Intelligent temperature control method and system for contactless cold chain transportation based on temperature prediction

Abstract

This invention belongs to the field of cold chain transportation temperature control technology, specifically involving a contactless intelligent temperature control method and system for cold chain transportation based on temperature prediction. The method includes: collecting the surface temperature of goods using temperature sensors and calculating the rate of change of the surface temperature of goods by combining historical temperature data; collecting the evaporator fan speed and airflow guide angle of the refrigeration unit; constructing a nonlinear response surface by fitting the evaporator fan speed, airflow guide angle, and the rate of change of the surface temperature of goods using nonlinear regression; substituting the preset parameter combination of evaporator fan speed and airflow guide angle into the nonlinear response surface to obtain an estimated value of the rate of change of the surface temperature of goods, and predicting the predicted temperature value of the surface temperature of goods; selecting parameter combinations whose predicted temperature values ​​do not exceed a preset upper temperature threshold to form a set of candidate combinations for compliance, ranking them to approximate the ideal solution to obtain the target operating parameter combination of the refrigeration unit, and then performing temperature regulation. This invention achieves precise temperature control in the cold chain transportation process.

Description

Technical Field

[0001] This invention belongs to the field of cold chain transportation temperature control technology, specifically relating to a contactless cold chain transportation intelligent temperature control method and system based on temperature prediction. Background Technology

[0002] Cold chain transportation is a core means of ensuring the quality and safety of food, pharmaceuticals, and fresh produce. A fundamental requirement of cold chain transportation is maintaining the temperature inside the vehicle compartment within a specified range throughout the entire transport process. As a key device for temperature control within the compartment, the operating parameters of the refrigeration unit directly determine the actual cooling effect inside the compartment.

[0003] However, in actual transportation scenarios, the relationship between refrigeration unit operating parameters and cargo temperature is not a simple, stable linear one, but rather exhibits significant nonlinear characteristics. This nonlinearity mainly stems from two aspects: First, the heat transfer process of airflow within the carriage is affected by multiple factors, including cargo type, stacking method, loading density, and the internal structure of the carriage, resulting in significant differences in cooling response under different cargo loading conditions for the same set of operating parameters. Second, during long-distance high-temperature transportation, the outer walls of the carriage continuously bear external heat loads, causing the temperature field inside the carriage to dynamically change with travel time and external environmental conditions. This results in the mapping relationship between operating parameters and cooling effect continuously drifting over time and under operating conditions. Existing temperature control methods struggle to effectively detect these dynamically changing nonlinear responses, making it impossible to predict and intervene in the changing trend of cargo temperature within the carriage in a timely and accurate manner. This can easily lead to problems such as localized overheating or excessive temperature fluctuations, posing a potential threat to cargo quality and safety. Summary of the Invention

[0004] (1) Technical problems to be solved The purpose of this invention is to provide a contactless intelligent temperature control method and system for cold chain transportation based on temperature prediction, in order to solve the problem that existing temperature control methods cannot effectively sense the nonlinear response relationship between the operating parameters of the refrigeration unit and the effective cooling of the carriage in high-temperature long-distance cold chain transportation, which leads to the inability to effectively control the temperature of the cold chain transportation process.

[0005] (2) Technical solution To achieve the above objectives, in one aspect, the present invention provides an intelligent temperature control method for contactless cold chain transportation based on temperature prediction, the method comprising: The surface temperature of the cargo is collected using a non-contact infrared temperature sensor; the rate of change of the cargo surface temperature is calculated by differential calculation based on the cargo surface temperature and the historical temperature data of the corresponding cargo location; the evaporator fan speed and airflow guide angle of the refrigeration unit are also collected.

[0006] The evaporator fan speed, airflow guide angle, and cargo surface temperature change rate were fitted using nonlinear regression to construct a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle.

[0007] Substituting the parameter combination of preset evaporator fan speed and airflow guide angle into the nonlinear response surface, an estimated value of the rate of change of cargo surface temperature is obtained; the estimated value of the rate of change of cargo surface temperature is then used to predict the temperature of the cargo surface using a linear extrapolation method.

[0008] A set of parameter combinations is obtained by selecting parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold; the compliant candidate combinations are then sorted using an algorithm that approximates the ideal solution to obtain the target operating parameter combination for the refrigeration unit; and temperature regulation is performed based on the target operating parameter combination for the refrigeration unit.

[0009] Furthermore, the method for constructing a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle through nonlinear regression fitting of the evaporator fan speed, airflow guide angle, and cargo surface temperature change rate includes: Within a preset data acquisition time window, the evaporator fan speed sequence, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence are obtained based on the evaporator fan speed, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence, respectively. Cross-correlation functions are calculated between the evaporator fan speed sequence, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence, respectively. The time delay corresponding to the maximum absolute value of each cross-correlation function is selected, and the evaporator fan speed sequence and the air delivery guide angle sequence are shifted and aligned to obtain time-aligned evaporator fan speed sequence and air delivery guide angle sequence.

[0010] Using the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable, a second-order polynomial regression was performed to calculate the Cook distance corresponding to each acquisition period. The Cook distance was then used to denoise the time-aligned evaporator fan speed sequence and airflow guide angle sequence to obtain the denoised evaporator fan speed sequence and airflow guide angle sequence.

[0011] The denoised evaporator fan speed sequence and airflow guide angle sequence are fitted with a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm.

[0012] Furthermore, the method for calculating the cross-correlation function between the evaporator fan speed sequence and the airflow guide angle sequence and the cargo surface temperature change rate sequence includes: First-order differences were performed on the evaporator fan speed sequence and the air delivery guide angle sequence to obtain the evaporator fan speed difference sequence and the air delivery guide angle difference sequence, respectively; linear detrending was performed on the cargo surface temperature change rate sequence to obtain the detrended temperature change rate sequence.

[0013] Calculate the cross-correlation function between the evaporator fan speed difference sequence and the detrended temperature change rate sequence; calculate the cross-correlation function between the air supply guide angle difference sequence and the detrended temperature change rate sequence.

[0014] Furthermore, the method for calculating the Cook distance corresponding to each acquisition cycle by performing a second-order polynomial regression with the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable includes: The parameter space corresponding to the aligned evaporator fan speed sequence and airflow guide angle sequence is divided into uniform grids according to the respective value ranges of the evaporator fan speed and airflow guide angle. The average value of the cargo surface temperature change rate corresponding to the acquisition period falling into the same uniform grid is taken, and the average value of the evaporator fan speed and airflow guide angle at the center of the grid and the corresponding cargo surface temperature change rate constitutes the balanced acquisition sample sequence.

[0015] Using the evaporator fan speed and airflow guide angle in the balanced sample sequence as independent variables and the mean rate of change of the corresponding cargo surface temperature as the dependent variable, a second-order polynomial regression is performed to calculate the Cook distance of the balanced sample sequence.

[0016] Furthermore, the method for fitting the nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm after denoising the evaporator fan speed sequence and airflow guide angle sequence includes: Using the denoised evaporator fan speed sequence and airflow guide angle sequence as independent variables and the corresponding cargo surface temperature change rate as dependent variable, a second-order complete polynomial containing the first and second terms of evaporator fan speed, the first and second terms of airflow guide angle, and their cross-product term are fitted by the least squares method to obtain the preliminary nonlinear response surface and the residual sequence of each acquisition period.

[0017] Calculate the arithmetic mean of the denoised evaporator fan speed sequence, and divide the denoised evaporator fan speed sequence into high speed segment and low speed segment using the arithmetic mean as the boundary; calculate the mean square residual of the residual sequence corresponding to the high speed segment and the low speed segment respectively, and determine the speed segment with larger mean square residual as the speed segment with significant nonlinear response.

