Distributed dynamic defrosting control method for air cooler

Abstract

The invention discloses an air cooler distributed dynamic defrosting control method, and relates to the technical field of air cooler distributed defrosting, and the method comprises the following steps: obtaining the spatial position and wind direction and flow direction information of an air cooler in a refrigeration environment and the defrosting exhaust state of an adjacent air cooler, building a heat exchange state space-time perception map based on a space coordinate relation, and obtaining a heat exchange state space-time perception map; the determination module is used for determining whether the air cooler is interfered by defrosting airflow of adjacent air coolers; acquiring temperature, humidity, current and air volume data of the air coolers which are determined to be interfered by defrosting airflow of the adjacent air coolers, constructing a heat exchange characteristic curve of the target air cooler in the current time period, and performing residual error fitting calculation on the heat exchange characteristic curve and a standard heat exchange curve in a historical non-defrosting state; and determining the local heat exchange parameter distortion performance of the air cooler caused by the defrosting airflow interference of the adjacent air cooler. The problem that defrosting is triggered by mistake due to the fact that the air cooler is interfered by defrosting airflow is solved, and accurate recognition and dynamic control optimization of the defrosting requirement are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of distributed defrosting technology for air coolers, and particularly relates to a distributed dynamic defrosting control method for air coolers. BACKGROUND

[0002] The distributed dynamic defrosting control for air coolers is an intelligent defrosting management method applied to multi-point air cooling systems. The core of the method is to deploy multiple distributed control nodes in the air cooler system, collect real-time data such as temperature and humidity, surface frosting state, and fan operating parameters in each region, and upload the data to the central control module or edge computing node through the communication network (such as CAN bus, Modbus or wireless communication). The system determines whether to start the defrosting program according to the environmental state and operating condition fed back by each node, and uses the preset control algorithm (such as fuzzy control, PID control or adaptive algorithm) to realize on-demand, dynamic and local defrosting of the frosting area. The whole process mainly includes five key links: data acquisition, data transmission, intelligent judgment, defrosting execution and feedback adjustment. Firstly, the distributed sensing nodes continuously collect temperature and humidity and frosting information and upload them in real time. Secondly, the system analyzes the data and determines whether each air cooler reaches the defrosting threshold. Thirdly, defrosting control instructions (such as switching the heat exchange mode, starting the heating device or stopping the defrosting) are sent to the air coolers that meet the defrosting conditions. Fourthly, the defrosting effect is continuously monitored and the defrosting strategy is dynamically adjusted during the defrosting process. Finally, the defrosting process and effect are fed back for subsequent strategy optimization. Compared with the traditional timed or unified control defrosting method, this method can improve the defrosting efficiency and system energy efficiency, avoid unnecessary energy consumption and performance loss, and is particularly suitable for air cooler system scenes with complex environment and large changes in operating load.

[0003] The prior art has the following disadvantages: In a cold air fan distributed dynamic defrosting control method in a refrigeration environment, the cold air fans are distributed by region, and each independently executes dynamic defrosting control by sensing local temperature, humidity, air volume, current and other parameters. When a cold air fan does not enter the defrosting state, if the region where the cold air fan is located is disturbed by hot and humid air flow due to the defrosting of the adjacent cold air fan, the temperature and humidity of the region where the cold air fan is located may be disturbed for a short time, thereby causing abnormal fluctuations in the heat exchange performance parameters sensed by the cold air fan. Since the current distributed dynamic defrosting control technology is mainly based on the node self-sensing and self-judgment mechanism, it lacks the ability to identify the external interference source of abnormal local sensing data, so that the system cannot distinguish whether the parameter fluctuation is caused by the frost accumulation of the local machine or by the defrosting exhaust behavior of the external cold air fan. Therefore, in the above case, the existing cold air fan distributed dynamic defrosting control technology cannot accurately determine whether the cold air fan needs to perform defrosting operation according to the local heat exchange parameter distortion caused by the defrosting air flow interference of the adjacent fan. This misjudgment will cause the cold air fan that does not need to defrost to be frequently triggered to defrost, which not only interferes with the normal operation rhythm, but also affects the coordinated judgment of other nodes due to the error state being included in the system calculation model, thereby causing chain mis-triggering of the defrosting control chain, resulting in increased system energy consumption, reduced heat exchange efficiency, and intensified regional temperature fluctuations, which seriously affects the stability and reliability of the dynamic defrosting control of the distributed cold air fan network.

[0004] The above information disclosed in the BACKGROUND section merely to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0005] The purpose of the present application is to provide a cold air fan distributed dynamic defrosting control method to solve the problems in the background art.

[0006] In order to achieve the above-mentioned purpose, the present application provides the following technical solution: a cold air fan distributed dynamic defrosting control method, specifically comprising the following steps: S1, obtaining the spatial position, wind direction and flow direction information of the cold air fan in the refrigeration environment and the defrosting exhaust state of the adjacent cold air fan, constructing a heat exchange state space-time perception map based on the spatial coordinate relationship, and determining whether the cold air fan is disturbed by the defrosting air flow of the adjacent cold air fan; S2, collecting temperature, humidity, current and air volume data of the cold air fan which has been determined to be disturbed by the defrosting air flow of the adjacent cold air fan, constructing a heat exchange characteristic curve of the target cold air fan in the current period, and performing residual fitting calculation with the standard heat exchange curve in the historical non-defrosting state to determine the local heat exchange parameter distortion caused by the defrosting air flow disturbance of the adjacent cold air fan; S3, fuse the local heat exchange parameter distortion performance with the real-time heat exchange data of the air cooler, correct the fusion result through interference weight, input the corrected fusion data into the defrosting credible evaluation model, generate a defrosting credible evaluation parameter, and use the parameter to judge whether the air cooler really needs to perform defrosting operation according to the local heat exchange parameter distortion performance caused by the air flow interference of the adjacent air cooler defrosting; S4, generate a judgment label based on the defrosting credible evaluation parameter, the historical operation characteristics of the air cooler and the load wave of the refrigeration area, divide the air cooler into three states of needing immediate defrosting, needing delayed observation and not needing defrosting; S5, execute a differentiated defrosting regulation process according to the judgment label corresponding to the air cooler, and use the data generated by each air cooler in the current round to update the heat exchange state space perception map, so as to realize defrosting control optimization under dynamic regulation.

[0007] Preferably, S1 specifically includes the following steps: S101, obtain the spatial position of each air cooler by deploying a space recognition unit in the refrigeration environment, obtain the wind direction flow information of the air cooler under different operation states in combination with the wind direction and speed detection components arranged at the outlet of the air cooler, and identify whether the adjacent air cooler is in a defrosting exhaust state based on the air cooler operation control logic and the heating start signal; S102, establish a spatial coordinate relationship according to the spatial position of the air cooler and the wind direction flow information, take the air cooler in the defrosting exhaust state as a hot and humid disturbance source point, calculate its influence radius and directional propagation path by calling a three-dimensional disturbance diffusion prediction model, and then construct a heat exchange state space perception map containing the air flow interaction relationship between the air coolers; S103, identify the interference area in the heat exchange state space perception map by analyzing the spatial angle between the air cooler and the disturbance source point, the air flow overlapping area and the disturbance intensity mapping relationship, and determine whether the air cooler is interfered by the defrosting air flow of the adjacent air cooler according to the preset interference judgment rule.

[0008] Preferably, S103 specifically includes: Calculate the spatial angle between the air cooler and the disturbance source point, take the coincidence degree of the angle range and the main direction of the air flow as the first interference index, and select the potential interference direction through the spatial path with an angle less than the set interference threshold; Extract the spatial volume of the air flow overlapping area, construct the disturbance intensity mapping relationship of the volume ratio and the air flow velocity superposition strength as the second interference index, and mark the air flow propagation track and the overlapping area of the intensity distribution in the heat exchange state space perception map to identify the interference area meeting the interference index weight threshold; The preset interference judgment rules include an included angle threshold judgment, a disturbance intensity coverage ratio judgment, and an airflow interference duration judgment. The included angle threshold judgment is based on whether an included angle between an air outlet direction of the air cooler and a disturbance source direction is lower than a preset interference threshold. The disturbance intensity coverage ratio judgment is based on whether a proportion of a disturbed area on a heat exchange surface of the air cooler exceeds a preset limited proportion. The airflow interference duration judgment is based on whether a disturbance influence duration exceeds a preset time window. The air cooler that meets at least two of the three judgment conditions is determined as an air cooler interfered by defrost airflow of a neighboring air cooler.

[0009] Preferably, S2 specifically includes the following steps: S201, collecting, by temperature sensors, humidity sensors, current sensors, and air volume detection components arranged on the heat exchanger surface and the air inlet and outlet path inside the air cooler, temperature, humidity, current, and air volume data of the air cooler determined to be interfered by defrost airflow of a neighboring air cooler in a current operation period, and performing time alignment and data missing repair processing on different types of sensing data to ensure continuity and consistency of the input data; S202, integrating the processed temperature, humidity, current, and air volume data in time sequence and inputting the data into a heat exchange behavior modeling engine, extracting heat exchange characteristic factors by a multivariate feature extraction algorithm, and constructing a heat exchange characteristic curve of the target air cooler in the current period in the same time dimension to fully reflect the heat load change and operation state linkage relationship; S203, performing residual fitting calculation on the constructed heat exchange characteristic curve of the target air cooler in the current period and a standard heat exchange curve in a non-defrost state selected from historical data, analyzing the offset amplitude, change trend, and duration of the residual interval, extracting local heat exchange parameter distortion caused by interference of the air cooler by defrost airflow of a neighboring air cooler, and inputting the distortion into subsequent defrost judgment logic.

