein the basic defrost trigger threshold is the minimum thermal inertia offset for defrosting the ultra-low temperature cold storage under standard operating conditions; obtaining the reference pressure difference of the ultra-low temperature cold storage in a completely frost-free state based on the ultra-low temperature cold storage; obtaining the safety inflection point value of the fan through the fan operation manual of the ultra-low temperature cold storage, wherein the safety inflection point value is the maximum allowable pressure difference of the fan; and calculating the defrost trigger threshold of the current ultra-low temperature cold storage at different times based on the basic defrost trigger threshold and the safety inflection point value, wherein the calculation formula for the defrost trigger threshold is: ； in, For ultra-low temperature cold storage at all times The defrost trigger threshold, Based on the basic defrost trigger threshold The attenuation coefficient is affected by the pressure difference. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine.

7. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 1, characterized in that, The process of calculating the pre-compensated cooling capacity and compressor operating frequency for the defrosting stage based on thermal inertia offset, obtaining the thermal deviation of the defrosting process based on the current thermal inertia state matrix, and determining refrigeration based on a preset defrosting termination threshold includes: before the ultra-low temperature cold storage enters the defrosting stage, a thermal inertia pre-compensation stage is first performed; the estimated defrosting duration is calculated based on the evaporator inlet and outlet pressure difference, the reference pressure difference, and the safety inflection point value; the pre-compensated cooling capacity for the defrosting stage is calculated based on the estimated defrosting duration and thermal inertia offset, wherein the pre-compensated cooling capacity is the amount of cooling capacity compensation required during cold storage defrosting; the compressor operating frequency of the ultra-low temperature cold storage is calculated based on the pre-compensated cooling capacity; and the cold storage operates according to the compressor operating frequency during the thermal inertia pre-compensation stage; after the thermal inertia pre-compensation stage ends, the defrosting stage begins. The current thermal inertia state matrix is ​​obtained as the thermal inertia prediction matrix, which is the predicted thermal inertia state matrix after defrosting. The thermal deviation of the defrosting process is calculated based on the difference between the thermal inertia prediction matrix and the thermal inertia state matrix. The defrosting change rate is calculated based on the thermal deviation. A preset defrosting termination threshold is set. The defrosting stage ends and the cooling stage begins by comparing the defrosting change rate with the defrosting termination threshold. The determination conditions are: if the defrosting change rate is less than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage ends and the cooling stage begins; if the defrosting change rate is greater than the defrosting termination threshold for three consecutive fixed sampling intervals, the defrosting stage continues while a warning is issued. The calculation formulas for defrosting duration, pre-compensated cooling capacity, compressor operating frequency, and defrosting change rate are as follows: ； ； ； ； ； in, For at any time The estimated defrosting time, Based on the defrosting time, This is the duration conversion factor. For ultra-low temperature cold storage at all times The pressure difference between the inlet and outlet of the evaporator, This is the baseline pressure difference for an ultra-low temperature cold storage facility under completely frost-free conditions. This is the safety inflection point value for the wind turbine. For at any time Pre-compensation cooling capacity, This refers to the thermal inertia node number. For the first The current temperature of each thermal inertial node. For the first The target set temperature for each thermal inertial node. This is the offset compensation coefficient. For at any time Thermal inertia offset, For the compressor operating frequency, For the maximum permissible frequency, The current operating frequency, This is the conversion factor for cooling capacity. For the first The equivalent heat capacity of each thermal inertial node. For at any time defrosting change rate For at any time Thermal deviation in the defrosting process For a fixed sampling interval, For at any time The thermal inertial state matrix, To predict the thermal inertial state matrix, It is the square of the norm.

8. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 1, characterized in that, The process of obtaining the recovery priority index of each execution unit during the cooling phase, obtaining the recovery unit sequence based on the recovery priority index, and restoring cold storage cooling based on the recovery unit sequence includes: initiating the cooling phase after the defrosting phase; obtaining control units by querying the control objects where the thermal inertia nodes in the ultra-low temperature cold storage are located; calculating the recovery priority index of each control unit, where the recovery priority index is the priority index for the control unit to recover from the defrosting state to the normal cooling state; sorting the control units in descending order based on the recovery priority index to obtain a collaborative scheduling queue; adjusting the recovery of different control units to the cooling phase based on the collaborative scheduling queue; and correcting the defrosting period by updating and iterating the thermal inertia weight matrix of the ultra-low temperature cold storage. The formula for calculating the recovery priority index is: ； in, For the first Each control unit at time Recovery priority index, The serial number of the control unit is 1. For the first Each control unit at time Temperature deviation, This represents the maximum temperature deviation among all units. For the first Thermal inertia relaxation time constant of each control unit The maximum relaxation time constant among all elements. For the first Each control unit at time Thermal inertia offset, This represents the maximum thermal inertia offset among all elements. , and This is the priority weighting coefficient.

9. The IoT-based method for coordinated control of defrosting and refrigeration in refrigeration equipment according to claim 8, characterized in that, The method of correcting the defrosting period by updating and iterating the thermal inertia weight matrix of the ultra-low temperature cold storage includes: flattening the thermal inertia weight matrix to obtain the thermal inertia weight vector; obtaining sensitivity vectors of different nodes and dimensions at different snapshot sampling times based on the snapshot sampling times within the optimization period; the optimization period being the total time from the start of data acquisition to the resumption of refrigeration; the snapshot sampling time being the interval for saving the thermal inertia weight matrix once; the sensitivity vector being the vector representing the change in the cold storage's sensitivity to heat with the number of iterations; calculating the Kalman gain vector at different snapshot sampling times based on the sensitivity vectors at different snapshot sampling times; the Kalman gain vector being the change in weight coefficients in each iteration; and calculating the thermal inertia weight vector at the current snapshot sampling time based on the Kalman gain vector, the sensitivity vector, and the thermal inertia weight vector at the previous snapshot sampling time. The formula for calculating the thermal inertia weight matrix at the current snapshot sampling time is as follows: ； ； ； ； in, For the current optimization cycle, the first Thermal inertia weight vector at the next snapshot sampling time For the current optimization cycle, the first The average temperature of the air inside the storage room at the time of the next snapshot sampling. Set the target temperature for the air inside the storage facility. For the current optimization cycle, the first The first snapshot sampling time Node at the Dimensional thermal inertia weighting coefficient For the current optimization cycle, the first The covariance matrix of the thermal inertia weight matrix at the next snapshot sampling time. It is the identity matrix. For the current optimization cycle, the first Kalman gain vector at the next snapshot sampling time For the current optimization cycle, the first The sensitivity vector at the next snapshot sampling time. For transpose, For the current optimization cycle, the first The first snapshot sampling time Node at the Sensitivity to dimensions.

10. An IoT-based defrosting and refrigeration coordinated control system for refrigeration equipment, used to implement the IoT-based defrosting and refrigeration coordinated control method for refrigeration equipment as described in any one of claims 1-9, characterized in that it further includes... include: The data acquisition module acquires multi-source data from the ultra-low temperature cold storage, constructs original observation vectors at different times based on the multi-source data, and constructs thermal inertia feature vectors at different times based on the original observation vectors. The feature input module presets thermal inertial nodes, obtains the feature input of thermal inertial nodes based on thermal inertial feature vectors and thermal inertial nodes, calculates the real-time thermal inertial matching degree based on the feature input of thermal inertial nodes, and constructs the thermal inertial state matrix. The defrosting module calculates the thermal inertia offset and offset change rate based on the thermal inertia state matrix, presets a defrosting trigger threshold, and determines defrosting based on a comparison between the thermal inertia offset, offset change rate, and defrosting trigger threshold. The refrigeration module is determined by calculating the pre-compensated refrigeration capacity and compressor operating frequency during the defrosting stage based on the thermal inertia offset, obtaining the thermal deviation of the defrosting process based on the thermal inertia state matrix at the current moment, and determining the refrigeration by setting a defrosting termination threshold. The refrigeration module is restored, the recovery priority index of each execution unit in the refrigeration stage is obtained, the recovery unit sequence is obtained based on the recovery priority index, and the refrigeration of the cold storage is restored based on the recovery unit sequence.

Images