[0018] Using the absolute value of the difference between the evaporator fan speed and the mean evaporator fan speed in each acquisition cycle within the significant speed range of the nonlinear response as the independent variable, the weight of each acquisition cycle is calculated using a Gaussian function. Weighted least squares regression is then performed on each acquisition cycle within the significant speed range of the nonlinear response using these weights to obtain a residual correction polynomial containing the first and second terms of the evaporator fan speed, the first and second terms of the airflow guide angle, and their cross-product. This residual correction polynomial is then added to the second-order complete polynomial of the initial nonlinear response surface within the corresponding speed interval of the significant speed range of the nonlinear response to obtain the correction polynomial for the significant speed range of the nonlinear response. Finally, the correction polynomial for the significant speed range of the nonlinear response is merged with the second-order complete polynomial of the initial nonlinear response surface outside the corresponding interval of the significant speed range of the nonlinear response to obtain the nonlinear response surface.

[0019] Furthermore, the method for setting the parameter combination of the preset evaporator fan speed and the airflow guide angle includes: Obtain the minimum stable operating speed and the rated maximum speed of the inverter driver of the refrigeration unit; perform arithmetic progression discretization on the speed range between the minimum stable operating speed and the rated maximum speed to obtain the evaporator fan speed discrete sequence; obtain the minimum deflection angle and the maximum mechanical deflection angle of the air supply guide mechanism, and perform arithmetic progression discretization on the angle range between the minimum deflection angle and the maximum mechanical deflection angle to obtain the air supply guide angle discrete sequence.

[0020] The gas flow pressure is obtained by applying the dynamic pressure formula to the evaporator fan speed.

[0021] From all combinations of the discrete sequences of evaporator fan speed and airflow guide angle, retain the combinations where the airflow pressure is not lower than the preset airflow pressure threshold to obtain the preset parameter combination of evaporator fan speed and airflow guide angle.

[0022] Furthermore, the method for predicting the temperature forecast value of the cargo surface temperature by using the estimated rate of change of cargo surface temperature to predict the cargo surface temperature through linear extrapolation includes: Obtain the historical cargo surface temperature change rate sequence within a preset number of historical collection periods, and obtain the temperature change rate trend slope through Theil-Sen slope estimation.

[0023] The length of the prediction time window is determined by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the trend of the rate of change of temperature.

[0024] The extrapolation termination rate is obtained by summing the product of the slope of the temperature change rate trend and the length of the prediction time window with the estimated value of the temperature change rate of the cargo surface. The temperature change within the prediction time window is integrated using the estimated value of the temperature change rate of the cargo surface and the extrapolation termination rate as the two endpoints to obtain the temperature change of the cargo surface.

[0025] The current surface temperature of the cargo is added to the change in surface temperature of the cargo to obtain the predicted surface temperature of the cargo.

[0026] Furthermore, the method for determining the prediction time window length by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the trend of the rate of change of temperature includes: The basic time window is obtained by dividing the absolute value of the estimated rate of change of cargo surface temperature by the absolute value of the slope of the trend of the rate of change of temperature; the sign of the thermodynamic state discrimination sign is obtained by the product of the estimated rate of change of cargo surface temperature and the slope of the trend of the rate of change of temperature.

[0027] When the thermodynamic state discrimination sign is negative, the base time window is used as the prediction time window length.

[0028] When the thermodynamic state discrimination sign is positive, the basic time window is multiplied by a preset time coefficient to obtain the prediction time window length.

[0029] Furthermore, the method for obtaining the target operating parameter combination of the chiller unit by sorting the candidate combinations of compliance using an algorithm that approximates the ideal solution includes: From the set of candidate combinations that meet the criteria, the positive ideal solution is formed by the minimum evaporator fan speed and the maximum predicted temperature; the negative ideal solution is formed by the maximum evaporator fan speed and the minimum predicted temperature.

[0030] Based on the range of evaporator fan speed and temperature prediction values ​​in the candidate combination set, the range of evaporator fan speed and temperature prediction values ​​for each parameter combination is normalized, and the Euclidean distance between the normalized parameter combination and the positive ideal solution and the negative ideal solution is calculated.

[0031] The relative closeness is obtained by dividing the Euclidean distance between each parameter combination and the negative ideal solution by the sum of the Euclidean distances between the positive and negative ideal solutions; the parameter combination with the largest relative closeness is determined as the target operating parameter combination of the refrigeration unit.

[0032] Based on the same inventive concept, this invention also provides a contactless cold chain transportation intelligent temperature control system based on temperature prediction, the system comprising: The data acquisition module is used to collect the surface temperature value of the cargo through a non-contact infrared temperature sensor; based on the surface temperature value of the cargo and the historical temperature data of the corresponding cargo location, the rate of change of the surface temperature of the cargo is calculated by differential calculation; and the evaporator fan speed and airflow guide angle of the refrigeration unit are collected.

[0033] The response surface modeling module is used to construct a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle by nonlinear regression fitting.

[0034] The temperature trend prediction module is used to substitute the parameter combination of preset evaporator fan speed and air delivery guide angle into the nonlinear response surface to obtain the estimated value of the cargo surface temperature change rate; and to predict the temperature prediction value of the cargo surface temperature by using the estimated value of the cargo surface temperature change rate through linear extrapolation.

[0035] The temperature control module is used to select parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold, thereby obtaining a set of candidate combinations that meet the standards; the candidate combinations that meet the standards are sorted by an algorithm that approximates the ideal solution to obtain the target operating parameter combination of the refrigeration unit; and the temperature is controlled according to the target operating parameter combination of the refrigeration unit.

[0036] (3) Beneficial effects Compared with the prior art, the beneficial effects of the present invention are: 1. By collecting the surface temperature of goods through a non-contact infrared temperature sensor, and combining the evaporator fan speed and airflow angle, a nonlinear response surface is constructed through nonlinear regression fitting. The predicted temperature value of the goods surface temperature is then predicted by linear extrapolation. Subsequently, a set of candidate combinations that meet the standards is screened, and the target operating parameter combination of the refrigeration unit is determined by an algorithm that approximates the ideal solution. This achieves effective perception and precise temperature control of the nonlinear response relationship between the operating parameters of the refrigeration unit and the effective cooling during cold chain transportation.

[0037] 2. When constructing the nonlinear response surface, the evaporator fan speed sequence and the air supply guide angle sequence are shifted and aligned by cross-correlation function, and noise is removed according to Cook distance. Then, the significant speed segment of the nonlinear response is locally refined by combining Gaussian function weighted piecewise residual correction, so as to obtain a high-precision nonlinear response surface, which improves the accuracy of subsequent temperature prediction and the reliability of the selection of refrigeration unit operating parameters. Attached Figure Description

[0038] Figure 1 This is a flowchart of the intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to Embodiment 1 of the present invention. Figure 2This is a schematic diagram of the module composition of the contactless cold chain transportation intelligent temperature control system based on temperature prediction according to Embodiment 2 of the present invention. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0040] Before providing examples, it is necessary to describe the application scenarios of this invention. This invention is applicable to the long-distance delivery of mobile cold chain transportation equipment such as refrigerated trucks and refrigerated containers, enabling contactless temperature control of temperature-sensitive goods that are unsuitable for wired probes. In high-temperature long-distance cold chain transportation scenarios, the aforementioned nonlinear response characteristics are particularly prominent: during long-distance transportation, the truck compartment is sealed for extended periods, and external high temperatures continuously transfer heat inward through the outer walls of the compartment. The refrigeration unit must maintain high power output to counteract the continuous heat load, and the evaporator fan operates under high load for extended periods. The heat exchange process between the airflow and the goods is subject to multiple coupling interferences from the cargo density, differences in packaging insulation, and the internal structure of the truck compartment. The cooling response caused by the same set of refrigeration unit operating parameters varies significantly depending on the location of the goods and the loading conditions.