[0010] Preferably, S202 specifically includes: aligning the temperature, humidity, current, and air volume data in the same time axis, extracting continuous time period data segments at equal intervals by a sliding window mechanism, and completing and correcting missing or abnormal fluctuation data by a bidirectional interpolation method to ensure synchronization and data integrity of the multi-source data in the time sequence at each time point; inputting the time sequence aligned data into the heat exchange behavior modeling engine, extracting heat exchange characteristic factors from the temperature, humidity, current, and air volume data by a joint principal component analysis algorithm and a clustering decomposition algorithm, and eliminating redundant and low correlation factors to form a characteristic vector set for describing the heat exchange behavior state of the target air cooler; Map the eigenvector set in time series, and construct the heat exchange characteristic curve of the target air cooler in the current period, wherein the vertical axis is the refined heat exchange characteristic factor group, and the horizontal axis is the time axis consistent with the sensing data. The heat exchange characteristic curve reflects the response law of the current heat exchange capacity and operation state of the target air cooler under the dynamic change of heat load.

[0011] Preferably, S203 specifically comprises: In the historical operation data, the time period closest to the current refrigeration environment load state and operation condition characteristic is screened, and the standard heat exchange curve of the target air cooler in the time period is extracted to ensure that the standard curve does not contain data interference during defrosting, and serves as a control reference for the current heat exchange characteristic curve. The heat exchange characteristic curve of the target air cooler in the current period is compared with the standard heat exchange curve point by point according to the time axis, residual fitting calculation is performed in a sliding residual window mode, the numerical deviation and directional change of the characteristic factor are calculated in each window, a residual time series is obtained, and abnormal peak disturbance values are removed. The residual time series is subjected to dynamic clustering and segmentation analysis, residual interval segments with a deviation amplitude exceeding a warning interval threshold are extracted, the one-way nature and continuous time length characteristics of the deviation change trend in these interval segments are combined, it is determined that the local heat exchange parameter distortion caused by the air flow interference of the adjacent air cooler defrosting represents the air cooler, and the distortion is transmitted to the subsequent defrosting judgment logic.

[0012] Preferably, S3 specifically comprises the following steps: S301, the local heat exchange parameter distortion caused by the air flow interference of the adjacent air cooler defrosting and the real-time heat exchange data collected by the air cooler in the current period are subjected to data fusion processing according to the time dimension, the feature vectors are subjected to weighted operation to add up the values of the same dimension in the fusion process, and the fusion heat exchange characteristic data capable of representing the heat exchange state of the air cooler are formed. S302, according to the spatial position relationship between the air cooler and the disturbance source point, the air flow superposition strength and the interference coverage ratio, the interference weight is calculated, the interference weight is applied to the feature dimensions selected from the fusion heat exchange characteristic data through the interference index weight threshold, the weighted correction is performed on each feature dimension, and the modified fusion data subjected to interference compensation processing is output. S303, input the corrected fusion data into the defrosting credibility evaluation model, generate defrosting credibility evaluation parameters by using the trained credibility classification algorithm and heat exchange performance deviation identification algorithm in the evaluation model, wherein the defrosting credibility evaluation parameters include heat exchange performance degradation degree, interference offset factor and duration index; according to whether the heat exchange performance degradation degree continuously exceeds the normal operation interval threshold value, whether the interference offset factor exceeds the interference source influence threshold value, and whether the duration index exceeds the preset time threshold value, the cold air blower satisfying at least two index conditions is determined as the cold air blower that really needs to perform defrosting operation.

[0013] Preferably, S302 specifically comprises: Obtain the spatial position relationship parameters between the cold air blower and the disturbance source point in the current period, calculate the included angle between the air outlet direction of the cold air blower and the propagation direction of the disturbance source, determine the air flow superposition intensity parameter in combination with the spatial intersection proportion of the air flow superposition area and the wind speed information of the propagation path of the disturbance source point, and collect the interference area volume proportion parameter covered by the heat exchange surface of the cold air blower to form an input data set for constructing the interference weight; Construct a weighted function based on the spatial position included angle, the air flow superposition intensity parameter and the interference coverage proportion parameter, generate an interference weight value representing the interference degree, and apply the interference weight value to each feature dimension in the fusion heat exchange feature data, extract the feature dimensions whose interference sensitivity coefficients exceed the interference index weight threshold value to constitute a set of to-be-corrected feature dimensions; Perform a weighted correction operation on each feature dimension in the set of to-be-corrected feature dimensions, weight and superimpose the interference weight value and the original value of the corresponding feature dimension to generate corrected feature data for replacing the original value, and finally output the corrected fusion data after interference compensation processing as the input data source of the subsequent defrosting credibility evaluation model.

[0014] Preferably, S4 specifically comprises: Based on the joint state of the heat exchange performance degradation degree, the interference offset factor and the duration index in the defrosting credibility evaluation parameters, extract the numerical features of each heat exchange parameter of the cold air blower in the historical operation period, construct a historical operation feature vector including the heat exchange efficiency drop amplitude, the current load response amplitude and the wet heat fluctuation stability, and use it to describe the heat exchange behavior of the cold air blower under different load conditions; Collect the load fluctuation data of the current refrigeration area, extract the feature factors including the spatial temperature and humidity distribution, the goods flow frequency and the transient cold load change rate, input the historical operation feature vector and the current load fluctuation factor into the defrosting state determination engine, and generate the current defrosting determination label of the cold air blower through the trained determination model; According to the determination tag output result, the cold air machine is divided into three states of immediate defrosting required state, delayed observation required state and no defrosting required state, wherein the immediate defrosting required state satisfies that the heat exchange performance degradation degree is continuously higher than the normal operation interval threshold value, and the interference offset factor exceeds the interference source influence threshold value and the continuous time index exceeds the preset time threshold value; the delayed observation required state corresponds to that the heat exchange performance degradation degree is in the upper limit interval within the normal operation interval threshold value, and the interference offset factor or the continuous time index satisfies one of them; and the no defrosting required state corresponds to that the heat exchange performance degradation degree, the interference offset factor and the continuous time index all do not exceed the respective limited threshold value.

[0015] Preferably, S5 is specifically: According to the determination tag corresponding to the cold air machine, a differentiated defrosting control process is performed, the cold air machine with the determination tag of immediate defrosting required state is included in the immediate defrosting control sequence and a defrosting execution signal is sent, the cold air machine with the determination tag of delayed observation required state is marked as a key monitoring object and the heat exchange data acquisition frequency is improved, and the cold air machine with the determination tag of no defrosting required state maintains the original operation parameter and retains the observation task record; The data generated by each cold air machine in the current round, including heat exchange feature data, interference identification index and operation state parameter, is input into a heat exchange state space perception graph update engine in a time index and space position mapping manner, and the cold air machine node state label, adjacent airflow interference relationship edge weight and space heat load distribution result in the graph are synchronously updated; Based on the updated heat exchange state space perception graph, the heat exchange behavior correlation network of each cold air machine in the current operation cycle is reconstructed, the heat load distribution state, defrosting interference path and control strategy adaptability among the cold air machines are comprehensively analyzed, the defrosting control priority ranking result facing the next round is generated, and defrosting control optimization under dynamic regulation is realized.

[0016] In the above technical solution, the technical effects and advantages provided by the present application are as follows: 1、The present application breaks through the limitation of the traditional single-point perception mechanism by constructing a heat exchange state space perception graph, fusing cold air machine space position, wind direction and flow direction and defrosting exhaust state information, and for the first time realizes spatial recognition and propagation path prediction of airflow interference sources in the cold air machine operation environment. Combined with temperature, humidity, current, air volume and other multi-source parameters, a heat exchange characteristic curve is constructed, and a residual fitting and interference offset analysis algorithm is introduced, which effectively extracts the local heat exchange parameter distortion caused by external hot and humid airflow interference of the cold air machine, and provides high credibility of basic data basis for judging whether there is a real defrosting demand. This method significantly enhances the identification and filtering ability of the system to the false defrosting signal, and reduces the energy consumption fluctuation and operation instability risk caused by false defrosting.

[0017] 2、The application introduces a multi-dimensional decision mechanism of defrosting credible evaluation parameters, and constructs a defrosting judgment label system combining historical operation characteristics and cold storage area load fluctuation information, supports the hierarchical control of "need to defrost immediately", "need to delay observation" and "no need to defrost" on the cold air fan, and further improves the accuracy and rhythm matching of defrosting regulation. On this basis, the data closed loop mechanism can feed back the core perception data, correction data and judgment results in each round of decision process to the heat exchange state space-time perception map in real time, realize the dynamic learning and optimization of the heat load distribution state, defrosting interference path and control strategy adaptability in the cold air fan network, make the defrosting control strategy have the ability of continuous evolution, and significantly improve the energy efficiency, heat exchange performance and operation reliability of the distributed cold air fan system under complex working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0019] Figure 1 The flowchart of the cold air fan distributed dynamic defrosting control method of the present application. DETAILED DESCRIPTION

[0020] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations may, however, be implemented in many different forms and should not be construed as limited to the examples set forth herein; rather, these example implementations are provided so that this disclosure will be thorough and complete, and will fully convey the inventive concept to those skilled in the art.

