A cold machine regulation method and system for cold chain logistics

Abstract

The application discloses a cold machine regulation and control method and system for cold chain logistics, and relates to the technical field of cold machine regulation and control. The cold machine regulation and control method for cold chain logistics collects environment data, cargo state data and cold machine operation state data of a cold chain carriage; performs feature mining and evaluation processing on the cold machine operation state data, extracts a cold machine health evaluation value, and estimates a cold machine effective refrigerating capacity; performs classification analysis and fusion processing on the environment data and the cargo state data to obtain a heat load demand of the cold chain carriage; based on the cold machine effective refrigerating capacity and the heat load demand, a comprehensive cost function is constructed, and multi-objective dynamic optimization solving processing is performed to obtain a cold machine control sequence. The cold machine is regulated and controlled through the cold machine control sequence, so that the cold machine output is dynamically adapted to the actual demand of the carriage, the deterioration of goods caused by insufficient refrigerating capacity is effectively avoided, the preservation effect in the process of goods transportation is improved, and the control precision is improved.

Description

Technical Field

[0001] This invention relates to the field of refrigeration control technology, specifically to a refrigeration control method and system for cold chain logistics. Background Technology

[0002] Cold chain logistics refers to logistics activities in which goods are kept at a specified temperature throughout the entire transportation process. It is widely used in the circulation of temperature-sensitive goods such as fresh agricultural products, pharmaceutical preparations, and frozen foods. As the core temperature control equipment in cold chain vehicles, the refrigeration unit's operating status directly determines the stability of the temperature field inside the vehicle, thereby affecting the quality of goods and transportation safety.

[0003] As the cold chain logistics industry continues to expand and transportation scenarios become increasingly complex, higher demands are being placed on the precision of refrigeration unit control. Different transportation tasks have different requirements for temperature control. For example, long-distance trunk transportation needs to focus on dealing with the continuous heat transfer of the building envelope, while urban delivery faces the heat load impact caused by frequent door opening and closing. At the same time, the performance of the refrigeration unit itself will degrade during long-term operation, and the refrigeration efficiency will gradually decrease. If it is still operated in the conventional way, it may lead to increased energy consumption, decreased temperature control capability, or even premature equipment failure.

[0004] Existing technologies, such as the refrigeration air conditioning control method and system based on dynamic adjustment algorithm disclosed in patent application CN120488585A, determine the low-pressure and high-pressure zones within the cold chain logistics transport cabinet by using real-time reference pressure data from a barometer. This involves receiving and analyzing the real-time reference pressure data. A sliding window is established, and based on this window, the parameter graph of the real-time reference pressure data is segmented. The real-time pressure data obtained within the sliding window reveals that when the amount of cold air delivered is excessive, a low-pressure zone forms within the cold chain logistics transport cabinet, absorbing more humid heat, leading to frost and ice formation inside the cabinet and consuming more effective energy. Conversely, when the amount of cold air delivered is insufficient, a high-pressure zone forms, preventing the preset temperature from being reached. By using pressure data, the temperature inside the cold chain logistics transport cabinet can be obtained relatively accurately.

[0005] Based on the above findings, the limitations of existing technologies include at least the following problems: Existing technologies fail to comprehensively analyze environmental data of the refrigerated truck compartment, cargo status data, and the refrigeration unit's own operating status data. This makes it difficult to accurately identify the health degradation of the refrigeration unit during operation, and also makes it difficult to accurately quantify the actual heat load demand of the refrigerated truck compartment. Consequently, refrigeration unit control decisions are difficult to match with the actual needs of the cargo, and it is even more difficult to achieve a dynamic balance between refrigeration effect, energy consumption cost, and equipment wear and tear. This can easily lead to a mismatch between refrigeration capacity and actual heat load, which in turn affects the preservation quality of the cargo or causes energy waste. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a refrigeration control method and system for cold chain logistics, which solves the problem that existing technologies are unable to accurately match refrigeration control with actual needs, easily leading to cargo damage or energy waste.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a refrigeration unit control method for cold chain logistics, comprising the following steps: collecting environmental data, cargo status data, and refrigeration unit operating status data of the refrigeration unit compartment; performing feature mining and evaluation processing on the refrigeration unit operating status data to extract refrigeration unit health assessment values ​​and estimate the effective cooling capacity of the refrigeration unit; classifying, analyzing, and fusing the environmental data and cargo status data to obtain the heat load demand of the refrigeration unit compartment; constructing a comprehensive cost function based on the effective cooling capacity of the refrigeration unit and the heat load demand, and performing multi-objective dynamic optimization to obtain a refrigeration unit control sequence; and controlling the refrigeration unit based on the refrigeration unit control sequence.

[0008] Further, the specific steps for extracting the chiller health assessment value are as follows: dimensionality reduction processing is performed on the chiller operating status data to extract the chiller principal component vector; and the historical chiller principal component vector sequence is obtained and time-series prediction processing is performed to obtain the chiller principal component prediction vector; residual analysis is performed on the chiller principal component vector and the chiller principal component prediction vector to determine the chiller health assessment value.

[0009] Further, the specific steps for obtaining the refrigeration unit principal component prediction vector are as follows: Based on the historical refrigeration unit principal component vector sequence, construct a vector autoregressive model; based on a preset estimation algorithm, perform parameter estimation processing on the vector autoregressive model to determine the model parameters; substitute the model parameters into the vector autoregressive model to extract the refrigeration unit principal component prediction vector.

[0010] Furthermore, the specific steps of residual analysis are as follows: based on the refrigeration unit principal component vector and the refrigeration unit principal component prediction vector, extract the refrigeration unit residual vector; based on the refrigeration unit residual vector, extract the refrigeration unit statistical feature set; based on a preset mapping function, perform mapping processing on the refrigeration unit statistical feature set to generate the refrigeration unit health assessment value.

[0011] Furthermore, the specific steps for estimating the effective cooling capacity of the chiller are as follows: Based on the preset mapping relationship between health and cooling capacity, determine the basic cooling capacity corresponding to the chiller health assessment value; based on the chiller operating status data, correct the basic cooling capacity to generate the effective cooling capacity of the chiller.

[0012] Further, the specific steps for obtaining the heat load requirement of the cold chain compartment are as follows: Based on the environmental data and combined with the compartment structure parameter set stored in the database, analyze the heat transfer load of the compartment enclosure; perform heat capacity analysis processing on the cargo status data to determine the sensible heat load of the cargo in the cold chain compartment; obtain the door opening status signal of the cold chain compartment and, combined with the environmental data, analyze the door opening ventilation heat load of the cold chain compartment; comprehensively process the heat transfer load of the compartment enclosure, the sensible heat load of the cargo, and the door opening ventilation heat load to obtain the heat load requirement of the cold chain compartment.

[0013] Furthermore, the specific steps for constructing the comprehensive cost function are as follows: determine the decision variables of the chiller and construct a set of chiller cost components, including energy consumption cost components, cargo loss risk components, and equipment loss components; determine a set of cost weights based on the effective cooling capacity of the chiller and the heat load demand; and construct the comprehensive cost function based on the set of chiller cost components and the set of cost weights.