[0041] Example 1: As Figure 1 As shown, this embodiment provides a contactless cold chain transportation intelligent temperature control method based on temperature prediction, the method including: S1. Collect the surface temperature value of the cargo using a non-contact infrared temperature sensor; calculate the rate of change of the cargo surface temperature by differential calculation based on the cargo surface temperature value and the historical temperature data of the corresponding cargo location; collect the evaporator fan speed and airflow guide angle of the refrigeration unit.

[0042] For example, non-contact infrared temperature sensors are installed on the top of the compartment, covering each stacked cargo area from an angled downward view. Taking a standard refrigerated compartment carrying nine pallets of goods as an example, six non-contact infrared temperature sensors are evenly distributed along the longitudinal direction of the compartment, covering six cargo locations respectively. Each sensor independently collects the surface temperature value of the covered cargo location. In cold chain transportation, goods are diverse and have various packaging forms. Non-contact infrared temperature sensors do not need to be in direct contact with the goods. They are installed on the compartment structural components and do not move with the loading and unloading of goods, allowing for continuous monitoring of fixed cargo locations.

[0043] Temperature data is collected every 10 seconds at each cargo location. The collected cargo surface temperature value is then compared with the historical temperature value from the previous collection cycle at the corresponding cargo location using a first-order difference method. The difference result is divided by the duration of the collection cycle to obtain the rate of change of the cargo surface temperature. The difference calculation uses the temperature values ​​from two adjacent collection cycles at the same cargo location, thus eliminating the interference of uneven temperature distribution between different cargo locations on the calculation of the rate of change.

[0044] The evaporator fan speed is calculated by reading the real-time output frequency of the refrigeration unit's inverter driver, and the air delivery guide angle is obtained by reading the real-time feedback from the air delivery guide mechanism's position sensor. Both are synchronized with the cargo surface temperature acquisition cycle.

[0045] S2. The evaporator fan speed, airflow guide angle and the rate of change of cargo surface temperature are fitted by nonlinear regression to construct a nonlinear response surface of the rate of change of cargo surface temperature with respect to the evaporator fan speed and airflow guide angle.

[0046] S3. Substitute the preset combination of evaporator fan speed and airflow guide angle into the nonlinear response surface to obtain the estimated value of the cargo surface temperature change rate; use the estimated value of the cargo surface temperature change rate to predict the temperature prediction value of the cargo surface temperature through linear extrapolation.

[0047] S4. Select parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold to obtain a set of candidate combinations for compliance; sort the candidate combinations for compliance using an algorithm that approximates the ideal solution to obtain the target operating parameter combination of the refrigeration unit; and regulate the temperature according to the target operating parameter combination of the refrigeration unit.

[0048] For example, the preset upper temperature threshold is determined based on the refrigeration requirements of the transported goods. For instance, the typical requirement for cold chain transportation of fresh food is that the surface temperature of the goods does not exceed 8°C, so the preset upper temperature threshold is set to 8°C. The predicted temperature values ​​corresponding to each parameter combination are compared with 8°C one by one, and only the parameter combinations in which the predicted temperature values ​​of all goods positions in the compartment do not exceed 8°C are retained, forming a set of candidate combinations for compliance.

[0049] The candidate combination set that meets the standard is sorted to approximate the ideal solution, and the target operating parameter combination of the chiller unit is determined to be evaporator fan speed of 1050 rpm and air delivery guide angle of 30°.

[0050] Accordingly, a speed command is sent to the inverter driver of the refrigeration unit to adjust the evaporator fan speed to 1050 rpm; a position command is sent to the drive actuator of the air supply guide mechanism to adjust the air supply guide angle to 30°, thus completing the temperature regulation of the current control cycle.

[0051] The method for constructing a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle through nonlinear regression fitting includes: Within a preset data acquisition time window, the evaporator fan speed sequence, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence are obtained based on the evaporator fan speed, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence, respectively. Cross-correlation functions are calculated between the evaporator fan speed sequence, the air delivery guide angle sequence, and the cargo surface temperature change rate sequence, respectively. The time delay corresponding to the maximum absolute value of each cross-correlation function is selected, and the evaporator fan speed sequence and the air delivery guide angle sequence are shifted and aligned to obtain time-aligned evaporator fan speed sequence and air delivery guide angle sequence.

[0052] Using the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable, a second-order polynomial regression was performed to calculate the Cook distance corresponding to each acquisition period. The Cook distance was then used to denoise the time-aligned evaporator fan speed sequence and airflow guide angle sequence to obtain the denoised evaporator fan speed sequence and airflow guide angle sequence.

[0053] The denoised evaporator fan speed sequence and airflow guide angle sequence are fitted with a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm.

[0054] For example, the preset collection time window is selected based on the principle of being able to cover multiple adjustments of evaporator fan speed and airflow angle while maintaining the overall thermal condition of the compartment relatively stable. It is set to 20 minutes, with a collection cycle of 10 seconds. The preset collection time window contains data from 120 collection cycles, thereby obtaining a sequence of evaporator fan speed, airflow angle, and cargo surface temperature change rate with a length of 120.

[0055] After the evaporator fan speed is adjusted, the cold air flows through the air supply channel to the cargo surface, creating a clear physical heat transfer path. This results in a time delay in the response of the cargo surface temperature change rate to the fan speed change. Similarly, after the air supply angle is adjusted, the airflow direction redistributes and establishes a new heat transfer distribution on the cargo surface, which also requires time. Directly pairing the evaporator fan speed, air supply angle, and cargo surface temperature change rate at the same data collection moment for regression is equivalent to forcibly associating the current actuator state with the lagging temperature change rate, introducing systematic bias into the regression results. To address this, a cross-correlation function is calculated between the evaporator fan speed sequence and the cargo surface temperature change rate sequence. The time delay corresponding to the maximum absolute value of the cross-correlation function is used to shift and align the evaporator fan speed sequence. Similarly, a cross-correlation function is calculated independently between the air supply angle sequence and the cargo surface temperature change rate sequence, and the corresponding time delay is used to shift and align the air supply angle sequence.

[0056] Taking an evaporator fan speed sequence delay of 3 acquisition cycles (30 seconds) and an airflow guide angle sequence delay of 2 acquisition cycles (20 seconds) as an example, the first... The rate of change of cargo surface temperature in the first collection cycle and the first Evaporator fan speed in the first collection cycle, the first The airflow guide angle is matched for each collection cycle. By taking values ​​from 4 to 120, we obtain a time-aligned evaporator fan speed sequence and airflow guide angle sequence with a length of 117.