[0021] The present application provides a cold air fan distributed dynamic defrosting control method as shown in Figure 1 The cold air fan distributed dynamic defrosting control method specifically comprises the following steps: S1, obtaining the spatial position of the cold air fan in the refrigeration environment, the wind direction and flow direction information, and the defrosting exhaust state of the adjacent cold air fan, constructing a heat exchange state space-time perception map based on the spatial coordinate relationship, for determining whether the cold air fan is disturbed by the defrosting air flow of the adjacent cold air fan; In this embodiment, S1 specifically comprises the following steps: S101, by deploying a space recognition unit in the refrigeration environment, obtaining the spatial position of each cold air fan, and combining the wind direction and wind speed detection components configured at the outlet of the cold air fan, obtaining the wind direction and flow direction information of the cold air fan under different operating states, and simultaneously identifying whether the adjacent cold air fan is in a defrosting exhaust state based on the cold air fan operation control logic and the heating start signal; The space recognition unit is deployed in a refrigeration environment to obtain the spatial position of each air cooler. Position reference marks can be preset in the refrigeration area, and an ultra-wideband positioning chip, a laser radar unit, or a high-precision inertial navigation device can be integrated into the air cooler body to determine the precise coordinate position of the air cooler in three-dimensional space through ranging feedback or coordinate fusion algorithms. The wind direction and speed detection components are arranged at the air outlet of the air cooler. A hot film anemometer and a differential pressure wind direction sensor are used to synchronously collect wind speed vector data and outlet direction angle information when the air cooler is in different operating states such as air supply, defrosting, and standby. The air cooler airflow behavior parameter library in each state is constructed. The operation control logic of the air cooler obtains the operation mode switching signal, such as air supply, defrosting start, and other control signals, by connecting the main controller. The heating start signal is used as a key trigger signal for the defrosting process. By collecting the on-off state of the electric heater or the heat pump reversing device, it is identified whether the current air cooler is in the defrosting exhaust process. The spatial position, wind direction flow information, and heating signal are logically mapped to determine in real time whether the air cooler is participating in the defrosting operation as a disturbance source of hot and humid airflow. This method can provide basic dynamic input data for subsequent construction of airflow propagation paths and identification of interference areas, avoiding misjudgment or lag caused by relying solely on static deployment information.

[0022] S102, establish a spatial coordinate relationship according to the spatial position of the air cooler and the wind direction flow information, take the air cooler in the defrosting exhaust state as a hot and humid disturbance source point, call a three-dimensional disturbance diffusion prediction model to calculate its influence radius and directional propagation path, and then construct a heat exchange state space perception map containing the airflow interaction relationship between air coolers; On the basis of obtaining the spatial position and wind direction flow information of the air cooler, the position data of each air cooler can be mapped into a three-dimensional space model by constructing a spatial coordinate grid. Combined with the wind direction angle and wind speed vector information, the main propagation direction and initial velocity field of the airflow are labeled in the coordinate model. For the air cooler currently in the defrosting exhaust state, it can be taken as a disturbance source point, and a three-dimensional disturbance diffusion prediction model based on CFD (Computational Fluid Dynamics) simulation optimization is called to simulate the disturbance in combination with its outlet velocity, wind direction angle, and environmental temperature and humidity background field. Based on the physical diffusion characteristics of the hot and humid air mass in the refrigeration environment, the model considers parameters such as turbulent diffusion, temperature difference convection, and airflow blocking structure to calculate the influence radius and propagation path of the disturbance source point within a certain time window. For example, when an air cooler discharges high-humidity and high-temperature airflow forward during defrosting, the model can predict that the airflow will form a disturbance field within a range of 2 meters in front and produce a certain intensity of residual propagation in the wind direction deflection area, eventually forming local temperature rise and humidity anomaly on the heat exchange surface. After superimposing the airflow propagation paths of multiple air coolers, a time-space perception map containing airflow overlap, interaction, and boundary reflection effects can be generated to further analyze the interference range and intensity distribution.

[0023] The spatial position of the air cooler is a core parameter for constructing a three-dimensional coordinate basis structure of the refrigeration environment, and can be obtained in real time by a spatial positioning device. The air direction and flow direction information is used to describe the air flow directionality characteristics of the air cooler in different operating states, and is an important basis for modeling the disturbance path. The air cooler in the defrost exhaust state is selected as the heat and moisture disturbance source point, because the high-temperature and high-humidity gas released by the air cooler in this state is most likely to cause an impact on the heat exchange perception of the adjacent equipment. The three-dimensional disturbance diffusion prediction model is an algorithm model trained based on the gas dynamics principle in the refrigeration environment, and has the ability to calculate the disturbance propagation boundary and path direction in real time. The influence radius represents the boundary range of the disturbance in space that produces obvious heat exchange parameter disturbance, and the directional propagation path represents the main flow direction and path line of the disturbance gas flow. The heat exchange state space-time perception map is a multi-dimensional visual data structure that integrates spatial coordinates, air direction and flow direction, disturbance path and intensity annotation, and can be used to accurately analyze the air flow interaction relationship between air coolers, and is a basis for subsequent identification of whether interference occurs.

[0024] S103, in the heat exchange state space-time perception map, identify the interference area by analyzing the spatial angle between the air cooler and the disturbance source point, the air flow overlap area and the disturbance intensity mapping relationship, and determine whether the air cooler is interfered by the defrost air flow of the adjacent air cooler according to the preset interference judgment rule.

[0025] The reason for identifying the interference area in the heat exchange state space-time perception map by analyzing the spatial angle between the air cooler and the disturbance source point, the air flow overlap area and the disturbance intensity mapping relationship, and determining whether the air cooler is interfered by the defrost air flow of the adjacent air cooler based on the preset interference judgment rule is that in the distributed dynamic defrosting control process of the air cooler, each air cooler needs to rely on local sensing parameters for defrosting determination. However, when the adjacent air cooler is releasing heat and humidity gas flow during defrosting, the gas flow will propagate along the space path and overlap with the air flow field of the air cooler that is not defrosting, causing abnormal fluctuations in local temperature, humidity and other perception data, thereby interfering with the accuracy of defrosting determination. By constructing the heat exchange state space-time perception map and considering the spatial angle, air flow overlap degree and disturbance intensity and other dimensional characteristics, the area affected by external disturbance can be accurately identified, and the source of abnormal perception data can be effectively distinguished. In combination with the preset interference judgment rule formed by multiple conditions, the reliability of identification can be further improved, and misjudgment as "frosting state" caused by short-term external disturbance can be avoided, thereby fundamentally improving the stability and energy efficiency of the distributed dynamic defrosting control of the air cooler. This strategy can help the system to abstract the disturbance source and influence path from the complex spatial interaction relationship, and provide a reliable and detailed basis for subsequent defrosting decision-making.

[0026] In this embodiment, S103 specifically comprises: The spatial angle between the air cooler and the disturbance source point is calculated, and the coincidence degree of the angle range and the main direction of the airflow is taken as a first interference index, and the potential disturbed direction is screened through the spatial path with an angle less than a set interference threshold; To determine whether the air cooler is in a potential disturbed direction, the air outlet direction vector of the air cooler and the main propagation direction vector of the airflow of the disturbance source point can be obtained first, and the spatial angle between the two sets of vectors is calculated based on a three-dimensional coordinate system. The spatial angle reflects the geometric relationship between the air cooler and the disturbance source point in the airflow propagation path. The smaller the angle, the more likely the two are collinear, and the more likely the direction is directly impacted by the airflow disturbance. To quantify this relationship, the coincidence degree of the angle range and the main propagation direction of the airflow of the disturbance source point can be taken as a first interference index, and a fixed angle threshold is set to screen the potential disturbed space path. For example, the angle threshold is set to 15 degrees. When the calculated angle is less than the threshold, it indicates that the air cooler is in the main channel of the airflow propagation of the disturbance source point, and has a high probability of being disturbed. The angle threshold is set through multiple rounds of simulation and regression analysis of measured data of the airflow behavior in the refrigeration environment, which can effectively distinguish the airflow disturbance significant path and the non-disturbance path. The angle calculation uses the three-dimensional vector angle formula, which involves vector projection operation of the air cooler outlet direction and the disturbance source point diffusion path. By screening all paths with an angle less than the threshold, the air coolers in the disturbance airflow influence direction in the space can be identified, thereby narrowing the scope of subsequent interference analysis and improving the judgment accuracy. This method can accurately depict the spatial directivity of airflow propagation, and is of key significance to determine whether the air cooler is truly disturbed by the defrosting airflow of the adjacent air cooler.

[0027] The spatial volume of the airflow overlap region is extracted, the volume ratio and the airflow velocity superposition strength are constructed into a disturbance intensity mapping relationship as a second interference index, and the airflow propagation trajectory and the intensity distribution overlap region are marked in the heat exchange state space perception map to identify the disturbed region that meets the interference index weight threshold; To further determine whether the cold air machine is actually disturbed by the defrosting airflow of the adjacent cold air machine, the spatial volume of the airflow overlapping area needs to be extracted in the three-dimensional space model. The spatial volume represents the intersection and overlapping part between the current cold air machine heat exchange influence range and the airflow propagation path of the disturbance source point. By establishing the heat exchange volume model of each cold air machine and the airflow propagation trajectory model of the disturbance source point, the overlapping space area of the two in the refrigeration environment can be extracted by using the three-dimensional volume Boolean intersection operation. To measure the degree of disturbance that the overlapping area may cause, the proportion of the overlapping area in the cold air machine heat exchange surface range can be taken as the volume proportion index, and the airflow velocity of the disturbance source point in the overlapping area is weighted and superimposed to construct the disturbance intensity mapping relationship, which is used to measure the actual influence of the disturbance intensity in the disturbed area. For example, when the overlapping area accounts for 40% of the volume of the cold air machine heat exchange area, and the disturbance airflow velocity maintains a high level in the area, the system will mark the area as a strong disturbance risk area. In the heat exchange state space perception map, these high-intensity disturbance areas are visually marked by different colors or symbols, providing intuitive judgment basis. The interference index weight threshold is a reference standard set through experimental statistics and energy efficiency change comparison analysis under a large number of airflow interference scenes. Usually, when the volume proportion exceeds 30% and the superimposed intensity exceeds the set value, the area is defined as an effective disturbed area. The threshold is used to filter low-intensity and occasional airflow overlapping areas, ensuring that the interference judgment has high confidence and actual response value. This method can avoid false judgments caused by marginal contact of airflow, and focuses on the key interference area that may actually affect the heat exchange performance, providing accurate basic data support for subsequent judgment of whether the cold air machine needs to be defrosted.