[0014] Further, the specific steps for determining the cost weight set are as follows: determine whether the effective cooling capacity of the chiller is higher than the heat load demand; if it is higher, analyze the cooling capacity surplus coefficient and determine the cost weight set; if it is not higher, analyze the cooling capacity gap value and determine the cost weight set.

[0015] Further, the specific steps for obtaining the chiller control sequence are as follows: comparing the effective cooling capacity of the chiller with the heat load demand; determining the set of constraints for optimization based on the comparison results; using the chiller decision variables as optimization variables and minimizing the comprehensive cost function as the optimization objective; and solving the optimization variables under the constraints of the set of constraints to obtain the chiller control sequence.

[0016] A refrigeration unit control system for cold chain logistics includes: a data acquisition module for collecting environmental data, cargo status data, and refrigeration unit operating status data of the refrigeration compartment; a health assessment and refrigeration prediction module for performing feature mining and assessment processing on the refrigeration unit operating status data, extracting refrigeration unit health assessment values, and predicting the effective cooling capacity of the refrigeration unit; a heat load analysis module for classifying, analyzing, and fusing the environmental data and cargo status data to obtain the heat load demand of the refrigeration compartment; a multi-objective optimization module for constructing a comprehensive cost function based on the effective cooling capacity of the refrigeration unit and the heat load demand, and performing multi-objective dynamic optimization to obtain a refrigeration unit control sequence; and a refrigeration unit control module for controlling the refrigeration unit based on the refrigeration unit control sequence.

[0017] The present invention has the following beneficial effects:

[0018] (1) The refrigeration control method for cold chain logistics comprehensively collects environmental data, cargo status data and refrigeration operating status data of the cold chain compartment. For the refrigeration side, the refrigeration health assessment value is accurately extracted through principal component analysis, time series prediction and residual analysis. Then, combined with the mapping relationship between health and cooling capacity and the correction of operating conditions, the effective cooling capacity of the refrigeration is accurately estimated, avoiding misjudgment of cooling capacity due to the decline of refrigeration health. For the compartment side, the heat load is calculated by analyzing environmental data, cargo heat capacity characteristics and door opening and ventilation status, and the real-time heat load demand is obtained. Based on the accurate matching of cooling capacity and heat load, the refrigeration output is dynamically adapted to the actual demand of the compartment by optimizing the control parameters, effectively avoiding cargo deterioration caused by insufficient cooling capacity or unnecessary loss caused by excessive cooling, improving the preservation effect of cargo during transportation, thereby improving the control accuracy.

[0019] (2) The refrigeration control method for cold chain logistics achieves multi-objective collaborative optimization by constructing a comprehensive cost function, analyzing the relationship between the effective cooling capacity of the refrigeration unit and the heat load demand, dynamically adjusting the cost weight set, and seeking the control sequence with the minimum comprehensive cost under constraints during the optimization process. This reduces economic losses caused by cargo damage through precise temperature control, reduces refrigeration unit energy consumption by reasonably adjusting operating parameters, and takes into account the health status of the refrigeration unit to reduce maintenance and replacement costs caused by excessive equipment wear. This reduces the comprehensive operating cost of cold chain transportation and improves the overall economic benefits.

[0020] (3) The refrigeration control method for cold chain logistics reduces the dimensionality of refrigeration operating status data through principal component analysis and combines it with vector autoregression model to achieve principal component vector time series prediction. It can complete the health assessment and refrigeration capacity prediction of refrigeration unit without manual intervention. During the heat load calculation, it automatically integrates multiple types of data and classifies and analyzes the load to adapt to the dynamic changes of different environments, goods and door opening scenarios. The cost weight set is dynamically adjusted according to the real-time comparison results of refrigeration capacity and heat load. The optimization solution process automatically adapts to the constraints and outputs the optimal control sequence, so as to quickly respond to environmental fluctuations, changes in the status of goods and refrigeration unit performance degradation during transportation, thereby reducing the intensity of manual operation and improving the stability and reliability of cold chain transportation.

[0021] (4) The refrigeration control system for cold chain logistics provides comprehensive basic data support for subsequent processes through the data acquisition module. The health assessment and refrigeration prediction module is responsible for the professional analysis of the refrigeration status. It extracts the refrigeration health assessment value through principal component analysis, time series prediction, residual analysis and other methods, and completes the accurate calculation of effective refrigeration capacity by combining the mapping relationship between health and refrigeration capacity. The heat load analysis module analyzes the load by classification and completes the fusion processing to obtain the heat load demand that fits the actual transportation scenario. The multi-objective optimization module focuses on minimizing the comprehensive cost. It constructs a comprehensive cost function based on the comparison results of refrigeration capacity and heat load, and obtains the optimal refrigeration control sequence through dynamic optimization. The refrigeration control module will accurately adjust the refrigeration based on the control sequence to achieve the optimized refrigeration output of the refrigeration.

[0022] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0023] Figure 1 This is a flowchart of a refrigeration control method for cold chain logistics according to the present invention.

[0024] Figure 2 This is a flowchart illustrating the specific steps involved in extracting refrigeration unit health assessment values ​​in a refrigeration unit control method for cold chain logistics according to the present invention.

[0025] Figure 3 This is a block diagram of a refrigeration control system for cold chain logistics according to the present invention. Detailed Implementation

[0026] Please see Figure 1 This invention provides a technical solution: a refrigeration unit control method for cold chain logistics, comprising the following steps: collecting environmental data, cargo status data, and refrigeration unit operating status data of the refrigeration unit compartment; performing feature mining and evaluation processing on the refrigeration unit operating status data to extract refrigeration unit health assessment values ​​and predict the (maximum) effective cooling capacity of the refrigeration unit; classifying, analyzing, and fusing the environmental data and cargo status data to obtain the heat load demand of the refrigeration unit compartment; constructing a comprehensive cost function based on the effective cooling capacity and heat load demand of the refrigeration unit, and performing multi-objective dynamic optimization solution processing to obtain the refrigeration unit control sequence; and controlling the refrigeration unit based on the refrigeration unit control sequence, namely: using the current optimal decision variable value in the refrigeration unit control sequence as a control command to adjust the compressor operating frequency, fan speed, and electronic expansion valve opening, so that the refrigeration unit performs refrigeration output according to the optimized operating parameters.