[0057] A second-order polynomial regression was performed using aligned evaporator fan speed and airflow guide angle sequences as independent variables and cargo surface temperature change rate sequences as dependent variables to calculate the Cook distance for each acquisition period. The Cook distance is a statistic in linear regression diagnostics that measures the influence of a single sample point on the overall regression fit. It quantifies the influence by comparing the change in the regression coefficient vector before and after deleting the point. Its value is determined by both the leverage effect (i.e., the degree to which the independent variable deviates from the mean) and the magnitude of the residual. A larger Cook distance indicates a more significant impact of that acquisition period on the overall regression result, and is more likely to be a high-influence point to be identified. Larger Cook distances typically correspond to acquisition periods where the evaporator fan speed or airflow guide angle is in a transitional adjustment process. At such times, the refrigeration unit's operating parameters are still changing, and the cargo surface temperature change rate reflects the lag response of the previous operating state and the overlap of the current adjustment action, which deviates significantly from the regression pattern under the denoised operating conditions. Therefore, it is considered a high-influence point in regression diagnostics. A Cook distance threshold was used. As a criterion, among which, The empirical threshold rule in the Cook distance removes the acquisition period with a Cook distance exceeding 0.034 from the time-aligned evaporator fan speed sequence and airflow guide angle sequence to complete the noise reduction, resulting in the noise-reduced evaporator fan speed sequence and airflow guide angle sequence.

[0058] The denoised evaporator fan speed sequence and airflow guide angle sequence were fitted with the nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using the response surface analysis algorithm. The specific fitting method is described later.

[0059] The method for calculating the cross-correlation function between the evaporator fan speed sequence, the airflow guide angle sequence, and the cargo surface temperature change rate sequence includes: First-order differences were performed on the evaporator fan speed sequence and the air delivery guide angle sequence to obtain the evaporator fan speed difference sequence and the air delivery guide angle difference sequence, respectively; linear detrending was performed on the cargo surface temperature change rate sequence to obtain the detrended temperature change rate sequence.

[0060] Calculate the cross-correlation function between the evaporator fan speed difference sequence and the detrended temperature change rate sequence; calculate the cross-correlation function between the air supply guide angle difference sequence and the detrended temperature change rate sequence.

[0061] For example, both the evaporator fan speed sequence and the airflow guide angle sequence are 120° long. The evaporator fan speed remains constant between two adjustment commands, and the sequence exhibits a stepped distribution. For instance, the fan speed is maintained at 900 rpm in the first 40 sampling cycles, adjusted to 1200 rpm in the 41st to 80th sampling cycles, and maintained at 1050 rpm in the 81st to 120th sampling cycles. After performing a first-order difference on the evaporator fan speed sequence, a pulse with an amplitude of +300 rpm appears at the boundary of the 40th to 41st sampling cycle, and a pulse with an amplitude of [missing value] appears at the boundary of the 80th to 81st sampling cycle. A pulse of 150 rpm was used, with the difference value of 0 for the remaining acquisition cycles, to obtain an evaporator fan speed difference sequence of length 119. The air supply guide angle sequence was processed in the same way to obtain an air supply guide angle difference sequence of length 119.

[0062] The rate of change of cargo surface temperature varies slowly and monotonically within a 20-minute window as the thermal equilibrium process of the cargo compartment progresses. For example, the rate of change of temperature within the window ranges from... Gradually changing from 0.06℃ / s to The temperature change rate is 0.03℃ / s, exhibiting a linear drift superimposed with random disturbances. A linear trend is fitted to the cargo surface temperature change rate sequence using the least squares method. Subtracting the fitted straight line from the cargo surface temperature change rate sequence yields the detrended temperature change rate sequence. The detrended sequence retains the dynamic response component of the temperature change rate fluctuating with control actions, while eliminating low-frequency interference from the gradual trend on the cross-correlation results.

[0063] Calculate the cross-correlation function between the evaporator fan speed difference sequence and the detrended temperature change rate sequence, with the cross-correlation function in time delay. The value at the position is the result of shifting the two sequences. The normalized inner product follows. If the absolute value of the cross-correlation function is in... The maximum value was obtained at the 30-second sampling period, indicating that the statistical correlation between the evaporator fan speed adjustment event sequence and the cargo surface temperature change rate sequence was highest when time-aligned with a 30-second time delay. This means the estimated time required for heat transfer to the cargo surface after a step adjustment of the evaporator fan speed and the formation of a stable and identifiable response in the temperature change rate sequence is 30 seconds, with the time delay determined to be 3 sampling periods. The cross-correlation function between the airflow guide angle difference sequence and the detrended temperature change rate sequence was calculated. If the maximum absolute value is at... The data was obtained at one acquisition cycle (20 seconds), and the latency was determined to be 2 acquisition cycles.

[0064] The method for calculating the Cook distance for each data collection period by performing a second-order polynomial regression with the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable includes: The parameter space corresponding to the aligned evaporator fan speed sequence and airflow guide angle sequence is divided into uniform grids according to the respective value ranges of the evaporator fan speed and airflow guide angle. The average value of the cargo surface temperature change rate corresponding to the acquisition period falling into the same uniform grid is taken, and the average value of the evaporator fan speed and airflow guide angle at the center of the grid and the corresponding cargo surface temperature change rate constitutes the balanced acquisition sample sequence.

[0065] Using the evaporator fan speed and airflow guide angle in the balanced sample sequence as independent variables and the mean rate of change of the corresponding cargo surface temperature as the dependent variable, a second-order polynomial regression is performed to calculate the Cook distance of the balanced sample sequence.

[0066] For example, after time alignment, the evaporator fan speed sequence ranges from 600 to 1500 rpm, and the airflow guide angle sequence ranges from 15° to 60°. The grid step size is determined based on the actual scheduling characteristics of the refrigeration unit in the cold chain transportation scenario: during cold chain transportation, the evaporator fan speed of the refrigeration unit is adjusted stepwise according to the variable frequency drive gear, and the speed difference between adjacent adjustment gears is usually not less than 150 rpm; the airflow guide mechanism switches according to fixed deflection gears, and the deflection angle difference between adjacent gears is usually not less than 15°. The grid step size is taken as the minimum spacing between adjacent gears in each dimension, so that the same grid only contains the collection period of the refrigeration unit under the same stable operating gear. The collection periods of different operating gears are not included in the same grid. The grid mean can accurately represent the average thermal response under each stable operating state, and the number of effective grids is equal to the number of gear combinations that actually occur within the collection time window. Accordingly, the evaporator fan speed range is divided into 6 segments in 150 rpm increments, and the air delivery guide angle range is divided into 3 segments in 15° increments, forming a 6×3 uniform grid of 18. The evaporator fan speeds at the center of each grid are 675, 825, 975, 1125, 1275, and 1425 rpm, respectively, and the air delivery guide angles are 22.5°, 37.5°, and 52.5°, respectively.

[0067] The 117 time-aligned acquisition cycles are assigned to corresponding uniform grids based on the actual values ​​of the evaporator fan speed and airflow guide angle. The arithmetic mean of the rate of change of cargo surface temperature corresponding to the acquisition cycles falling into the same uniform grid is taken. A balanced acquisition is formed by the average of the evaporator fan speed, airflow guide angle and the corresponding rate of change of cargo surface temperature at the center of the grid. Finally, a balanced acquisition sample sequence of up to 18 is formed (if no acquisition cycle falls into a certain grid, the grid is not generated). In this embodiment, the evaporator fan speed goes through four levels within a 20-minute window: 750 rpm, 900 rpm, 1050 rpm, and 1200 rpm. The airflow guide angle goes through three levels: 30°, 45°, and 60°. The actual effective sampling period is generated by 12 grids: (750 rpm, 30°), (750 rpm, 45°), (900 rpm, 30°), (900 rpm, 45°), (900 rpm, 60°), (1050 rpm, 30°), (1050 rpm, 45°), (1200 rpm, 30°), (1200 rpm, 45°), (1200 rpm, 60°), (1050 rpm, 60°), and (750 rpm, 60°). The balanced sampling sample sequence contains 12 sequences with 6 degrees of freedom, which fully supports the Cook distance calculation of 6-parameter second-order polynomial regression. Within the preset acquisition time window, the refrigeration unit typically undergoes switching between more than 3 speed gears and more than 3 angle gears, and the effective number of grids is usually no less than 9. The number of samples collected in a balanced manner fully supports the Cook distance calculation of the 6-parameter second-order polynomial regression.