[0028] The angle threshold judgment, disturbance intensity coverage ratio judgment, and airflow interference duration judgment are used as preset interference judgment rules. The angle threshold judgment is based on whether the angle between the cold air machine outlet direction and the disturbance source direction is less than the set interference threshold. The disturbance intensity coverage ratio judgment is based on whether the proportion of the disturbed area in the cold air machine heat exchange surface exceeds the preset limited proportion. The airflow interference duration judgment is based on whether the disturbance influence duration exceeds the preset time window. The cold air machine that meets at least two of the above three judgments is determined as the cold air machine disturbed by the defrosting airflow of the adjacent cold air machine.

[0029] To achieve accurate judgment of whether the cold air fan is disturbed by the defrosting airflow of the adjacent cold air fan, three preset disturbance judgment rules, i.e., an included angle threshold judgment, a disturbance intensity coverage ratio judgment, and an airflow disturbance duration judgment, can be used for joint determination. The included angle threshold judgment is to analyze the spatial included angle between the outflow direction of the cold air fan and the airflow propagation direction of the disturbance source. If the included angle is less than a preset interference angle threshold, it indicates that the two airflows have a high degree of consistency in the height direction, which increases the possibility of overlapping interference. The angle threshold is generally determined based on the refrigerated space structure and fan layout characteristics, for example, taking 15 degrees as an empirical reference. The disturbance intensity coverage ratio judgment takes the proportion of the disturbed area in the cold air fan heat exchange surface range as the core index. The percentage of the overlapping volume to the entire heat exchange volume is usually automatically calculated by a three-dimensional modeling software. If the percentage exceeds a preset limited percentage, it is considered that the cold air fan will be substantially affected in the heat exchange process. The selection of the limited percentage is based on the sensitive threshold of the decrease of the cold air fan heat exchange efficiency, which is usually set to 30% or more. The airflow disturbance duration judgment is to continuously monitor the exhaust state of the disturbance source point and the duration of its continuous effect on the target cold air fan space area, and to accumulate statistics in the time window combined with the data acquisition period. When the duration exceeds a preset time window, for example, 5 minutes or 10 minutes, it indicates that the disturbance is not an occasional instantaneous phenomenon, but has a sustained impact feature. The three judgment indexes are independent of each other and complement each other. Among them, the included angle threshold reflects the spatial trend, the coverage ratio reflects the influence degree, and the duration reflects the action intensity. The system is set to satisfy at least two of them to be considered as effective disturbance, which can effectively avoid the misjudgment of a single index and improve the accuracy and reliability of disturbance identification. The joint judgment mechanism not only enhances the anti-interference ability, but also provides a more reliable data basis for subsequent dynamic defrosting strategies.

[0030] S2, collecting temperature, humidity, current, and air volume data of the cold air fan determined to be disturbed by the defrosting airflow of the adjacent cold air fan, constructing a heat exchange characteristic curve of the target cold air fan in the current period, and performing residual fitting calculation with a standard heat exchange curve in a historical non-defrosting state to determine the local heat exchange parameter distortion caused by the disturbance of the defrosting airflow of the adjacent cold air fan. In this embodiment, S2 specifically includes the following steps: S201, collecting temperature, humidity, current, and air volume data of the cold air fan determined to be disturbed by the defrosting airflow of the adjacent cold air fan in the current operation period through temperature sensors, humidity sensors, current sensors, and air volume detection components arranged on the internal heat exchanger surface and the inlet and outlet air paths of the cold air fan, and performing time alignment and data missing repair processing on different types of sensor data to ensure the continuity and consistency of the input data. In a refrigeration environment, to achieve high-precision perception of the target cold air fan heat exchange state, a thermocouple array temperature sensor can be uniformly arranged on the surface of the heat exchanger fins of the cold air fan, high-speed response humidity sensors and air volume detection components can be installed on both sides of the air inlet and air outlet, and a current collection module can be installed on the power input end to obtain the temperature, humidity, air volume, current and other operating parameters of the target cold air fan at different times. The collection process is carried out at a unified sampling frequency, and a time stamp synchronization mechanism is accessed by a data interface to align the time axis of different types of sensor data. To solve the problem of data discontinuity caused by network fluctuations or temporary sensor malfunction, a linear regression interpolation algorithm based on a sliding window is used to fill in missing values, and adjacent point slope constraints are used to remove abnormal points to ensure the time continuity and numerical stability of the data. For example, if the air volume data is missing for 3 seconds in a sampling, a linear fitting interpolation can be performed using the sliding window values of the previous and next 5 seconds, and the reasonableness of the fitted value can be judged according to the air volume change trend to avoid error propagation into the subsequent modeling process.

[0031] The deployment position of the temperature sensor is located at the center of the heat exchanger fins and the transition area on both sides, which can reflect the temperature gradient distribution of the cold air fan in the heat load response process; the humidity sensor selects a fast-response thin-film capacitor structure and is installed inside the cold air fan outlet channel to effectively sense the humidity characteristics of the cold air output; the air volume detection component is a impeller speed type or differential pressure type structure, arranged between the air inlet and air outlet, for synchronous detection of air volume fluctuation state; the current sensor uses a Hall-type closed-loop measurement scheme to capture the current response characteristics of the cold air fan under load changes. The data generated by the above four types of sensors is recorded at a millisecond level and enters a unified data buffer channel, and high-precision time stamps are used to ensure the consistency of the data timing between channels. When repairing data, a two-way moving average and central difference method is used to identify discontinuous segments, and a conditional regression fitting interpolation is used to retain the true operating characteristics while minimizing data distortion, providing a high-quality, low-noise input basis for subsequent heat exchange behavior modeling.

[0032] S202, after the processed temperature, humidity, current and air volume data are integrated in time sequence, they are input into the heat exchange behavior modeling engine, the heat exchange characteristic factors are refined through multivariate feature extraction algorithm, and the heat exchange characteristic curve of the target cold air fan in the current period is constructed in the same time dimension, so that the heat load change and the running state linkage relationship are fully reflected; S203, the heat exchange characteristic curve of the target cold air fan in the current period is fitted with the standard heat exchange curve in the non-defrosting state selected from the historical data, the offset amplitude, change trend and duration of the residual error interval are analyzed, the local heat exchange parameter distortion caused by the defrosting airflow interference of the adjacent cold air fan is extracted, and is used as the input of the subsequent defrosting judgment logic.

[0033] The fundamental purpose of this is to distinguish the heat transfer performance parameter fluctuation caused by the interference of the defrosting air flow of the adjacent cold air machine from the real defrosting demand caused by the accumulation of frost on the machine in the refrigeration environment where the cold air machine is located, so as to improve the accuracy of defrosting control decision. In the background of complex air flow interference, the heat transfer parameters of the cold air machine may appear short-term temperature rise, current change or abnormal air volume. The traditional distributed defrosting control depends on the single-node self-sensing mechanism and is difficult to identify whether these disturbances are caused by external interference or actual frost blockage. By fitting the residual of the current cold air machine heat transfer characteristic curve with the historical standard curve, the difference between the two in actual operation can be quantified. The shift range, trend change and duration of the residual can reveal whether these differences are systematic and persistent, and whether they meet the typical characteristics of air flow interference. This analysis not only provides a clear decision basis for whether to trigger the defrosting operation subsequently, but also builds a reliable judgment chain connecting the sensing data and the control execution, avoiding the problem of cascading false triggering of defrosting link caused by misjudgment, and ensuring the stability of temperature control and energy efficiency in refrigeration environment.

[0034] In this embodiment, S202 is specifically: Align the temperature, humidity, current and air volume data in the same time axis, extract the continuous time period data segment at equal intervals by using the sliding window mechanism, complete and correct the missing or abnormal fluctuation data by using the bidirectional interpolation method, and ensure the synchronization and data integrity of the multi-source data in the time sequence at each time point. To realize the sequence alignment processing of temperature, humidity, current and air volume data on the unified time axis, first, the data of the four types of sensors should be standardized and synchronized based on a unified timestamp system, all sensor data should be remapped to a millisecond-level time axis, and a sampling period reference point should be set as the alignment reference. Using the sliding window mechanism, continuous and fixed-length data segments can be divided at a set time step, for example, a window of 30 seconds, and each time sliding 10 seconds, so as to generate multiple overlapping data segments of time periods, which is convenient for local trend judgment and anomaly identification. Within the window, if the air volume or current value at a certain time point is missing or a mutation jump point appears, the bidirectional interpolation method can be used to repair the value at this point. Bidirectional interpolation refers to selecting several effective data points before and after the target point, calculating the predicted value from the forward and backward directions respectively through linear interpolation or cubic spline interpolation, and taking the weighted average result to fill in the missing data, so as to improve the smoothness and robustness of interpolation. For example, during the operation of a certain air cooler, the air volume is missing at t=80s due to signal fluctuation, then effective data points can be extracted between t=75s to t=79s and t=81s to t=85s, forward and backward interpolation models are constructed, and the weighted average value of the two is taken as the completion value at t=80s. When correcting abnormal fluctuation values, the mean and standard deviation information within the sliding window are combined to determine whether it is an abnormal peak value, and then the adjacent point trend is corrected based on the adjacent point trend to ensure the synchronization, continuity and integrity of multi-source data at the same time point, and to avoid noise interference on subsequent heat transfer behavior modeling. The sliding window mechanism can dynamically adapt to the local variation characteristics of the data, and the bidirectional interpolation method can take into account the context trend information, and the combination of the two can significantly improve the time series fusion quality of multi-source sensor data.