[0027] Specifically, such as Figure 2As shown, the specific steps for extracting the chiller health assessment value are as follows: Dimensionality reduction is performed on the chiller operating status data to extract the chiller principal component vector. Specifically, this involves constructing a historical training sample set of the chiller operating status. The sample set contains The data includes chiller operating status data at each historical sampling point (including compressor operating frequency, fan speed, electronic expansion valve opening, chiller evaporation temperature, chiller condensation temperature, chiller basic standby power consumption, operating pressure, current, voltage, and heat exchange efficiency). The data for each point in time includes... The dimension, denoted as the first The first historical sampling moment Data for each dimension ( ; );

[0028] Secondly, the sample set is preprocessed for standardization to eliminate the influence of different dimensions of data. The standardization formula is as follows:

[0029] ;in, , for the first Historical statistical mean of data in each dimension , for the first Historical statistical standard deviation of data in each dimension;

[0030] Next, calculate the covariance matrix of the standardized sample set. The formula is: ;

[0031] in, For the first The standardized data vector at each time point ( ), Standardize the mean vector of the sample set ( );

[0032] For covariance matrix Perform eigenvalue decomposition to obtain eigenvalues and corresponding eigenvectors (Each feature vector has a dimension of 1) ); Select the top contributors whose cumulative contribution rate reaches a preset threshold (preferably 85%). The eigenvectors corresponding to the eigenvalues ​​constitute the principal component loading matrix. ( );

[0033] Finally, the current moment Collected chiller operating status data ( The standardized data vector for the current moment is obtained by standardizing the data according to the above standardization formula. Combine it with the load matrix Perform matrix multiplication to obtain the principal component vector of the chiller at the current time. ( );

[0034] The system obtains the historical refrigeration unit principal component vector sequence, performs time series prediction processing to obtain the refrigeration unit principal component prediction vector, and performs residual analysis on the refrigeration unit principal component vector and the refrigeration unit principal component prediction vector to determine the refrigeration unit health assessment value.

[0035] The specific steps to obtain the refrigeration unit principal component prediction vector are as follows:

[0036] Based on the historical refrigeration unit principal component vector sequence, a vector autoregressive (VAR) model is constructed, specifically as follows: the order of the VAR model is selected as... That is: set the candidate set of orders as This range covers a reasonable fitting interval for the time-series correlation of the principal component vectors of the chiller, avoiding both underfitting due to excessively low order and computational redundancy due to excessively high order.

[0037] For each candidate order Calculate the corresponding AIC value. The AIC criterion formula is as follows: (in (This refers to the model's log-likelihood function value). The smaller the AIC value, the better the overall model performance. By traversing the candidate set, the order with the smallest AIC value is selected as the optimal lag order. ( (Determined by the AIC criterion); if there are multiple orders with an AIC value difference ≤ 0.5, the smaller order is preferred to reduce computational complexity and avoid overfitting. The VAR model of order 1 has the following expression:

[0038] ;

[0039] in, All The autoregressive coefficient matrix, for The random error term satisfies (Expected value is 0) (covariance matrix is) , for Positive definite matrix);

[0040] Based on a pre-defined estimation algorithm, parameter estimation is performed on the vector autoregressive (VAR) model to determine the model parameters. Specifically, the VAR model is rewritten in matrix form. ;

[0041] in, For the endogenous variable matrix, for The lagged terms matrix is ​​expressed as follows:

[0042] ;

[0043] For the estimated coefficient matrix, the parameter estimation formula is:

[0044] ;

[0045] The estimation coefficient matrix is ​​obtained by calculating using the above formula. That is, to determine all the parameters of the VAR model;

[0046] Substituting the model parameters into the vector autoregression model, the refrigeration unit principal component prediction vector is extracted. Specifically, the estimated coefficient matrix is... Substitution The most recent VAR model is selected. Historical principal component vector at each moment As input, the principal component vector of the chiller at the current moment is predicted using the following formula:

[0047] ;

[0048] in, for The corresponding autoregressive coefficient matrix is ​​calculated as follows: This is the prediction vector of the principal components of the refrigeration unit.

[0049] The specific steps of residual analysis are as follows:

[0050] Based on the refrigeration unit principal component vector and the refrigeration unit principal component prediction vector, the refrigeration unit residual vector is extracted. Specifically, this involves calculating the refrigeration unit principal component vector at the current moment. Principal component prediction vectors at corresponding times The residual vector is obtained by calculating the difference between the residual vector at the current time and the residual vector at the current time. The formula is:

[0051] ;

[0052] in, , The residual vector is the first Each component ( The magnitude of the residual vector reflects the degree of deviation between the current principal component vector and the predicted vector. The larger the deviation, the more abnormal the chiller's operating status.

[0053] Based on the chiller residual vector, a statistical feature set of the chiller is extracted, specifically by collecting data from the current time and the most recent time. The residual vector at each sampling time point ( (Preferred value: 20) Extract the four core statistical features of all residual vectors within this window to form the refrigeration unit statistical feature set. , The mean of the residuals reflects the overall degree of deviation of the residuals. The standard deviation of the residuals reflects the degree of dispersion of the residuals. The maximum residual value reflects the degree of maximum deviation. Residual variance reflects the amplitude of residual fluctuations;

[0054] Based on a preset mapping function, the statistical feature set of the chiller is mapped to generate chiller health assessment values. The mapping function is as follows:

[0055] ;

[0056] in, For the feature weight vector, For bias terms; The range of values ​​is , The closer the value is to 1, the better the health of the refrigeration unit. The closer the value is to 0, the worse the health of the refrigeration unit, and the more timely the maintenance is required.

[0057] Among them, the feature weight vector and bias terms The prediction data is trained using historical health data of the chiller and corresponding statistical feature sets. The training process employs gradient descent to minimize the prediction error. Once trained, the prediction data is fixed for use.

[0058] Collect historical operating data of the chiller, covering historical samples of all health states including normal operation, minor faults, and serious faults. The number of samples should be no less than 100 sets (to ensure training accuracy), and each set of samples should contain two core data points:

[0059] Input data: The statistical feature set of the chiller at the corresponding time point, which is completely consistent with the dimensions and definitions of the statistical feature set extracted from the chiller health assessment;

[0060] Tag data: Actual health status of the chiller at the corresponding moment. (Dimensionless, value range [0, 1]), manually marked by professional technicians based on chiller operation test reports and fault records. The marking rule is: when the chiller is fault-free and operating normally, When the chiller has a minor malfunction that does not affect normal operation, When the chiller has a serious malfunction that affects the control effect, ;

[0061] The collected training samples are preprocessed to ensure data validity: the 3σ criterion is used to remove outlier samples from the statistical feature set (if a certain feature value of a sample exceeds the mean of all samples of that feature ± 3 times the standard deviation, it is judged as an outlier sample and removed).

[0062] The preprocessed statistical feature set is normalized to map each feature value to the [0, 1] interval to eliminate the influence of dimensions. The preprocessed sample is divided into a training set (70%) and a validation set (30%) in a 7:3 ratio. The training set is used for parameter training, and the validation set is used to verify the training effect and prevent overfitting.

[0063] For feature weight vector and bias terms Perform initial assignments to ensure that gradient descent converges quickly:

[0064] Feature weight vector The initial values ​​are randomly initialized, with a range of [0.1, 0.3], to ensure that the initial weights of each feature are balanced and to avoid the initial weight of a certain feature being too large and affecting the training results.

[0065] Bias term The initial value is set to 0.5 (dimensionless) as the baseline offset of the mapping function to speed up training convergence.

[0066] Set the learning rate (range [0.01, 0.1], preferably 0.05) to control the step size of gradient descent; set the maximum number of iterations. (Value range [1000, 5000], preferably 3000) to prevent training from getting stuck in infinite iteration; set a convergence threshold (dimensionless, value range [10^{-4}, 10^{-3}]) to determine whether training is complete;

[0067] Minimizing the prediction error of the mapping function, i.e., minimizing the deviation between the predicted health assessment values ​​and the actual health status labels in the training set, involves the following steps:

[0068] Calculate the predicted value: Substitute the normalized statistical feature set from the training set into the preset mapping function to calculate the predicted health assessment value for each sample. The mean squared error (MSE) is used as the loss function to quantify the deviation between the predicted value and the actual label. The loss function is then applied to the feature weight vector. and bias terms Find the partial derivatives to obtain the gradient vector, and then update the parameters using gradient descent.