[0068] In actual transportation, refrigeration units may operate continuously for extended periods under certain combinations of evaporator fan speed and airflow angle, resulting in uneven distribution of data acquisition periods in the parameter space. Dense parameter intervals have a far greater impact on the regression results than sparse intervals. By uniformly representing each grid interval with the evaporator fan speed and airflow angle values ​​at the grid center, each parameter interval participates in the regression with equal weight, ensuring that the contribution of each parameter interval to the regression is not imbalanced by the number of data acquisition periods.

[0069] Using the evaporator fan speed and airflow angle in the balanced sample sequence as independent variables and the mean rate of change of the corresponding cargo surface temperature as the dependent variable, a second-order polynomial regression was performed, including the first and second terms of evaporator fan speed, the first and second terms of airflow angle, and their cross-product, to calculate the Cook distance for each.

[0070] The method for fitting the nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm after the noise reduction of the evaporator fan speed sequence and the airflow guide angle sequence includes: Using the denoised evaporator fan speed sequence and airflow guide angle sequence as independent variables and the corresponding cargo surface temperature change rate as dependent variable, a second-order complete polynomial containing the first and second terms of evaporator fan speed, the first and second terms of airflow guide angle, and their cross-product term are fitted by the least squares method to obtain the preliminary nonlinear response surface and the residual sequence of each acquisition period.

[0071] Calculate the arithmetic mean of the denoised evaporator fan speed sequence, and divide the denoised evaporator fan speed sequence into high speed segment and low speed segment using the arithmetic mean as the boundary; calculate the mean square residual of the residual sequence corresponding to the high speed segment and the low speed segment respectively, and determine the speed segment with larger mean square residual as the speed segment with significant nonlinear response.

[0072] Using the absolute value of the difference between the evaporator fan speed and the mean evaporator fan speed in each acquisition cycle within the significant speed range of the nonlinear response as the independent variable, the weight of each acquisition cycle is calculated using a Gaussian function. Weighted least squares regression is then performed on each acquisition cycle within the significant speed range of the nonlinear response using these weights to obtain a residual correction polynomial containing the first and second terms of the evaporator fan speed, the first and second terms of the airflow guide angle, and their cross-product. This residual correction polynomial is then added to the second-order complete polynomial of the initial nonlinear response surface within the corresponding speed interval of the significant speed range of the nonlinear response to obtain the correction polynomial for the significant speed range of the nonlinear response. Finally, the correction polynomial for the significant speed range of the nonlinear response is merged with the second-order complete polynomial of the initial nonlinear response surface outside the corresponding interval of the significant speed range of the nonlinear response to obtain the nonlinear response surface.

[0073] For example, the denoised evaporator fan speed sequence and the airflow guide angle sequence retain a total of 94 acquisition cycles after denoising. The evaporator fan speed in each acquisition cycle of the denoised evaporator fan speed sequence is used as the basis for this calculation. (rpm) and airflow guide angle (°) is the independent variable, corresponding to the rate of change of cargo surface temperature. (℃ / s) is the dependent variable. A second-order complete polynomial is fitted using the least squares method: ; For example, the fitting yields... , , , , , The preliminary nonlinear response surface was obtained, and the residuals for each acquisition cycle were recorded. ,in, To measure the rate of change of cargo surface temperature, These are the predicted values ​​for the initial nonlinear response surface.

[0074] Calculate the arithmetic mean of the evaporator fan speed over 94 data collection periods. ,by To demarcate the noise-reduced evaporator fan speed sequence, it is divided into a high-speed segment. ) and low speed range ( ), calculate the mean square residuals of the two corresponding residual sequences respectively. Taking rpm, high-speed range with 41 acquisition cycles, and low-speed range with 53 acquisition cycles as examples, the mean square residual of the high-speed range... (℃ / s)², mean square residual at low speed range (℃ / s)², The high-speed range was identified as the range with significant nonlinear response. When the evaporator fan operates in the high-speed range, the airflow is fully turbulent, and the nonlinearity of the heat transfer coefficient with respect to speed is significantly higher than that in the low-speed range. The global second-order polynomial of the preliminary nonlinear response surface is insufficient to describe the local response characteristics in the high-speed range, resulting in a systematically large residual in the high-speed range.

[0075] Average evaporator fan speed in the high-speed range of the nonlinear response rpm, taking the standard deviation of evaporator fan speed within the speed range where nonlinear response is significant. rpm, calculate the weight of each acquisition cycle: ; Rotation speed close to The weight of the sampling period for rpm approaches 1, and the speed deviates. Weight decay occurs at longer acquisition periods. (At segment boundaries...) Taking rpm as an example, The weight at the boundary has been reduced to 0.21, and the correction magnitude of the residual correction polynomial fitted by the weighted least squares regression near the boundary of the significant speed range of the nonlinear response converges accordingly, causing the correction polynomial of the significant speed range of the nonlinear response to converge with the initial nonlinear response surface at the boundary speed. The splicing jump at rpm is within an acceptable range.

[0076] With weights of 41 acquisition cycles Weighted least squares regression was performed on the significant speed range of the nonlinear response, using the residuals of each acquisition cycle. The dependent variable is the corresponding evaporator fan speed. With air supply and airflow angle Using the independent variable, fit the residual correction polynomial: ; The residual correction polynomial With the initial nonlinear response surface In the speed range corresponding to the significant nonlinear response speed segment ( By summing the values ​​within the range of rpm, a corrected polynomial with significant nonlinear response is obtained. The nonlinear response is significantly modified by the polynomial in the speed range. The initial nonlinear response surface in the significant speed range of the nonlinear response ( The second-order complete polynomial of the interval corresponding to rpm) By merging, a nonlinear response surface is obtained. : ; The method for setting the parameter combination of the preset evaporator fan speed and airflow guide angle includes: Obtain the minimum stable operating speed and the rated maximum speed of the inverter driver of the refrigeration unit; perform arithmetic progression discretization on the speed range between the minimum stable operating speed and the rated maximum speed to obtain the evaporator fan speed discrete sequence; obtain the minimum deflection angle and the maximum mechanical deflection angle of the air supply guide mechanism, and perform arithmetic progression discretization on the angle range between the minimum deflection angle and the maximum mechanical deflection angle to obtain the air supply guide angle discrete sequence.

[0077] The gas flow pressure is obtained by applying the dynamic pressure formula to the evaporator fan speed.

[0078] From all combinations of the discrete sequences of evaporator fan speed and airflow guide angle, retain the combinations where the airflow pressure is not lower than the preset airflow pressure threshold to obtain the preset parameter combination of evaporator fan speed and airflow guide angle.

[0079] For example, the minimum stable operating speed of the inverter drive of the chiller unit is 600 rpm, and the rated maximum speed is 1500 rpm. Discretizing the speed in 150 rpm increments yields a discrete sequence of evaporator fan speeds: {600, 750, 900, 1050, 1200, 1350, 1500} rpm, totaling 7 discrete values. The minimum deflection angle of the airflow guiding mechanism is 15°, and the maximum mechanical deflection angle is 60°. Discretizing the angle in 15° increments yields a discrete sequence of airflow guiding angles: {15°, 30°, 45°, 60°}, totaling 4 discrete values. The total number of combinations is 7 × 4 = 28.