[0035] The time series aligned data is input into the heat transfer behavior modeling engine, the heat transfer characteristic factors are extracted from the temperature, humidity, current and air volume data through the joint principal component analysis algorithm and clustering decomposition algorithm, and the redundant and low correlation factors are removed to form a characteristic vector set for describing the heat transfer behavior state of the target air cooler. To extract the heat exchange characteristic factors that can accurately characterize the heat exchange behavior of the air cooler from temperature, humidity, current and air volume data, the data aligned by time series first needs to be input into the heat exchange behavior modeling engine, and the joint principal component analysis algorithm and clustering decomposition algorithm are used for feature dimension reduction and structure recognition processing. Principal component analysis algorithm can transform multiple high-dimensional variables into a small number of uncorrelated principal components while maximizing information, thereby highlighting the main trend of change in the heat exchange process. For example, when inputting multiple time series indicators including inlet air temperature, outlet air humidity, fan power, current load, air speed, etc., the principal component analysis algorithm can identify the main cause of the greatest impact on heat load change, such as the "high temperature, high humidity, high load" working condition mode. Clustering decomposition algorithm further classifies the distribution of these principal components in different time periods, classifies time periods with similar operating states into the same class, and extracts the most representative statistical characteristics (such as mean change, fluctuation amplitude, periodic waveform) in this class as heat exchange characteristic factors to represent the heat exchange capacity performance of the air cooler under the current working condition. For example, identifying clustering samples with synchronous current amplitude rise accompanied by stable air speed and sudden humidity rise in multiple window periods can extract the heat exchange characteristic factor of "evaporation side heat transfer efficiency decline". By combining these two algorithms, not only can redundant information and low correlation variables be eliminated, but also key factors reflecting the change of the air cooler's heat exchange capacity can be retained, ultimately forming a feature vector set containing multiple high-weight heat exchange characteristic factors, providing a strong model basis for subsequent heat exchange curve construction and anomaly identification. This modeling method not only improves the accuracy of heat exchange state representation, but also enhances the response comparison ability of air coolers in dynamic environments.

[0036] Map the feature vector set by time series to construct the heat exchange characteristic curve of the target air cooler in the current period, where the vertical axis is the extracted heat exchange characteristic factor group, and the horizontal axis is the time axis consistent with the sensor data. The response law of the target air cooler's current heat exchange capacity and operating state under dynamic heat load changes is reflected through the heat exchange characteristic curve.

[0037] In order to construct the heat exchange characteristic curve of the target air cooler in the current period, the feature vector set composed of the heat exchange characteristic factors extracted by the principal component analysis algorithm and the clustering decomposition algorithm needs to be sequentially mapped according to the corresponding time stamp. In the mapping process, each feature vector represents the heat exchange behavior state of the air cooler at a certain time point. By arranging the feature vectors of multiple consecutive time points in sequence, a heat exchange characteristic curve can be formed in a two-dimensional coordinate system. The horizontal axis of the coordinate system is the standard time axis consistent with the time stamp of the sensor data, and the vertical axis is the multi-dimensional feature projection value composed of the heat exchange characteristic factor group corresponding to each time point. Through this construction method, the response rhythm and state change pattern of the target air cooler in the current operating period when facing environmental heat load changes and adjacent air cooler airflow interference can be fully expressed. For example, in the case of a sudden increase in heat load in the refrigeration area, if the air cooler can quickly respond by increasing the air volume and reducing the evaporation temperature, its heat exchange characteristic curve will show a gradual stabilization after the rapid shift of the characteristic factors in a certain time period. If the reaction is delayed due to external disturbance, the curve will show delayed shift and increased fluctuations. By constructing this curve, not only can the instantaneous fluctuation trend of the heat exchange performance be captured intuitively, but also a clear structured input can be provided for subsequent comparison with the standard heat exchange behavior, so that the causal relationship between the air cooler operating state and the interference factors can be clearly analyzed in the time dimension. This method focuses on the time sequence and structure of the data in the construction process, ensuring that the feature curve formed has high-resolution behavior description capability.

[0038] In this embodiment, S203 is specifically: In the historical operating data, the time period closest to the current refrigeration environment load state and operating condition characteristics is selected, and the standard heat exchange curve of the target air cooler in this time period is extracted to ensure that the standard curve does not contain data interference during defrosting, serving as a reference for the current heat exchange characteristic curve. In order to extract the time period closest to the current cold storage environment load state and operating condition characteristics from historical operation data, it is necessary to first establish an operating condition characteristic index system for similarity matching. The index system can include the thermal load level of the target cold air fan served area in the cold storage space, the environment temperature and humidity, the operating frequency of the cold air fan, the air volume output curve, the current change trend and other parameters. Using a multi-dimensional operating condition similarity matching algorithm, such as Mahalanobis distance combined with a time window weighting strategy, a plurality of time periods closest to the current operating condition can be selected from the historical data as candidates. Subsequently, each candidate time period is filtered for defrosting state, and by identifying the heating current peak value, control signal marker or defrosting log record, the segment containing defrosting interference is removed, and finally a historical time period with the highest operating condition similarity and without defrosting behavior interference is determined, and the heat exchange characteristic curve of the target cold air fan is extracted from the time period as the standard heat exchange curve. The standard curve is used to construct the reference for the current heat exchange characteristic curve, ensuring that the comparison is the ideal operating state during residual fitting, thereby improving the accuracy of subsequent error analysis and the reliability of interference identification. Through this process, not only is the reference curve guaranteed in terms of thermal load consistency, operating stability and defrosting interference, but the ability to distinguish between abnormal heat exchange behavior and real frosting phenomenon is also effectively enhanced.

[0039] The heat exchange characteristic curve of the target cold air fan in the current period is compared with the standard heat exchange curve point by point along the time axis, and residual fitting calculation is performed in a sliding residual window mode. The numerical deviation and directional change of the characteristic factor are calculated in each window to obtain the residual time series and remove abnormal peak disturbance values. In order to effectively compare the heat transfer characteristic curve of the target air cooler with the standard heat transfer curve in the current time period, it is necessary to ensure that the two curves are completely aligned in the time dimension, i.e. each time node corresponds to a unique characteristic factor vector. After alignment, a sliding residual window with a fixed width is set to gradually slide through the time axis, each window covering a number of consecutive time points. For each time point, the difference between its corresponding characteristic factor group and the same time point in the standard heat transfer curve is calculated, including the numerical deviation and the trend difference of the change direction. Based on the mean and variance of the residuals of all characteristic factors in the window, a continuous residual time series is formed. To eliminate false positives caused by occasional fluctuations, the median filter and mutation detection algorithm, such as the IQR outlier identification strategy, can be used to clean up the prominent spikes in the residual sequence, thereby preserving the continuous error passages that truly reflect the abnormal trend of heat transfer. The entire residual fitting process not only captures the local change trend and overall deviation degree, but also separates short-term fluctuations and long-term deviations through the sliding window, improving the sensitivity and discrimination of the air cooler's running state abnormality. This provides a high-resolution data basis for subsequent error trend identification and interference distortion judgment. This method is especially suitable for nonlinear fluctuation frequent refrigeration environment heat transfer process, and can effectively filter out short-term irregular fluctuations and highlight the characteristics of persistent deviation.

[0040] The residual time series is dynamically clustered and segmented, and the residual interval segments with a deviation amplitude exceeding the warning interval threshold are extracted. Combined with the one-wayness and duration characteristics of the deviation trend in these interval segments, it is determined that they represent the local heat transfer parameter distortion caused by the interference of the defrosting airflow of the adjacent air cooler, and this distortion is passed to the subsequent defrosting judgment logic.

[0041] In order to identify the key segment of the cold air machine heat exchange behavior from the residual time series, it is necessary to introduce a dynamic clustering segmentation analysis method to divide the entire residual series into multiple segments with similar characteristic change patterns. Dynamic clustering segmentation analysis is usually realized by sliding distance measurement algorithm and trend direction similarity index, which can adaptively divide the residual curve into several trend-uniform interval segments. The data inside each segment has high consistency in numerical offset direction and change speed, which helps to exclude non-continuous disturbance caused by local noise. At the same time, combined with the set warning interval threshold, the average offset amplitude of each residual segment is compared, and if it exceeds the threshold, it is considered as a potential interference area. The threshold is usually set by the historical normal state data distribution range, which ensures that the abnormal identification has statistical significance. Further, the segments that meet the threshold requirement also need to verify whether the offset trend is unidirectional and continuous, and whether the continuous length meets the set time length threshold, and finally filter out the distortion interval that meets all the characteristics. These intervals are identified as local heat exchange parameter distortion caused by the interference of the defrosting airflow of the adjacent cold air machine, which can be used as the key variable input to the subsequent defrosting logic judgment process to effectively control the false trigger risk. This method improves the accuracy of abnormal identification and the rigor of the judgment logic by modeling the behavior trend of residual data and performing statistical significance test.