[0069] Repeat the above steps until one of the following two conditions is met, then stop training:

[0070] The maximum number of iterations has been reached. The change in the loss function is less than the convergence threshold. .

[0071] The feature weight vector after training and bias terms Substitute the values ​​into the validation set, calculate the predicted health assessment values ​​for the validation set, and calculate the mean squared error of the validation set. ;like This indicates that the parameter training is successful and the fitting accuracy meets the requirements for refrigeration unit health assessment; if If the learning rate and maximum number of iterations are not met, the training process is repeated until the validation is successful.

[0072] In this implementation plan, the dimensionality of the chiller operating status data is first reduced to extract principal component vectors, and redundant information is removed while core features are retained, making the chiller status representation more accurate. Then, based on the vector autoregression model, the historical principal component vectors are predicted over time to match the time-series correlation characteristics of chiller operation, resulting in a principal component prediction vector that is close to reality. The deviation between the actual and predicted vectors is explored through residual analysis, and multi-dimensional statistical features are extracted to construct a feature set, which comprehensively reflects the degree of chiller operation abnormality. Finally, the feature set is transformed into health assessment values ​​by the trained mapping function. The model parameters are trained and validated in multiple rounds to ensure the reliability of the assessment results.

[0073] Specifically, based on a preset mapping relationship between health and cooling capacity, the basic cooling capacity corresponding to the chiller's health assessment value is determined. This preset mapping relationship reflects the impact of the chiller's health status on the reduction of cooling capacity, and can be expressed as:

[0074] ;

[0075] in, For the chiller to be in a healthy state ( The rated cooling capacity (W) under =1) is determined by the factory parameters of the chiller; This is the cooling capacity attenuation coefficient;

[0076] The core purpose of pre-setting the mapping relationship between health and cooling capacity is to quantify the health assessment value of the chiller. The better the refrigeration unit's health condition, the more significant the impact of the reduction in cooling capacity. The closer to 1, the closer the cooling capacity is to the rated cooling capacity; the worse the health condition of the chiller ( The closer the value is to 0, the more significant the decrease in cooling capacity. The preset steps are as follows: Refrigeration unit health assessment value The value range is [0, 1] (dimensionless), and the chiller is in a fully healthy state ( When =1), the cooling capacity reaches the factory-rated cooling capacity. (W); The chiller is in a complete failure state ( When the temperature is 0, the cooling capacity approaches 0, which is consistent with the actual operating law of the refrigeration unit (a decline in health will lead to the performance degradation of core components such as the compressor and heat exchanger, thereby reducing the cooling capacity).

[0077] Considering the actual characteristics of cooling capacity decay in refrigeration systems (the effect of health status on cooling capacity decay is non-linear, not linearly decreasing), a power function form is preferred as the mapping relationship:

[0078] Initial formal assumption: Assume the mapping relationship is in the form of an exponential function. ,in This represents the basic cooling capacity (W) of the chiller under its current healthy condition. Rated cooling capacity (W). These are the health assessment values ​​for the refrigeration unit. This is the cooling capacity attenuation coefficient (dimensionless), used to adjust the rate of cooling capacity attenuation.

[0079] By substituting the two extreme health states of the chiller into the hypothetical form, we can verify its reasonableness:

[0080] When the chiller is completely healthy ( When =1), This is consistent with the preset premises and conforms to the actual operating status of the chiller;

[0081] When the cold machine completely fails ( When =0), This is consistent with the preset premise and fits the actual situation where there is no cooling output when the chiller fails;

[0082] Collect 3-5 groups of different health states ( We took the actual cooling capacities corresponding to 1.0, 0.8, 0.6, 0.4, and 0.2 respectively, and substituted them into the assumed power function form. Preliminary verification showed that this form could effectively fit the nonlinear relationship between "health level" and "cooling capacity," with a fitting error of less than 5%, confirming the rationality of the mapping relationship. The final preset mapping relationship between health and cooling capacity is:

[0083] ;

[0084] After the preset is completed, the fitting accuracy of the mapping relationship is further verified by combining experimental data of different health states of the chiller: if the fitting error (difference between actual cooling capacity and the calculated mapping value ÷ actual cooling capacity) ≤ 5%, the preset mapping relationship is confirmed to be effective; if the fitting error > 5%, the form of the mapping relationship is adjusted (the power function is still preferred, and only the attenuation coefficient is optimized). The initial calibration range is determined until the fitting accuracy requirement is met, and the above mapping relationship is finally determined for subsequent basic cooling capacity calculation.

[0085] Cooling capacity attenuation coefficient The steps to obtain it are as follows:

[0086] Acquire several sets of chiller health assessment values ​​and corresponding actual cooling capacity data. Substitute the actual cooling capacity and health assessment value of each set into the preset mapping relationship formula, perform logarithmic transformation on the formula to convert it into a linear form, and use the least squares method to fit the transformed linear relationship. By calculating the slope of the linear relationship, obtain the initial calibration value of the cooling capacity attenuation coefficient.

[0087] Substitute the initial calibration values ​​into the original mapping relationship, calculate the theoretical cooling capacity for each group under healthy conditions, and compare the relative error between the theoretical and actual cooling capacities. If the relative error for all groups is ≤5%, then... The initial calibration values ​​are valid; if there are groups with a relative error exceeding 5%, add 2-3 more healthy groups and repeat the above fitting steps until the relative errors of all groups meet the requirements, thus confirming that the accuracy requirements are met. Final calibration value;

[0088] Based on the chiller's operating status data, the basic cooling capacity is corrected to generate the effective cooling capacity of the chiller. Specifically, this involves selecting the evaporation temperature from the chiller's operating status data. Condensation temperature As a correction parameter, a cooling capacity correction coefficient is constructed. The correction factor reflects the impact of the deviation between the actual operating conditions and the rated operating conditions on the cooling capacity. The formula is:

[0089] ;

[0090] in, The rated evaporation temperature of the chiller (°C). The rated condensing temperature (°C) of the chiller is determined by the chiller's factory specifications; correction factor. The range of values ​​is When the actual evaporation temperature and condensation temperature deviate slightly from the rated values, A value close to 1 indicates that the operating condition has little impact on the cooling capacity.

[0091] Multiplying the base cooling capacity by the correction factor yields the maximum effective cooling capacity of the chiller, using the following formula:

[0092] ;

[0093] This value represents the maximum effective cooling capacity that the chiller can stably output under its current healthy state and operating conditions.