[0080] Evaporator fan speed (rpm) and evaporator outlet gas flow pressure The pressure (Pa) and pressure (Pa) are proportional to the square of each other, indicating that the gas flow pressure is directly proportional to the pressure (Pa). Therefore and Proportional, that is Proportionality coefficient The calibration is performed at the rated operating point, with the rated maximum speed of 1500 rpm corresponding to an airflow pressure of 22 Pa as the calibration point. The corresponding air pressure at each speed is as follows: 3.5 Pa at 600 rpm, 5.5 Pa at 750 rpm, 7.9 Pa at 900 rpm, 10.8 Pa at 1050 rpm, 14.1 Pa at 1200 rpm, 17.8 Pa at 1350 rpm, and 22.0 Pa at 1500 rpm.

[0081] The larger the deflection angle, the stronger the adverse pressure gradient experienced by the airflow on the convex side of the blade, and the greater the minimum gas flow pressure required to maintain stable airflow along the blade surface. After blade aerodynamic testing and calibration, the minimum gas flow pressures corresponding to each deflection angle in the discrete sequence of supply air guide angles {15°, 30°, 45°, 60°} are 1.5 Pa, 4.2 Pa, 9.8 Pa, and 18.5 Pa, respectively. The preset gas flow pressure threshold is the maximum value among these minimum gas flow pressures, i.e., 18.5 Pa corresponding to a 60° deflection angle, to ensure that the retained parameter combinations meet the stable airflow adhesion condition at all deflection angles in the discrete sequence of supply air guide angles. Of all 28 combinations, only the gas flow pressure of 22.0 Pa corresponding to 1500 rpm meets the condition of not less than 18.5 Pa. The 27 combinations corresponding to 600 rpm to 1350 rpm were discarded, resulting in four preset parameter combinations of evaporator fan speed and supply air guide angle.

[0082] The method for predicting the temperature forecast value of the cargo surface temperature by using the estimated rate of change of cargo surface temperature to predict the cargo surface temperature using a linear extrapolation method includes: Obtain the historical cargo surface temperature change rate sequence within a preset number of historical collection periods, and obtain the temperature change rate trend slope through Theil-Sen slope estimation.

[0083] The length of the prediction time window is determined by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the trend of the rate of change of temperature.

[0084] The extrapolation termination rate is obtained by summing the product of the slope of the temperature change rate trend and the length of the prediction time window with the estimated value of the temperature change rate of the cargo surface. The temperature change within the prediction time window is integrated using the estimated value of the temperature change rate of the cargo surface and the extrapolation termination rate as the two endpoints to obtain the temperature change of the cargo surface.

[0085] The current surface temperature of the cargo is added to the change in surface temperature of the cargo to obtain the predicted surface temperature of the cargo.

[0086] For example, the preset number of historical data collection periods is set to 30 (corresponding to 5 minutes of historical data), thus obtaining a historical cargo surface temperature change rate sequence of length 30, denoted as the th... The rate of change in cargo surface temperature corresponding to each historical data collection period is ( During the temperature control process in the cargo compartment, the temperature change rate sequence of the cargo surface is affected by factors such as the adjustment action of the refrigeration unit and fluctuations in the external wall heat load, and the temperature change rate readings in some collection periods may show temporary deviations. Theil-Sen slope estimation is a nonparametric slope estimation method with high robustness to outliers. Its principle is to independently calculate the slope of each pair of data points in the sequence, and then take the median of the slopes of all pairs of points as the final trend slope, thus making the impact of individual deviations on the overall slope estimation result negligible. Theil-Sen slope estimation calculates the slope by pairwise pairing of 30 historical values, obtaining a total of pairwise slopes ( The median of the 435 slopes was taken as the slope of the temperature change rate trend. Median ensembles are performed with each slope as an independent estimator, thus minimizing the impact of individual outlier collection periods. The impact is negligible, and the slope obtained by directly performing least-squares linear fitting on 30 historical values ​​is more robust to noise. The temperature change rate over the 30 historical data collection periods is used as an example. 0.006 to Taking a gradually changing slope between 0.003℃ / s as an example, the median of 435 pairwise slopes. ℃ / s², with a positive sign, indicates that the rate of change of the cargo surface temperature is approaching zero, meaning that the cooling rate is continuously slowing down.

[0087] Prediction time window length Estimated value from the rate of change of cargo surface temperature slope of the trend of temperature change rate The absolute value of the ratio is determined, and thus we obtain Second.

[0088] Extrapolation termination rate The slope of the temperature change rate trend With the length of the prediction time window The product plus the estimated rate of change of cargo surface temperature : ; ℃ / s indicates that at At the end of the second, the rate of change of cargo surface temperature approaches zero, meaning the cooling process tends to stop at the end of the prediction window. Estimated rate of change of cargo surface temperature. With extrapolation termination rate The rate values ​​at both ends of the prediction time window, under the assumption of a linear trend, represent the rate of change of cargo surface temperature within the prediction window. For a linear function of time, the trapezoidal rule is used... exist The integral over is equal to the exact value: ; Based on the current surface temperature of the cargo Taking ℃ as an example, the predicted temperature value for the surface temperature of the cargo is: ; The method for determining the prediction time window length by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the rate of change trend includes: The basic time window is obtained by dividing the absolute value of the estimated rate of change of cargo surface temperature by the absolute value of the slope of the trend of the rate of change of temperature; the sign of the thermodynamic state discrimination sign is obtained by the product of the estimated rate of change of cargo surface temperature and the slope of the trend of the rate of change of temperature.

[0089] When the thermodynamic state discrimination sign is negative, the base time window is used as the prediction time window length.

[0090] When the thermodynamic state discrimination sign is positive, the basic time window is multiplied by a preset time coefficient to obtain the prediction time window length.

[0091] For example, the estimated rate of change of cargo surface temperature. ℃ / s, slope of the rate of temperature change ℃ / s², base time window: ; The base time window is based on the current rate of change of cargo surface temperature. With linear trend Under the combined effect, the rate of change of cargo surface temperature from The time required to extrapolate the trend to zero, i.e. Time corresponding .

[0092] Thermodynamic state discrimination symbol plus or minus sign: The discriminant sign is negative. mean and The signs are opposite: the current rate of change of the cargo surface temperature is negative (continuous cooling), while the slope of the temperature change rate trend is positive (the cooling rate is continuously slowing down). The rate of change of the cargo surface temperature is approaching zero, and the thermodynamic process is in a self-converging deceleration state. In this decelerating convergence state, That is, the moment when the cooling rate tends to stop. The end point of the prediction time window has a clear physical meaning; the deviation between the extrapolated linear rate within the prediction window and the actual thermodynamic process decreases over time rather than accumulates. A negative sign indicates the length of the prediction time window. Second.

[0093] by ℃ / s、 ℃ / s² represents another scenario: The discriminant sign is positive. and With the same sign, both the cooling rate and the slope of the temperature change rate are negative, and the temperature change rate of the cargo surface is continuously shifting towards a more significant negative value, indicating that the thermodynamic process is in an accelerated deviation state. (Base time window under accelerated deviation state) Second, The corresponding physical endpoint is no longer the moment when the speed approaches zero, but rather the moment when the speed decreases from zero. Accelerate to ℃ / s At a time of ℃ / s, linear extrapolation over the entire 625-second interval will amplify the accelerating deviation trend, and the prediction error will accumulate significantly as the time window increases. Multiplying the base time window by a preset time coefficient of 1 / 2... The time factor is set to 1 / 2, narrowing the prediction range to the near end of the extrapolation linearity range where the extrapolation linearity is relatively reliable. The basis for selecting a preset time coefficient of 1 / 2 is that the error term and... Proportional Halving the time coefficient reduces the error to 1 / 4 of its original value, achieving a balance between predictive time-domain effectiveness and error controllability. Halving it to 1 / 3 reduces the error to 1 / 9 (T_w = 208 seconds), but the prediction time domain is too short, resulting in insufficient margin for predicting slow heating processes. Halving it to 2 / 3 reduces the error to only 4 / 9 (T_w = 417 seconds), showing limited reduction. Therefore, 1 / 2 is chosen as the preset time coefficient to balance prediction time domain and error control.