[0042] S3, fuse the local heat exchange parameter distortion performance with the real-time heat exchange data of the cold air machine, correct the fusion result by interference weight, input the corrected fusion data into the defrosting credibility evaluation model, generate defrosting credibility evaluation parameters, and judge whether the cold air machine really needs to perform defrosting operation according to the local heat exchange parameter distortion performance caused by the interference of the defrosting airflow of the adjacent cold air machine; In this embodiment, S3 specifically includes the following steps: S301, fuse the local heat exchange parameter distortion performance caused by the interference of the defrosting airflow of the adjacent cold air machine with the real-time heat exchange data collected by the cold air machine in the current period according to the time dimension, and perform numerical superposition on the same dimension features by using feature vector weighted operation during the fusion process to form fusion heat exchange feature data that can represent the heat exchange state of the cold air machine; In order to accurately characterize the current heat transfer behavior of the cold air machine, the local heat transfer parameter distortion caused by the interference of the defrosting air flow of the adjacent cold air machine needs to be fused with the real-time heat transfer data collected by the cold air machine in the current period. The fusion process first aligns the two types of data in a unified time dimension, and ensures the synchronization of each feature data at each time by constructing time series with the same time granularity. On this basis, the heat transfer features of the same dimension are subjected to numerical superposition processing, and the real-time heat transfer feature values and the distorted feature values are combined and calculated according to the preset fusion weight by using the feature vector weighting algorithm to form fusion heat transfer feature data with time continuity. For example, for the feature dimension of the cold air machine outlet air volume, if the real-time monitoring value is 2.8 m³ / min, the distortion correction value is -0.3 m³ / min, and the fusion weight is set to 0.7 and 0.3, then the fusion value is the weighted sum of 2.59 m³ / min. Through this fusion method, the original features of the actual operating conditions of the cold air machine can be effectively retained, while the parameter drift information caused by the air flow interference is introduced, so that the fusion data can fully reflect the heat transfer state of the cold air machine under disturbance.

[0043] The local heat transfer parameter distortion caused by the interference of the defrosting air flow of the adjacent cold air machine refers to the phenomena of heat transfer efficiency decline, abnormal fluctuation of air speed or deviation of temperature and humidity parameters under specific interference. These data are usually derived from the abnormal feature recognition results in the residual fitting analysis process. The real-time heat transfer data collected by the cold air machine in the current period include temperature, humidity, current, air volume and other multi-dimensional sensor information, and the sampling frequency and time stamp are uniformly set to minute level. Data fusion processing refers to the fusion calculation of real-time heat transfer data and distortion performance data at the same time point at the feature level based on the time stamp. Feature vector weighting operation is an algorithm that linearly combines feature values of the same dimension by using weighting coefficients, which aims to strengthen the expression ability of interference in the fusion data and maintain the smoothness of the overall numerical change. Fusion heat transfer feature data refers to a set of time series data containing multiple heat transfer feature dimensions, which is used as input for subsequent interference correction and credibility evaluation model to ensure the uniformity and integrity of data in model training and inference process.

[0044] S302, according to the spatial position relationship between the cold air machine and the disturbance source point, the air flow superposition strength and the interference coverage ratio, the interference weight is calculated, the interference weight is applied to the feature dimensions selected by the interference index weight threshold in the fusion heat transfer feature data, the weighted correction is performed on each feature dimension, and the corrected fusion data after interference compensation processing is output; S303, input the corrected fusion data into a defrosting credibility evaluation model, and generate defrosting credibility evaluation parameters by using a credibility classification algorithm and a heat exchange performance deviation identification algorithm trained in the evaluation model, wherein the defrosting credibility evaluation parameters include a heat exchange performance degradation degree, an interference offset factor, and a duration index; and determine the cold air blower that meets at least two index conditions as the cold air blower that truly needs to perform the defrosting operation according to whether the heat exchange performance degradation degree continuously exceeds a normal operation interval threshold value, whether the interference offset factor exceeds an interference source influence threshold value, and whether the duration index exceeds a preset time threshold value.

[0045] In order to determine whether the cold air blower truly needs to perform the defrosting operation, the corrected fusion data obtained through the previous processing needs to be input into a defrosting credibility evaluation model. The model is constructed in the training stage and includes a credibility classification algorithm and a heat exchange performance deviation identification algorithm, and can comprehensively evaluate the difference characteristics between the current heat exchange state and the historical heat exchange performance of the cold air blower. In the specific implementation process, first, the feature vectors representing the heat exchange efficiency, temperature response, current change, and air volume stability in the corrected fusion data are input into the model, and the model internally extracts three types of evaluation parameters, i.e., the heat exchange performance degradation degree, the interference offset factor, and the duration index, through clustering boundary identification, probability distribution fitting, and error deviation regression algorithms. For example, the heat exchange performance degradation degree can be scored by comparing the deviation amplitude of the heat exchange curve in the standard state and the current state curve on the key features, the interference offset factor is calculated based on the weighted fitting result of the feature curve change rate in the disturbance influence area, and the duration index is the time window length of the abnormal heat exchange state continuously exceeding the reference curve, and finally a complete set of defrosting credibility evaluation parameters is formed.

[0046] When determining whether to perform the defrosting operation, joint determination based on the above three parameters and the set threshold value standard is needed. The normal operation interval threshold value is used to limit the maximum allowable deviation range of the heat exchange performance degradation degree index, which is usually determined according to the feature fluctuation range statistics collected under the condition of stable operation of the cold air blower in multiple cycles; the interference source influence threshold value defines the maximum offset amplitude that can be accepted by the disturbed feature factor, which is set by comparing the fitting residual range during the defrosting period and the non-defrosting period; and the preset time threshold value is used to determine whether the duration of the heat exchange abnormal state exceeds the normal window of the natural fluctuation of the heat exchange behavior, which is generally determined in combination with the refrigeration working condition and the unit response period experience. When performing the determination, the system calculates whether each of the three indexes meets the determination condition, and when at least two of the indexes exceed the corresponding threshold value standard, it is determined that the current cold air blower has substantially decreased in heat exchange performance due to interference, and needs to perform the defrosting operation. Through this joint determination strategy, false judgments caused by local interference or short-term fluctuations can be effectively avoided, thereby ensuring the accuracy of the defrosting operation and the efficiency of the refrigeration system operation.

[0047] In this embodiment, S302 specifically comprises: Obtain the spatial positional relationship parameter between the air cooler and the disturbance source point in the current period, calculate the included angle between the air outlet direction of the air cooler and the propagation direction of the disturbance source, combine the spatial intersection proportion of the airflow superposition area and the wind speed information of the propagation path of the disturbance source point, determine the airflow superposition intensity parameter, and collect the interference area volume proportion parameter covered by the heat exchange surface of the air cooler to form an input data set for constructing the interference weight; To accurately evaluate the degree of interference of the air cooler by the defrost airflow of the adjacent air cooler, the spatial positional relationship parameter between the air cooler and the disturbance source point needs to be obtained and quantitatively analyzed. First, the coordinate positions of the air cooler and the disturbance source point in the refrigeration space are obtained by a three-dimensional positioning sensor, and the included angle between the air outlet direction of the air cooler and the propagation direction of the disturbance source is calculated based on a geometric model. The smaller the included angle, the higher the risk of airflow direction overlap. Then, based on the airflow simulation results or the wind speed sensor array, the spatial range of the airflow superposition area is identified, and the intersection volume of the area and the working airflow path of the air cooler is calculated to obtain the spatial intersection proportion of the airflow superposition. At the same time, combined with the wind speed data on the propagation path of the disturbance source point, the airflow superposition intensity parameter is constructed to reflect the influence degree of interference energy. In addition, the spatial model of the heat exchange surface of the air cooler needs to be analyzed to identify the area covered by the disturbed airflow and calculate its volume proportion in the total heat exchange surface area to form the interference area volume proportion parameter. By modeling the three types of physical quantities of spatial included angle, superposition intensity, and interference coverage proportion and constructing the input data set, the basis for subsequent interference weight calculation is provided, thereby realizing the quantitative evaluation of the interference intensity of the air cooler. For example, when the included angle between the air cooler and the disturbance source point is 15 degrees, the airflow overlap volume proportion is 40%, the maximum value of the wind speed superposition is 3.2 m / s, and 25% of the area of the heat exchange surface of the air cooler is in the overlap range, the input data set constructed by the three parameters can fully reflect the spatial influence characteristics of the airflow interference on the operation state of the air cooler, providing an accurate basis for subsequent fusion data correction.

[0048] Based on the spatial positional included angle, the airflow superposition intensity parameter, and the interference coverage proportion parameter, a weighted function is constructed to generate an interference weight value representing the degree of interference. The interference weight value is applied to each feature dimension in the fusion heat exchange feature data, and the feature dimensions with interference sensitivity coefficients exceeding the interference index weight threshold are extracted to form a set of feature dimensions to be corrected. To achieve accurate correction of the cold air machine fusion heat transfer characteristic data, a weighted function is constructed based on the spatial position angle, airflow superposition intensity parameter and interference coverage ratio parameter, and the interference weight value for quantifying the interference degree is generated. The spatial position angle is used to measure the geometric alignment degree of the cold air machine outflow direction and the propagation direction of the disturbance source. The smaller the angle, the more direct the potential interference path. The airflow superposition intensity parameter reflects the energy overlap degree of the overlapping area between the disturbance source airflow and the normal operation airflow in space, which can be realized by the wind speed distribution superposition model. The interference coverage ratio parameter represents the proportion of the area affected by the disturbed airflow in the heat transfer surface of the cold air machine, which is usually calculated by spatial distribution mapping and volume modeling. The three types of parameters are input into the weighted function, and the linear or nonlinear weighted model is used to generate the interference weight value, which is used to measure the possibility of disturbance of each feature dimension in the fusion heat transfer characteristic data. Then, the interference sensitivity of each dimension in the fusion heat transfer characteristic data is analyzed, the response degree of the airflow disturbance under the historical operation state is calculated, and it is compared with the pre-set interference index weight threshold value, which can be set by historical data training or expert rules, to filter out the most sensitive feature dimension to the interference change. For example, if the temperature gradient change rate dimension deviates significantly under the interference condition and exceeds the interference index weight threshold value, it is identified as the target dimension to be corrected, and finally forms the set of feature dimensions to be corrected, which is used for subsequent weighted processing. Through this method, all features can be avoided to be corrected without distinction, and the pertinence and efficiency of interference correction are improved.