[0094] In this implementation plan, a mapping relationship between health status and cooling capacity that closely matches the operating pattern of the chiller is preset. The nonlinear decay impact of health status on cooling capacity is accurately quantified. After multiple rounds of verification to ensure the fit of the mapping relationship, the decay coefficient is calibrated to ensure that the basic cooling capacity calculation closely matches the actual performance of the chiller. At the same time, a correction coefficient is constructed by combining the evaporation and condensation temperatures in the actual operation of the chiller to compensate for the deviation between the rated operating conditions and the actual operating conditions, and the effective cooling capacity is obtained. This plan considers both the performance decay caused by the health status of the chiller and the real-time impact of the actual operating conditions, so that the final cooling capacity result is closer to the actual cooling output capacity of the chiller and avoids unreasonable control decisions due to the deviation of the cooling capacity prediction.

[0095] Specifically, the steps to obtain the heat load requirement of the cold chain truck compartment are as follows: Based on environmental data (including external temperature, internal temperature, air density, and specific heat capacity of air), and combined with the truck compartment structural parameter set stored in the database, the heat transfer load of the cold chain truck compartment enclosure is analyzed. Specifically, based on the external temperature of the truck compartment in the environmental data... Temperature inside the carriage Combined with the carriage structure parameters stored in the database (heat transfer coefficient of carriage enclosure structure) Heat transfer area of ​​the carriage enclosure structure According to the basic heat transfer formula, the heat transfer load of the carriage enclosure is calculated. The formula is:

[0096] ;

[0097] Among them, when the external temperature Higher than the internal temperature At that time, heat was transferred into the carriage from the outside. A positive value indicates a negative value indicates a negative value indicates a negative value. In this method, the absolute value is taken, and only the magnitude of the heat load is considered.

[0098] The sensible heat load of the cargo in the refrigerated truck compartment is determined by performing heat capacity analysis on the cargo status data (including the mass, actual temperature, and unit value of each type of cargo). Specifically:

[0099] Select the first Quality of Class of Goods Specific heat capacity (Retrieved by matching from the database according to the type of goods), actual temperature Target temperature (Based on the pre-set requirements for cargo preservation, i.e., by obtaining the suitable preservation temperature range for this type of cargo through channels such as cold chain logistics industry databases, cargo preservation technology manuals, and cargo characteristic descriptions provided by suppliers. If the logistics cycle is short, such as ≤24 hours, the median value of the suitable range is selected first; if the logistics cycle is long, such as ≥3 days, the lower limit of the suitable range is selected first; if the cargo is a special category that is sensitive to temperature (such as pharmaceutical cold chain, fresh sashimi), the fixed value specified in the industry standard is directly adopted); the formula for calculating the sensible heat load of a single type of cargo is:

[0100] ;

[0101] All inside the carriage The total sensible heat load of this type of cargo is:

[0102] ;

[0103] in, For the first Sensible heat load (W) of this type of cargo. The sensible heat load (W) of the cargo reflects the amount of cooling required for the cargo to cool from its initial temperature to its target temperature.

[0104] The system acquires the door opening status signal of the refrigerated truck compartment and, in conjunction with environmental data, analyzes the heat load of the refrigerated truck compartment's door opening ventilation. Specifically:

[0105] Obtain the door opening status signal of the refrigerated truck compartment (Collected synchronously in real-time, with the same frequency as the chiller operating status data and environmental data), where the door opening status signal is defined as: This indicates that the cold chain truck door is open at the current real-time sampling moment. This indicates that the cold chain truck door is closed at the current real-time sampling moment;

[0106] Simultaneously collect environmental data at the current real-time sampling moment and the internal volume of the cold chain vehicle stored in the database. (Fixed parameters, real-time invocation), combined with Real-time calculation of heat load from door opening and ventilation. The real-time calculation formula is:

[0107] ;

[0108] in, The current sampling time is the heat load of door opening and ventilation (unit: W), which is fully adapted to the real-time control rhythm of the chiller. The real-time sampling step size (s) of the binarized door opening state signal is consistent with the chiller data acquisition step size. Real-time calculation results when (door not open) No additional statistics are needed; the real-time heat load value is directly output, ensuring synchronization with the real-time control of the chiller.

[0109] The heat load requirements of refrigerated truck compartments are obtained by comprehensively processing the heat transfer load of the compartment enclosure, the sensible heat load of the cargo, and the heat load of door opening and ventilation. Specifically, it is as follows:

[0110] ;

[0111] in, , , These are the weighting coefficients for the heat transfer load of the carriage enclosure, the sensible heat load of the cargo, and the heat load of door opening and ventilation, respectively, and they satisfy the following conditions: ;

[0112] in, , , The steps to obtain it are as follows:

[0113] Select a cold chain truck with the same model and similar operating scenario as the current cold chain truck, and collect its operating data for the past 6 months. The sample size should be no less than 100 groups to ensure that the sample covers different seasons (summer, winter, spring and autumn), different types of goods (high-value sensitive goods, ordinary heat-resistant goods), and different door opening frequencies (multi-site delivery, direct transportation), so as to avoid the weight bias caused by a single sample.

[0114] From each set of operational data, the measured values ​​of three types of heat loads are accurately extracted, and the collected sample data are preprocessed using the 3σ criterion to remove abnormal samples.

[0115] Based on the preprocessed valid sample set, the initial candidate values ​​of the weighting coefficients are obtained by calculating the average proportion of each type of load. The total load of a single sample group is then calculated: For each valid sample group, the measured value of the total heat load of that group is calculated using the following formula:

[0116] ;

[0117] in This is the serial number of the valid sample;

[0118] Calculate the proportion of each load in a single sample group: For each valid sample group, calculate the proportion of the three types of heat loads in the total load of the group, and the sum of the proportions of the three types of loads in each sample group is 1;

[0119] The average percentage of each type of load in all valid samples is calculated as the initial candidate values ​​for the weighting coefficients. These initial candidate weights are calculated based on the historical sample averages and need to be adjusted according to the actual situation of current cold chain transportation to ensure that the weights align with the needs of the current transportation task. The adjustment process strictly adheres to... The specific modification rules for the constraints are as follows:

[0120] Adjustments based on cargo type: If the current shipment is high-value, temperature-sensitive cargo (such as fresh sashimi, pharmaceutical products, etc.), priority should be given to ensuring the weight of the cargo's sensible heat load to reduce the risk of damage. For example:

[0121] ;

[0122] ;

[0123] Remain unchanged;

[0124] If the goods being transported are low-value and heat-resistant (such as common fruits and vegetables, or goods converted from ambient temperature to cold chain), the weight of the heat transfer load of the building envelope can be appropriately increased, for example:

[0125] ;

[0126] ;

[0127] Remain unchanged;

[0128] After the correction is completed, it is necessary to verify whether the corrected weights satisfy the condition that the sum of the weights is 1, and at the same time ensure that the weight of each type is not less than 0.05, to avoid a type of heat load being ignored because its weight is 0. (It should be noted that when...) When the corresponding weight is 0, if the verification conditions are not met, the correction range needs to be fine-tuned until the requirements are met.

[0129] In this implementation plan, the heat transfer of the carriage enclosure, the sensible heat of the cargo, and the heat load of door ventilation are analyzed separately. Each type of load is accurately calculated based on corresponding data. At the same time, the heat load of door ventilation is calculated in real time to adapt to the real-time rhythm of refrigeration unit control and ensure the timeliness of load data. Based on this, the weight coefficient of the load is determined. Initial weights are first obtained based on historical operating data, and then dynamically adjusted in combination with actual transportation scenarios such as cargo type, so that the weight allocation fits the current transportation needs. Various types of loads are combined with their corresponding weights to complete a comprehensive calculation, so that the final heat load demand can fully reflect the actual cooling capacity demand of the refrigerated carriage and take into account the actual impact of various loads under different scenarios.