[0094] The method for obtaining the target operating parameter combination of the chiller unit by sorting the candidate combinations of compliance using an algorithm that approximates the ideal solution includes: From the set of candidate combinations that meet the criteria, the positive ideal solution is formed by the minimum evaporator fan speed and the maximum predicted temperature; the negative ideal solution is formed by the maximum evaporator fan speed and the minimum predicted temperature.

[0095] Based on the range of evaporator fan speed and temperature prediction values ​​in the candidate combination set, the range of evaporator fan speed and temperature prediction values ​​for each parameter combination is normalized, and the Euclidean distance between the normalized parameter combination and the positive ideal solution and the negative ideal solution is calculated.

[0096] The relative closeness is obtained by dividing the Euclidean distance between each parameter combination and the negative ideal solution by the sum of the Euclidean distances between the positive and negative ideal solutions; the parameter combination with the largest relative closeness is determined as the target operating parameter combination of the refrigeration unit.

[0097] For example, the predicted temperature values ​​for all parameter combinations in the candidate set of compliant combinations have met the constraint of not exceeding 8°C, ensuring cargo safety within the set range. Under the premise of meeting safety requirements, the lower the evaporator fan speed, the lower the power consumption of the refrigeration unit; the higher the predicted temperature value (closer to but still below 8°C), the smaller the cooling effect on the cargo and the less over-cooling. Therefore, minimizing the evaporator fan speed and maximizing the predicted temperature value are the two optimization directions for extracting from the candidate set of compliant combinations. rpm and ℃ constitutes a positive ideal solution. rpm and ℃ constitutes a negative ideal solution. and The positive ideal solution comes from different combinations of parameters in the set. The reference point does not correspond to any actual parameter combination in the set; the same applies to the negative ideal solution.

[0098] Range normalization was performed on the predicted evaporator fan speed and temperature for each parameter combination: ; The coordinates of the normalized positive ideal solution are The coordinates of the negative ideal solution are The Euclidean distance and relative closeness between each parameter combination and the two ideal solutions are as follows: ; Taking an evaporator fan speed of 1050 rpm and an airflow angle of 30° as an example, the predicted temperature value is... ℃, after normalization , , , , Taking an evaporator fan speed of 1350 rpm and an airflow angle of 30° as an example, the predicted temperature value is... ℃, after normalization , The normalized coordinates coincide with the negative ideal solution. , After calculating each of the six parameter combinations, the relative closeness is found to be 1050 rpm for the evaporator fan speed and 30° for the airflow guide angle. The maximum value among the six parameter combinations is determined as the target operating parameter combination for the refrigeration unit. Based on this, the evaporator fan speed is adjusted to 1050 rpm and the air delivery guide angle is adjusted to 30° for temperature control.

[0099] Example 2: Based on the same inventive concept, such as Figure 2As shown, this embodiment also provides a contactless cold chain transportation intelligent temperature control system based on temperature prediction, the system comprising: The data acquisition module is used to collect the surface temperature value of the cargo through a non-contact infrared temperature sensor; based on the surface temperature value of the cargo and the historical temperature data of the corresponding cargo location, the rate of change of the surface temperature of the cargo is calculated by differential calculation; and the evaporator fan speed and airflow guide angle of the refrigeration unit are collected.

[0100] The response surface modeling module is used to construct a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle by nonlinear regression fitting.

[0101] The temperature trend prediction module is used to substitute the parameter combination of preset evaporator fan speed and air delivery guide angle into the nonlinear response surface to obtain the estimated value of the cargo surface temperature change rate; and to predict the temperature prediction value of the cargo surface temperature by using the estimated value of the cargo surface temperature change rate through linear extrapolation.

[0102] The temperature control module is used to select parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold, thereby obtaining a set of candidate combinations that meet the standards; the candidate combinations that meet the standards are sorted by an algorithm that approximates the ideal solution to obtain the target operating parameter combination of the refrigeration unit; and the temperature is controlled according to the target operating parameter combination of the refrigeration unit.

[0103] It should be noted that the specific ways in which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0104] Finally, it should be noted that although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A contactless cold chain transportation intelligent temperature control method based on temperature prediction, characterized in that, The method includes: The surface temperature of the cargo is collected using a non-contact infrared temperature sensor; the rate of change of the cargo surface temperature is calculated by differential calculation based on the cargo surface temperature and the historical temperature data of the corresponding cargo location; the evaporator fan speed and airflow guide angle of the refrigeration unit are also collected. The nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle was constructed by nonlinear regression fitting of the evaporator fan speed, airflow guide angle and cargo surface temperature change rate. Substituting the parameter combination of preset evaporator fan speed and airflow guide angle into the nonlinear response surface, an estimated value of the rate of change of cargo surface temperature is obtained; the estimated value of the rate of change of cargo surface temperature is then used to predict the temperature of the cargo surface using a linear extrapolation method. A set of parameter combinations is obtained by selecting parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold; the compliant candidate combinations are then sorted using an algorithm that approximates the ideal solution to obtain the target operating parameter combination for the refrigeration unit; and temperature regulation is performed based on the target operating parameter combination for the refrigeration unit.

2. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 1, characterized in that, The method for constructing a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle through nonlinear regression fitting includes: Within a preset data acquisition window, the evaporator fan speed sequence, airflow guide angle sequence, and cargo surface temperature change rate sequence are obtained based on the evaporator fan speed, airflow guide angle, and cargo surface temperature change rate. Cross-correlation functions are calculated between the evaporator fan speed sequence, airflow guide angle sequence, and cargo surface temperature change rate sequence. The time delay corresponding to the maximum absolute value of each cross-correlation function is selected, and the evaporator fan speed sequence and airflow guide angle sequence are shifted and aligned to obtain time-aligned evaporator fan speed sequence and airflow guide angle sequence. Using the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable, a second-order polynomial regression was performed to calculate the Cook distance corresponding to each acquisition period; based on the Cook distance, the time-aligned evaporator fan speed sequence and airflow guide angle sequence were denoised to obtain the denoised evaporator fan speed sequence and airflow guide angle sequence. The denoised evaporator fan speed sequence and airflow guide angle sequence are fitted with a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm.

3. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 2, characterized in that, The method for calculating the cross-correlation function between the evaporator fan speed sequence, the airflow guide angle sequence, and the cargo surface temperature change rate sequence includes: First-order differences were performed on the evaporator fan speed sequence and the airflow guide angle sequence to obtain the evaporator fan speed difference sequence and the airflow guide angle difference sequence, respectively; linear detrending was performed on the cargo surface temperature change rate sequence to obtain the detrended temperature change rate sequence. Calculate the cross-correlation function between the evaporator fan speed difference sequence and the detrended temperature change rate sequence; calculate the cross-correlation function between the air supply guide angle difference sequence and the detrended temperature change rate sequence.

4. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 2, characterized in that, The method for calculating the Cook distance for each data collection period by performing a second-order polynomial regression with the aligned evaporator fan speed sequence and airflow guide angle sequence as independent variables and the cargo surface temperature change rate sequence as dependent variable includes: The parameter space corresponding to the aligned evaporator fan speed sequence and air delivery guide angle sequence is divided into uniform grids according to the respective value ranges of the evaporator fan speed and air delivery guide angle. The average value of the cargo surface temperature change rate corresponding to the collection period falling into the same uniform grid is taken, and the average value of the evaporator fan speed and air delivery guide angle at the center of the grid and the corresponding average value of the cargo surface temperature change rate are used to form a balanced collection sample sequence. Using the evaporator fan speed and airflow guide angle in the balanced sample sequence as independent variables and the mean rate of change of the corresponding cargo surface temperature as the dependent variable, a second-order polynomial regression is performed to calculate the Cook distance of the balanced sample sequence.

5. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 2, characterized in that, The method for fitting the nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle using a response surface analysis algorithm after the noise reduction of the evaporator fan speed sequence and the airflow guide angle sequence includes: Using the denoised evaporator fan speed sequence and air delivery guide angle sequence as independent variables and the corresponding cargo surface temperature change rate as dependent variable, a second-order complete polynomial containing the first and second terms of evaporator fan speed, the first and second terms of air delivery guide angle, and their cross-product term are fitted by the least squares method to obtain the preliminary nonlinear response surface and the residual sequence of each acquisition period. Calculate the arithmetic mean of the denoised evaporator fan speed sequence, and divide the denoised evaporator fan speed sequence into high speed segment and low speed segment using the arithmetic mean as the boundary; calculate the mean square residual of the residual sequence corresponding to the high speed segment and the low speed segment respectively, and determine the speed segment with larger mean square residual as the speed segment with significant nonlinear response. Using the absolute value of the difference between the evaporator fan speed and the mean evaporator fan speed in each acquisition cycle within the significant speed range of the nonlinear response as the independent variable, the weight of each acquisition cycle is calculated using a Gaussian function. Weighted least squares regression is then performed on each acquisition cycle within the significant speed range of the nonlinear response using these weights to obtain a residual correction polynomial containing the first and second terms of the evaporator fan speed, the first and second terms of the airflow guide angle, and their cross-product. This residual correction polynomial is then added to the second-order complete polynomial of the initial nonlinear response surface within the corresponding speed interval of the significant speed range of the nonlinear response to obtain the correction polynomial for the significant speed range of the nonlinear response. Finally, the correction polynomial for the significant speed range of the nonlinear response is merged with the second-order complete polynomial of the initial nonlinear response surface outside the corresponding interval of the significant speed range of the nonlinear response to obtain the nonlinear response surface.

6. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 1, characterized in that, The method for setting the parameter combination of the preset evaporator fan speed and airflow guide angle includes: Obtain the minimum stable operating speed and the rated maximum speed of the inverter driver of the refrigeration unit; perform arithmetic progression discretization on the speed range between the minimum stable operating speed and the rated maximum speed to obtain the evaporator fan speed discrete sequence; obtain the minimum deflection angle and the maximum mechanical deflection angle of the air supply guide mechanism; perform arithmetic progression discretization on the angle range between the minimum deflection angle and the maximum mechanical deflection angle to obtain the air supply guide angle discrete sequence. The gas flow pressure is obtained by applying the dynamic pressure formula to the evaporator fan speed; From all combinations of the discrete sequences of evaporator fan speed and airflow guide angle, retain the combinations where the airflow pressure is not lower than the preset airflow pressure threshold to obtain the preset parameter combination of evaporator fan speed and airflow guide angle.

7. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 1, characterized in that, The method for predicting the temperature forecast value of the cargo surface temperature by using the estimated rate of change of cargo surface temperature to predict the cargo surface temperature using a linear extrapolation method includes: Obtain the historical cargo surface temperature change rate sequence within a preset number of historical collection periods, and obtain the temperature change rate trend slope through Theil-Sen slope estimation; The length of the prediction time window is determined by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the trend of the rate of change of temperature. The extrapolation termination rate is obtained by summing the product of the slope of the temperature change rate trend and the length of the prediction time window with the estimated value of the temperature change rate of the cargo surface. The temperature change within the prediction time window is integrated using the estimated value of the temperature change rate of the cargo surface and the extrapolation termination rate as the two endpoints to obtain the temperature change of the cargo surface. The current surface temperature of the cargo is added to the change in surface temperature of the cargo to obtain the predicted surface temperature of the cargo.

8. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 7, characterized in that, The method for determining the prediction time window length by the absolute value of the ratio of the estimated rate of change of cargo surface temperature to the slope of the rate of change trend includes: The basic time window is obtained by dividing the absolute value of the estimated rate of change of cargo surface temperature by the absolute value of the slope of the trend of the rate of change of temperature; the sign of the thermodynamic state discrimination sign is obtained by the product of the estimated rate of change of cargo surface temperature and the slope of the trend of the rate of change of temperature. When the thermodynamic state discrimination sign is negative, the base time window is used as the prediction time window length; When the thermodynamic state discrimination sign is positive, the basic time window is multiplied by a preset time coefficient to obtain the prediction time window length.

9. The intelligent temperature control method for contactless cold chain transportation based on temperature prediction according to claim 1, characterized in that, The method for obtaining the target operating parameter combination of the chiller unit by sorting the candidate combinations of compliance using an algorithm that approximates the ideal solution includes: From the set of candidate combinations that meet the criteria, the positive ideal solution is formed by the minimum evaporator fan speed and the maximum predicted temperature; the negative ideal solution is formed by the maximum evaporator fan speed and the minimum predicted temperature. Based on the range of evaporator fan speed and temperature prediction value in the candidate combination set, the range of evaporator fan speed and temperature prediction value for each parameter combination is normalized, and the Euclidean distance between the normalized parameter combination and the positive ideal solution and the negative ideal solution is calculated. The relative closeness is obtained by dividing the Euclidean distance between each parameter combination and the negative ideal solution by the sum of the Euclidean distances between the positive and negative ideal solutions; the parameter combination with the largest relative closeness is determined as the target operating parameter combination of the refrigeration unit.

10. A contactless cold chain transportation intelligent temperature control system based on temperature prediction, used to perform the method according to any one of claims 1 to 9, characterized in that, The system includes: The data acquisition module is used to collect the surface temperature value of the cargo using a non-contact infrared temperature sensor; based on the cargo surface temperature value and the historical temperature data of the corresponding cargo location, the rate of change of the cargo surface temperature is calculated by differential calculation; and the evaporator fan speed and airflow guide angle of the refrigeration unit are collected. The response surface modeling module is used to construct a nonlinear response surface of the cargo surface temperature change rate with respect to the evaporator fan speed and airflow guide angle by nonlinear regression fitting of the evaporator fan speed, airflow guide angle and cargo surface temperature change rate. The temperature trend prediction module is used to substitute the parameter combination of preset evaporator fan speed and air delivery guide angle into the nonlinear response surface to obtain the estimated value of the cargo surface temperature change rate; and to predict the temperature prediction value of the cargo surface temperature by using the estimated value of the cargo surface temperature change rate through linear extrapolation. The temperature control module is used to select parameter combinations in which the predicted temperature values ​​inside the carriage do not exceed the preset upper temperature threshold, thereby obtaining a set of candidate combinations that meet the standards; the candidate combinations that meet the standards are sorted by an algorithm that approximates the ideal solution to obtain the target operating parameter combination of the refrigeration unit; and the temperature is controlled according to the target operating parameter combination of the refrigeration unit.

Images