[0049] The weighted correction operation is performed on each feature dimension in the set of feature dimensions to be corrected, the interference weight value and the original value of the corresponding feature dimension are weighted and superimposed to generate the corrected feature data for replacing the original value, and finally the interference compensation processed corrected fusion data is output as the input data source of the subsequent defrosting credibility evaluation model.

[0050] In processing each feature dimension in the set of feature dimensions to be corrected, a weighted correction operation needs to be performed to achieve targeted numerical adjustment of the features affected by the interference of the defrosting air flow of the adjacent cold air fan in the cold air fan heat exchange data. The implementation of the weighted correction is to superimpose the original numerical value of each feature dimension with the corresponding interference weight value, calculate a new corrected numerical value, and use it to replace the original data to ensure that the fused data more truly reflects the actual heat exchange state of the cold air fan under the interference condition. For example, when the temperature change rate is identified as a sensitive feature dimension, its original curve shows abnormal fluctuations during the interference period, and by weighting with the interference weight value, the deviation trend caused by interference is suppressed, and a more stable corrected temperature change rate is obtained. The interference weight value is derived from the weighting function constructed in the previous analysis, and its numerical value is determined according to the spatial angle between the cold air fan and the disturbance source, the air flow overlap intensity and the interference coverage ratio of the heat exchange surface. In the correction process, to avoid excessive compensation or data distortion, a proportional limit or error suppression mechanism can be used to control the weighted ratio within a certain range. All the corrected feature data are finally integrated into a corrected fused data set and organized in time series for input into the defrosting credible evaluation model to provide interference-corrected decision basis for judging whether the cold air fan truly needs to perform defrosting operation. This processing strategy significantly improves the recognition ability of the model between real heat exchange performance degradation and disturbance misjudgment, and helps to reduce unnecessary defrosting behavior and improve the operating efficiency of the refrigeration system.

[0051] S4, based on the defrosting credible evaluation parameters, the historical running features of the cold air fan and the load wave of the refrigeration area, a judgment label is generated to divide the cold air fan into three states of immediate defrosting, delayed observation and no defrosting; In this embodiment, S4 is specifically: Based on the joint state of the heat exchange performance degradation degree, the interference offset factor and the duration index in the defrosting credible evaluation parameters, the numerical features of each heat exchange parameter of the cold air fan in the historical running period are extracted, and a historical running feature vector containing the heat exchange efficiency decline amplitude, the current load response amplitude and the humidity and heat fluctuation stability is constructed to describe the heat exchange behavior of the cold air fan under different load conditions; In the process of executing the cold air blower defrosting state intelligent judgment, first of all, a feature expression reflecting the historical trend of the heat exchange behavior of the cold air blower needs to be constructed. To achieve this goal, the defrosting credible evaluation parameters can be taken as a reference, the historical operation data of the target cold air blower in different time periods is traced back, and the numerical evolution process of the key heat exchange parameters in the sensor data such as temperature, humidity, current and air volume is extracted through window analysis. The heat exchange efficiency decline amplitude can be quantified by the change gradient of the difference between the outlet air temperature and the inlet air temperature of the cold air blower in the continuous time period, the current load response amplitude can be calculated by the average value of the current and the load fluctuation synchronism index, and the wet heat fluctuation stability can be evaluated according to the standard deviation change of the inlet and outlet humidity difference in different load stages. The above heat exchange efficiency, current fluctuation, wet heat stability and other parameters are vectorized and combined, and are uniformly mapped to the time dimension, so as to form a historical operation feature vector for describing the stability of the thermodynamic performance of the cold air blower. The feature vector can be matched with the real-time state in the subsequent model, providing basic data support for the identification of abnormal trends of the cold air blower heat exchange.

[0052] The heat exchange performance degradation degree is used to measure the decline amplitude of the heat exchange capacity of the cold air blower under a certain load compared with the historical reference value, which is calculated by comparing the current outlet air temperature difference with the average heat exchange curve under the historical normal state; the interference offset factor is used to measure the deviation between the current heat exchange characteristic curve and the curve after the interference characteristic is superimposed, which can be represented by the Euclidean distance or Mahalanobis distance in the feature dimension; the duration index is used to capture whether the degradation state or the offset state is stable, which can be measured by the time length of the deviation in the sliding time window being greater than the threshold value. The current load response amplitude is used to represent the current adjustment sensitivity of the cold air blower under the change of thermal load, which can be calculated by the correlation coefficient between the current curve and the temperature change rate. The wet heat fluctuation stability can be described by the mean square deviation of the humidity change per unit time in the time series. The historical operation feature vector composed of these parameters is input into the subsequent classification model after standardization, providing statistical and physical response basis for judging whether the cold air blower meets the defrosting state.

[0053] The load fluctuation data of the current refrigeration area is collected, the feature factors including spatial temperature and humidity distribution, goods circulation frequency and transient cold load change rate are extracted, the historical operation feature vector and the current load fluctuation factor are jointly input into the defrosting state judgment engine, and the defrosting judgment label of the current cold air blower is generated through the trained judgment model; In order to realize the intelligent judgment of whether the air cooler needs to perform defrosting operation, first of all, the current load fluctuation data in the refrigeration area need to be collected, which can be synchronously obtained through the multi-point temperature and humidity sensor array, door access sensor and flow trajectory recording device arranged in the refrigeration space. Among them, the space temperature and humidity distribution reflects the balance of the cold and hot distribution of the environment, the flow frequency of goods can be quantified through the door opening and closing record and the reading frequency of the goods handling label, and the transient cold load change rate can be calculated by the temperature and humidity change gradient in unit time. After the fusion of these multi-source data on the time axis, the load fluctuation factor vector is formed, and the historical running characteristic vector of the air cooler is jointly input. In the data processing process, a defrosting state judgment engine with feature dimension reduction and pattern recognition capability is introduced, the mapping relationship between the current load behavior and the heat exchange trend of the air cooler is established through feature cross combination and time sequence dependent modeling, so as to complete the classification of the running state of the air cooler.

[0054] The defrosting state judgment engine is a data reasoning system based on machine learning method, and its core component is a trained judgment model. The model adopts a supervised learning method to complete training based on a large number of historical running samples, and the model structure can select an integrated decision tree, a support vector machine or a time sequence neural network with strong generalization ability. In the training process, the historical characteristic vectors of the air cooler under various load fluctuation scenarios are taken as inputs, and the defrosting decision results labeled by experts are taken as output labels to complete the fitting of model parameters. In actual application, after the current joint characteristic data of the air cooler is input into the model, the defrosting judgment label will be output according to the law recognized by the historical pattern. The label contains three states, respectively corresponding to three levels of immediate defrosting, delayed observation and no defrosting, to guide the implementation of the defrosting control strategy of the system and improve the accuracy and energy efficiency of the defrosting operation.

[0055] According to the output result of the judgment label, the air cooler is divided into three states of immediate defrosting state, delayed observation state and no defrosting state, wherein the immediate defrosting state satisfies that the heat exchange performance degradation degree continuously exceeds the threshold value of the normal running interval, and the interference offset factor exceeds the influence threshold value of the interference source and the continuous time index exceeds the preset time threshold value; the delayed observation state corresponds to that the heat exchange performance degradation degree is within the upper limit interval of the normal running interval, and the interference offset factor or the continuous time index satisfies one of them; the no defrosting state corresponds to that the heat exchange performance degradation degree, the interference offset factor and the continuous time index all do not exceed the respective limited threshold value.

[0056] When intelligently classifying the defrosting status of evaporative air coolers, the system can categorize them into three states: requiring immediate defrosting, requiring delayed observation, and not requiring defrosting, based on the judgment labels output by the defrosting status judgment model. The judgment model comprehensively assesses the system based on three dimensions of evaluation parameters: the degree of heat exchange performance degradation, the interference offset factor, and the duration of the interference. If the degree of heat exchange performance degradation exceeds the upper limit of the normal operating range for multiple consecutive time periods, and the interference offset factor exceeds a preset interference threshold, and the duration of this interference exceeds a predefined time threshold, the evaporative air cooler is classified as requiring immediate defrosting. This indicates that the heat exchange capacity is severely limited, and immediate defrosting intervention is necessary. If only the interference offset factor or the duration meets the specified conditions, and the degree of heat exchange performance degradation is still near the upper limit but not exceeding it, the system will classify it as requiring delayed observation, indicating a potential latent defrosting need but no immediate action required. If none of the three indicators exceed their respective set threshold ranges, the system classifies it as not requiring defrosting to avoid unnecessary energy waste.

[0057] The degree of heat exchange performance degradation is determined by comparing the residual distribution between the current heat exchange characteristic curve of the evaporative cooler and the standard curve formed by historical normal heat exchange behavior. A higher index value indicates a more significant decrease in actual heat exchange efficiency. The disturbance offset factor measures the degree of influence of airflow disturbance on the heat exchange parameters of the evaporative cooler, typically calculated by quantifying the impact of superimposed airflow on key characteristic dimensions. The duration index is an assessment value recording the duration of the disturbance effect, statistically analyzed over the entire evaporative cooler operating cycle using a sliding time window. These three indices correspond to three risk sources: performance degradation, disturbance impact, and impact persistence. Judgment is based on a logical combination of whether each exceeds its respective threshold. This three-state classification helps the system distinguish defrosting needs of varying urgency, thereby improving the accuracy and response efficiency of defrosting decisions while ensuring system energy efficiency.