[0130] Specifically, the steps for constructing the comprehensive cost function are as follows: Determine the chiller decision variables and construct a set of chiller cost components, including energy consumption cost components, cargo loss risk components, and equipment loss components. Specifically, the chiller decision variables are the adjustable parameters of chiller operation, denoted as...

[0131] ,in, For compressor operating frequency, This refers to the fan speed. For the opening degree of the electronic expansion valve;

[0132] Energy consumption cost component is denoted as The energy consumption cost of the chiller during a single control cycle is represented by the following formula: ,in This represents the real-time power consumption of the chiller. , This refers to the basic standby power consumption of the chiller (unit: kW). , All are dimensionless power consumption coefficients. This refers to the compressor's maximum operating frequency (unit: Hz). This is the maximum speed of the fan. The unit energy consumption equivalent cost (unit: yuan / kWh) (for fuel-powered vehicles, this value is calculated from the real-time diesel / gasoline unit price and the cold engine energy consumption-fuel conversion coefficient; for pure electric vehicles, this value corresponds to the unit cost of discharging the power battery, which can be calculated by referring to the charging electricity price or operating cost).

[0133] in, , The steps to obtain it are as follows:

[0134] Historical data records of the chiller under stable operating conditions were extracted from the chiller's historical operation database. Stable operating conditions refer to a condition where the compressor operating frequency and fan speed fluctuate by less than 5% for at least 30 consecutive seconds, and the chiller is not in a start-stop transient or defrosting phase. For each data record that meets the stability conditions, the measured values ​​of compressor operating frequency, fan speed, and chiller real-time power consumption were collected, along with the corresponding maximum allowable compressor frequency and maximum allowable fan speed. A total of no less than 200 sets of valid data points were collected to form a sample set.

[0135] The standby periods when neither the compressor nor the fan is running are selected from historical data. The power consumption data of the whole machine during these periods is extracted and the arithmetic mean is calculated as the basic standby power consumption of the chiller.

[0136] For each set of data in the sample set, the compressor operating frequency is divided by its maximum allowable frequency to obtain the dimensionless compressor contribution variable; the fan speed is divided by the maximum allowable fan speed to obtain the dimensionless fan contribution variable. At the same time, the total net power consumption of the chiller corresponding to the set of data is calculated, which is the measured real-time power consumption of the whole machine minus the basic standby power consumption. Thus, the sample points required for regression analysis are constructed. Each sample point contains three values: compressor contribution variable, fan contribution variable, and chiller net power consumption.

[0137] A bivariate linear regression model is established with the net power consumption of the chiller as the dependent variable and the contributions of the compressor and fan as independent variables. The least squares method is used to estimate the parameters of the regression model. By solving the normal equations, the regression coefficients that minimize the sum of squared prediction errors for all sample points are obtained. These two regression coefficients are the desired power consumption coefficients. , ;

[0138] Cargo damage risk component is denoted as This represents the loss incurred by goods due to temperature deviating from the preservation threshold. The fan speed affects the efficiency of cold air circulation within the cargo compartment, and the opening of the electronic expansion valve affects the refrigeration capacity output of the refrigeration unit, thus collectively determining the cargo temperature. The calculation formula is as follows:

[0139] ,in, For the first Unit value of goods (unit: yuan / kg) For the first Weight of goods (unit: kg) For the first Dimensionless damage coefficient for this type of cargo For the first Actual temperature of the goods (unit: °C). For the first Critical preservation temperature for a type of goods (unit: °C) (the highest permissible temperature for maintaining quality for refrigerated goods or the lowest permissible temperature for frozen goods. This value is preset according to the type of goods and stored in the database. For common fresh produce, pharmaceuticals, and other goods, the preset temperature can be made with reference to national standards, industry specifications, or preservation parameters provided by the supplier). The fan uniformity coefficient represents the effect of fan speed on mitigating the risk of cargo damage.

[0140] in, The acquisition steps are as follows: For the same type of goods, collect historical data from no less than thirty independent transport batches. The data for each batch should include two parts: First, the temperature time series data of the representative location of the goods recorded by the refrigeration control system at fixed sampling intervals during transportation, as well as the critical preservation temperature corresponding to the goods; Second, the arrival quality inspection data recorded by the quality inspectors after the goods arrive at the destination, in accordance with national standards or enterprise specifications, including the total mass of the batch of goods and the mass of goods damaged due to excessive temperature. The criteria for judging the damage should be clearly defined and unified in advance according to the type of goods. For example, for fruits and vegetables, the criteria are that the surface area of ​​visible rot and mold exceeds a certain limit; for frozen products, the criteria are that the core temperature exceeds the allowable range and causes thawing; and for medicines, the criteria are that the cumulative time of temperature recording exceeding the allowable range exceeds the specified threshold.

[0141] For each transport batch, based on its temperature time series data, the cumulative effect of temperature exceeding the critical preservation temperature during the entire transport process is calculated. That is, for each sampling time, the difference between the temperature at that time and the critical preservation temperature is calculated. If the difference is positive, it is retained; if the difference is negative or zero, it is zero. The difference is multiplied by the sampling interval time and then summed to obtain the cumulative temperature deviation of the batch. This cumulative amount reflects the total heat load borne by the batch of goods during the entire transport process.

[0142] For each shipping batch, the actual damage ratio is calculated based on the incoming quality inspection data. That is, the mass of the damaged part in the batch is divided by the total mass of the batch, and then multiplied by 100% to obtain a damage ratio value ranging from zero to 100%.

[0143] The cumulative temperature deviation and the proportion of cargo damage calculated for each batch are used as a sample point to form a regression sample set. The number of sample points is the total number of batches participating in the calibration, which shall not be less than thirty.

[0144] A univariate linear regression model is established with the cumulative temperature deviation as the independent variable and the proportion of cargo damage as the dependent variable. The least squares method is used to estimate the model parameters. By solving for the regression coefficient that minimizes the sum of squared prediction errors for all sample points, the slope of the regression line is obtained. This slope represents the probability of loss or damage to this type of cargo. ;

[0145] The steps to obtain it are as follows:

[0146] Data records of the chiller under stable operating conditions were extracted from the historical operation database of the chiller. The collected data included fan speed and measured values ​​of temperature sensors at multiple representative locations inside the carriage.

[0147] The collected historical data was divided into several speed ranges according to the fan speed. For each speed range, the average standard deviation of the temperature at all measuring points within that range was calculated as the temperature uniformity index for that speed range. The larger the standard deviation, the more uneven the temperature distribution. The temperature uniformity index corresponding to the lowest speed range was designated as the baseline uniformity.

[0148] For each speed range, the improvement rate of temperature uniformity relative to the baseline operating condition is calculated. A linear regression is performed with the speed ratio as the abscissa and the uniformity improvement rate as the ordinate. The slope obtained from the regression is the fan uniformity coefficient. ;

[0149] Equipment loss components are denoted as The cost of loss of the chiller equipment within a single control cycle is represented by the following formula:

[0150] ,in Replacement cost of the chiller (unit: yuan) (This is the total cost required to purchase a new chiller of the same model and install it into a working condition. You can obtain the market price of the current model by asking the chiller manufacturer, or you can refer to the original value in the financial depreciation record). , All are dimensionless loss coefficients; This refers to the real-time health assessment value of the chiller;

[0151] in, , The steps to obtain it are as follows:

[0152] Collect historical data of the entire process from commissioning the chiller in brand new condition to its first major overhaul or scrapping, including the compressor operating frequency, health assessment value, final actual service life and corresponding replacement cost at each sampling time. For chillers that are still running, data on their running time can be collected, and the current cumulative running time can be used as a reference for part of the life loss.

[0153] Historical data is divided into multiple sample intervals according to runtime periods. Each interval corresponds to a continuous runtime. For each sample interval, the average compressor frequency and average health level within that interval are calculated, and the actual runtime of that interval is recorded.

[0154] Using the health loss portion (1 - average health) and normalized frequency for each sample interval as independent variables, and the corresponding equipment loss cost ratio as the dependent variable, a multiple linear regression method is used to fit the sample set, establishing a regression model with health loss and compressor frequency as independent variables and equipment loss ratio as the dependent variable. The obtained regression coefficients are then calculated. , ;

[0155] Determine the cost weight set based on the effective cooling capacity of the chiller and the heat load demand;

[0156] Based on the cost component set and cost weight set of the chiller, a comprehensive cost function is constructed, specifically as follows: The comprehensive cost function is denoted as... The comprehensive economic cost of refrigeration unit control is quantified using the following formula: .

[0157] The specific steps for determining the cost weight set are as follows: Determine whether the effective cooling capacity of the chiller is higher than the heat load demand, i.e., determine... Is it valid?

[0158] If it is higher, then analyze the excess cooling capacity coefficient and determine the cost weight set, specifically: calculate the excess cooling capacity coefficient. Based on this, the weighting of energy consumption cost components is analyzed. Cargo damage risk weighting Equipment loss component weight :

[0159] ;

[0160] ;

[0161] ;

[0162] The cost weight set is denoted as And satisfy ;

[0163] If it is not higher, then analyze the cooling capacity gap value and determine the cost weight set, specifically: calculate the cooling capacity gap value. , ; Calculate the gap coefficient Based on this, the weighting of energy consumption cost components is analyzed. Cargo damage risk weighting Equipment loss component weight :

[0164] ;

[0165] ;

[0166] .

[0167] The specific steps to obtain the chiller control sequence are as follows: Compare the effective cooling capacity of the chiller with the heat load demand; based on the comparison results, determine the set of constraints for optimization, specifically: the set of constraints includes... and The two together constitute a complete set of constraints, as follows: General limit constraint: Each decision variable is within its own physical limit range, i.e. , , ,in, , These are the minimum and maximum operating frequencies of the compressor, respectively. , These are the minimum and maximum speeds of the fan, respectively. , These represent the minimum and maximum opening degrees of the electronic expansion valve, respectively.

[0168] Targeted constraints based on comparison results: If the comparison result is... If the cooling capacity is insufficient, then increase the cooling capacity to compensate for the constraint. ,in The maximum allowable refrigeration capacity shortfall (the maximum allowable refrigeration capacity shortfall when the actual refrigeration capacity of the refrigeration unit is insufficient to meet the heat load demand; exceeding this value will cause the cargo temperature to rise beyond the allowable range, resulting in unacceptable cargo damage risk) is defined based on the type of cargo transported in the compartment, specifying its tolerance to temperature fluctuations. For refrigerated cargo, the temperature is usually required not to exceed a certain critical preservation temperature; for frozen cargo, the temperature is usually required not to fall below a certain critical temperature. Taking refrigerated cargo as an example, if the current cargo temperature is close to the critical preservation temperature, the allowable temperature rise is relatively small; if the current temperature is far below the critical value, the allowable temperature rise can be appropriately relaxed. This allowable temperature rise can be... Referring to national standards, industry specifications, and transportation requirements provided by suppliers, the total heat capacity of the air and cargo inside the wagon is calculated based on the internal volume of the wagon, the specific heat capacity of the air, and the total mass and specific heat capacity of the cargo. The total heat capacity of the wagon reflects the heat absorbed or released per unit temperature change. Multiplying the maximum allowable temperature rise by the total heat capacity of the wagon yields the maximum heat that can be accumulated within the allowable temperature rise range. This heat value represents the upper limit of net heat input that the wagon can withstand without causing unacceptable cargo damage. Dividing the maximum allowable heat accumulation by the refrigeration unit control cycle length yields the maximum allowable average refrigeration capacity gap, which is the maximum allowable refrigeration capacity gap, to avoid excessive cargo damage risk.

[0169] If the comparison result is If there is excess cooling capacity, then the upper limit constraint on cooling capacity will be increased. ,in This is the maximum effective cooling capacity of the chiller (unit: W), to avoid energy waste;

[0170] Using the refrigeration unit decision variables as optimization variables and minimizing the overall cost function as the optimization objective, specifically: the optimization variables are the refrigeration unit decision variables.

[0171] The optimization objective is to minimize the overall cost function. ,Right now: ;

[0172] Under the constraints of the set of conditions, the optimization variables are solved to obtain the chiller control sequence, which is as follows:

[0173] The Particle Swarm Optimization (PSO) algorithm is used as the optimization solution algorithm. First, the core parameters of the algorithm are set, such as: the particle swarm size is 30-80, the number of iterations is 50-150, the inertia weight is 0.6-0.9, and the acceleration factor is 1.4-1.8 (the two acceleration factors can be the same or different).

[0174] During the solution process, the current time t is used as the optimization time domain, and the chiller decision variables are... As the position vector of the particle, each particle corresponds to a set of candidate values ​​for decision variables, and all candidate values ​​strictly follow the above constraint set and must not exceed the constraint range; through iterative calculation by the algorithm, the particle position is updated generation by generation, and the particle position that minimizes the value of the comprehensive cost function is selected. The value of the decision variable corresponding to this particle is the specific value of the optimal decision variable at the current moment.

[0175] The chiller control sequence refers to the specific values ​​of the chiller decision variables at the current moment that satisfy all constraints and minimize the overall cost function. Its expression is: ,in The specific value for the optimal operating frequency of the chiller; The specific value for the optimal speed of the chiller; The specific value for the optimal opening degree of the electronic expansion valve of the chiller.

[0176] In this implementation plan, the costs of energy consumption, cargo loss, and equipment wear and tear are broken down. Each component is calibrated with key coefficients based on actual operating data, and precisely quantified by combining refrigeration unit decision variables with actual scenario characteristics. This ensures that the cost calculation closely matches the actual cost structure of cold chain operations. Furthermore, the cost weights are dynamically adjusted based on the matching of refrigeration capacity and heat load, allowing the weight allocation to adapt to the control requirements of different refrigeration conditions. The constructed comprehensive cost function can fully reflect the comprehensive economic cost of refrigeration unit control. When solving the problem, constraints are set in combination with physical limits and operating condition characteristics. Within the constraints, the optimal decision variables are sought, ensuring that the resulting control sequence not only meets the basic requirements of refrigeration unit operation and cargo preservation but also minimizes the comprehensive cost. This makes refrigeration unit control decisions more accurate, precisely balancing various operating costs and refrigeration effects.

[0177] Please see Figure 3 This invention provides a technical solution: a refrigeration unit control system for cold chain logistics, comprising: a data acquisition module for collecting environmental data, cargo status data, and refrigeration unit operating status data of the refrigeration compartment; a health assessment and refrigeration prediction module for performing feature mining and assessment processing on the refrigeration unit operating status data, extracting refrigeration unit health assessment values, and predicting the effective refrigeration capacity of the refrigeration unit; a heat load analysis module for classifying, analyzing, and fusing environmental data and cargo status data to obtain the heat load demand of the refrigeration compartment; a multi-objective optimization module for constructing a comprehensive cost function based on the effective refrigeration capacity and heat load demand of the refrigeration unit, and performing multi-objective dynamic optimization to obtain the refrigeration unit control sequence; and a refrigeration unit control module for controlling the refrigeration unit based on the refrigeration unit control sequence.

[0178] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0179] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A refrigeration unit control method for cold chain logistics, characterized in that, Includes the following steps: Collect environmental data, cargo status data, and refrigeration unit operating status data from refrigerated truck compartments; The chiller operating status data is subjected to feature mining and evaluation processing to extract chiller health assessment values ​​and predict the effective cooling capacity of the chiller. The environmental data and cargo status data are classified, analyzed, and fused to obtain the heat load requirements of the cold chain compartment; Based on the effective cooling capacity of the chiller and the heat load demand, a comprehensive cost function is constructed and multi-objective dynamic optimization is performed to obtain the chiller control sequence. The specific steps for constructing the comprehensive cost function are as follows: Determine the decision variables for the refrigeration unit and construct a set of refrigeration unit cost components, including energy consumption cost components, cargo damage risk components, and equipment loss components; Based on the effective cooling capacity of the chiller and the heat load demand, a cost weight set is determined; Based on the cost component set and cost weight set of the chiller, a comprehensive cost function is constructed; The specific steps to obtain the chiller control sequence are as follows: The effective cooling capacity of the chiller is compared with the heat load requirement; Based on the comparison results, the set of constraints for optimization is determined; Using the refrigeration unit decision variables as optimization variables and minimizing the comprehensive cost function as the optimization objective; Under the constraints of the aforementioned set of conditions, the optimization variables are solved to obtain the chiller control sequence, which is as follows: Particle swarm optimization algorithm is used as the optimization solution algorithm; During the solution process, the current time t is taken as the optimization time domain, the refrigeration decision variables are taken as the position vectors of the particles, each particle corresponds to a set of candidate values ​​of decision variables, and all candidate values ​​strictly follow the above constraint set and must not exceed the constraint range; Through iterative calculations, the particle positions are updated generation by generation, and the particle position that minimizes the comprehensive cost function is selected. The value of the decision variable corresponding to this particle is the specific value of the optimal decision variable at the current moment. The chiller control sequence refers to the specific values ​​of the chiller decision variables at the current moment that satisfy all constraints and minimize the overall cost function. Its expression is: ,in The specific value for the optimal operating frequency of the chiller; The specific value for the optimal speed of the chiller; The specific value for the optimal opening degree of the electronic expansion valve of the chiller; Based on the aforementioned chiller control sequence, the chiller is regulated and controlled.

2. The refrigeration control method for cold chain logistics according to claim 1, characterized in that, The specific steps for extracting the health assessment values ​​of the chiller are as follows: The chiller operating status data is subjected to dimensionality reduction processing to extract the chiller principal component vector; The historical refrigeration unit principal component vector sequence is obtained and time series prediction processing is performed to obtain the refrigeration unit principal component prediction vector. Residual analysis is performed on the refrigeration unit principal component vector and the refrigeration unit principal component prediction vector to determine the refrigeration unit health assessment value.

3. The refrigeration control method for cold chain logistics according to claim 2, characterized in that, The specific steps to obtain the refrigeration unit principal component prediction vector are as follows: Based on the historical refrigeration unit principal component vector sequence, a vector autoregressive model is constructed; Based on a preset estimation algorithm, the vector autoregression model is subjected to parameter estimation processing to determine the model parameters; Substitute the model parameters into the vector autoregressive model to extract the refrigeration unit principal component prediction vector.

4. The refrigeration control method for cold chain logistics according to claim 2, characterized in that, The specific steps of residual analysis are as follows: Based on the refrigeration unit principal component vector and the refrigeration unit principal component prediction vector, extract the refrigeration unit residual vector; Based on the refrigeration unit residual vector, extract the refrigeration unit statistical feature set; Based on a preset mapping function, the statistical feature set of the chiller is mapped to generate a chiller health assessment value.

5. The refrigeration control method for cold chain logistics according to claim 1, characterized in that, The specific steps for estimating the effective cooling capacity of a chiller are as follows: Based on the preset mapping relationship between health and cooling capacity, the basic cooling capacity corresponding to the health assessment value of the chiller is determined; Based on the chiller's operating status data, the basic cooling capacity is corrected to generate the chiller's effective cooling capacity.

6. The refrigeration control method for cold chain logistics according to claim 1, characterized in that, The specific steps to obtain the heat load requirements of refrigerated truck compartments are as follows: Based on the environmental data and the set of carriage structure parameters stored in the database, the heat transfer load of the cold chain carriage enclosure is analyzed. The cargo status data is processed by thermal capacity analysis to determine the sensible heat load of the cargo in the refrigerated truck compartment; Acquire the door opening status signal of the cold chain compartment and, in conjunction with the environmental data, analyze the heat load of the cold chain compartment's door opening ventilation. The heat load requirements of the cold chain compartment are obtained by comprehensively processing the heat transfer load of the compartment enclosure, the sensible heat load of the cargo, and the heat load of the door opening and ventilation.

7. The refrigeration control method for cold chain logistics according to claim 1, characterized in that, The specific steps for determining the cost weight set are as follows: Determine whether the effective cooling capacity of the chiller is higher than the heat load requirement; If it is higher, then analyze the excess cooling capacity coefficient and determine the cost weight set; If it is not higher, then analyze the cooling capacity gap value and determine the cost weight set.

8. A refrigeration control system for cold chain logistics, employing the refrigeration control method for cold chain logistics as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to collect environmental data, cargo status data, and refrigeration unit operating status data of the refrigerated truck compartment; The health assessment and cooling prediction module is used to perform feature mining and assessment processing on the chiller operating status data, extract the chiller health assessment value, and predict the effective cooling capacity of the chiller. The heat load analysis module is used to classify, analyze, and fuse the environmental data and cargo status data to obtain the heat load requirements of the cold chain compartment. The multi-objective optimization module is used to construct a comprehensive cost function based on the effective cooling capacity of the chiller and the heat load demand, and to perform multi-objective dynamic optimization to obtain the chiller control sequence. The chiller control module is used to control the chiller based on the chiller control sequence.

Images