[0058] S5. Based on the judgment tag corresponding to the evaporative air cooler, a differentiated defrosting control process is executed, and the data generated by each evaporative air cooler in the current cycle is used to update the spatiotemporal perception map of the heat exchange status, so as to realize the optimization of defrosting control under dynamic regulation.

[0059] In this embodiment, S5 specifically refers to: Based on the judgment tag corresponding to the evaporative air cooler, a differentiated defrosting control process is executed. Evaporative air coolers with the judgment tag indicating that they need to be defrosted immediately are included in the instant defrosting control sequence and a defrosting execution signal is sent. Evaporative air coolers with the judgment tag indicating that they need to be observed later are marked as key monitoring objects and the heat exchange data acquisition frequency is increased. Evaporative air coolers with the judgment tag indicating that they do not need to be defrosted maintain their original operating parameters and retain the observation task records. To achieve accurate control of the defrosting operation of the cold air machine, differentiated defrosting regulation processes need to be performed according to the current determination label of the cold air machine. The defrosting task scheduling controller can be constructed to bind the obtained determination label with the unique identification of the cold air machine, and the classification processing is performed in the task scheduling process. For the cold air machine marked as needing immediate defrosting, a defrosting execution signal is issued to the corresponding cold air machine through the control interface, the electric heating element or the reverse mode of the heat pump is started, and the defrosting trigger time point of this round is recorded in the system log. For the cold air machine in the delayed observation state, the current state is recorded and added to the observation monitoring list, and the system automatically increases the sampling frequency of the sensor data of the cold air machine to capture the subtle changes in its heat exchange performance, so as to determine whether it needs to be further upgraded to the immediate defrosting state. For the cold air machine without defrosting, the current running parameters remain unchanged, and the running data is recorded in the defrosting evaluation period for subsequent verification. This differentiated regulation method ensures that the response strategies of cold air machines in different states are targeted, effectively improving the real-time and accuracy of defrosting control.

[0060] The determination label is a classification result, which is generated by classification according to the defrosting credible evaluation parameters output by the evaluation model, and specifically includes three states of immediate defrosting, delayed observation and no defrosting. The defrosting control sequence refers to the list of cold air machines that prefer to perform defrosting operation in task scheduling, and its adjustment in and out is controlled by the dynamic change of label state. The improvement of heat exchange data collection frequency is mainly realized by adjusting the sampling interval of temperature, current, humidity, air volume and other sensors, which is usually set to more than twice the standard sampling frequency to improve the detection sensitivity. The observation task record refers to the data buffer area used to trace the state evolution process of the cold air machine in subsequent evaluation. These records support model optimization and trend prediction analysis, and provide data support for the defrosting decision of the next cycle.

[0061] The data generated by each cold air machine in the current round, including heat exchange feature data, interference identification index and running state parameters, is input into the heat exchange state space perception graph update engine according to the time index and space position mapping method, and the cold air machine node state label, adjacent airflow interference relationship edge weight and spatial heat load distribution result in the graph are updated synchronously; To achieve dynamic monitoring and intelligent control of the operation state of each air cooler in the refrigeration system, the data generated by each air cooler in the current round, including heat exchange characteristic data, interference identification indicators, and operation state parameters, can be uniformly input into the heat exchange state space perception map update engine according to the time index and spatial location mapping rules. In the implementation process, first, based on the physical layout position, air outlet direction, and number information of the air cooler, the data is positioned to the corresponding node in the map structure; then, based on the data timestamp, the time dimension of the node is aligned, and the data reflecting the heat exchange performance, disturbance situation, and operation mode of the air cooler in the current round is written into the node attribute. For the airflow interference relationship between adjacent air coolers that has been identified, the weight value on the connection edge is updated to reflect the latest airflow interference strength and impact direction. At the same time, combined with the heat exchange data of each air cooler node in the entire space region, the heat load distribution state in the space grid is recalculated to realize the continuous evolution and dynamic reflection of the heat exchange state space perception map, thereby providing real-time data support for subsequent defrosting control strategies and cold chain energy efficiency optimization.

[0062] The heat exchange characteristic data is a heat exchange capacity evaluation index constructed by fusing the temperature, humidity, current, and air volume data collected by the air cooler in real time, reflecting the heat exchange level under the current working condition. The interference identification indicator is used to represent whether the air cooler is affected by the defrosting airflow of the adjacent air cooler, usually including interference offset factor, spatial airflow superposition degree, etc. The operation state parameter includes whether the current air cooler is in the refrigeration, standby, defrosting, or observation state, etc. The time index is a standard timestamp used to locate the data belonging to the time period, and the spatial location mapping method positions the node according to the installation coordinates and spatial layout relationship of the air cooler in the refrigeration area. The heat exchange state space perception map is a dynamic state expression model of the air cooler based on graph structure, whose node represents the air cooler entity, the edge represents the airflow interference relationship, the node attribute includes heat exchange capacity, interference situation, and operation state, etc., and the edge weight reflects the airflow interference strength between air coolers. Through the continuous update of the map, the running coordination relationship and heat load distribution state of the air cooler in the refrigeration area can be mastered in real time, so as to realize more refined defrosting control and energy efficiency management.

[0063] Based on the updated heat exchange state space perception map, the heat exchange behavior association network of each air cooler in the current operation period is reconstructed, the heat load distribution state, defrosting interference path, and control strategy adaptability between air coolers are comprehensively analyzed, and the defrosting control priority ranking result for the next round is generated to realize defrosting control optimization under dynamic regulation.

[0064] To achieve the dynamic optimization of defrosting control, the heat exchange behavior correlation network of each air cooler in the current operation cycle in the refrigeration system can be reconstructed based on the updated heat exchange state spatiotemporal perception graph. The implementation process includes modeling and fusion analysis of the heat load conduction relationship between air cooler nodes, interference transmission path, and historical defrosting execution effect. First, according to the heat exchange feature vector and spatial position information of each node in the graph, the heat exchange coupling relationship between air coolers is extracted through graph mining algorithm, and the heat exchange behavior correlation network is established. Then, the paths that cause significant heat exchange disturbance to other air coolers after defrosting are identified, and a defrosting interference path graph is constructed. At the same time, based on the execution feedback of each air cooler under the previous control strategy, including the heat exchange recovery time after defrosting, energy consumption change and refrigeration environment response result, the adaptability of the control strategy is evaluated. Comprehensive heat load distribution, interference path strength and control strategy adaptability parameters, all air cooler nodes are executed defrosting control priority score and sorting, forming the defrosting execution plan for the next round, so as to realize the defrosting control optimization under dynamic working condition.

[0065] The heat exchange state spatiotemporal perception graph is a graph structure model that records the heat exchange characteristics, interference relationship and operation history of each air cooler node with time and space as dual indexes. The heat exchange behavior correlation network is a directed graph constructed based on the similarity of heat exchange characteristics and spatial proximity between nodes, which is used to express the heat exchange interaction mode between air coolers. The heat load distribution state is formed by spatial mapping of the heat exchange capacity of multiple air coolers in the refrigeration area, which expresses the heat flow and concentration situation. The defrosting interference path is the path expression of the crosstalk direction and strength caused by defrosting airflow in the network, which is used to judge whether a certain air cooler defrosting may interfere with other air coolers. The control strategy adaptability is a multi-dimensional evaluation index generated based on the historical control records of air coolers, which is used to measure the response effect of the existing strategy under specific working conditions. The defrosting control priority sorting result is the basis for decision whether to defrost first in the subsequent control cycle of each air cooler, which combines the running state, environmental feedback and energy efficiency performance, guiding the system to execute an orderly and efficient defrosting plan.

[0066] The above-described embodiments can be implemented in part or in whole through software, hardware, firmware or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When loaded and executed by a computer, the computer instructions or computer programs cause the computer to perform all or part of the processes or functions described in the embodiments of the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired or wireless (for example, infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0067] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0068] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0069] In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the above-described embodiments are only illustrative, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed each other can be indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0070] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may also be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0071] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0072] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for distributed dynamic defrosting control of a cold air machine, characterized in that, Specifically comprising the following steps: S1, obtain the spatial position of the air cooler in the refrigeration environment, the wind direction and flow information, and the defrost exhaust state of the adjacent air cooler, construct a heat exchange state space-time perception map based on the spatial coordinate relationship, and determine whether the air cooler is disturbed by the defrost air flow of the adjacent air cooler; S2, collect temperature, humidity, current, and air volume data of the air cooler determined to be disturbed by the defrost air flow of the adjacent air cooler, construct a heat exchange characteristic curve of the target air cooler in the current period, and perform residual fitting calculation with the standard heat exchange curve in the historical non-defrost state to determine the local heat exchange parameter distortion caused by the defrost air flow disturbance of the adjacent air cooler; S3, fuse the local heat exchange parameter distortion with the real-time heat exchange data of the air cooler, correct the fusion result through the interference weight, input the corrected fusion data into the defrost credibility evaluation model, generate defrost credibility evaluation parameters, and determine whether the air cooler needs to perform defrosting operation according to the local heat exchange parameter distortion caused by the defrost air flow disturbance of the adjacent air cooler; S4, generate a judgment label based on the defrost credibility evaluation parameters combined with the historical operation characteristics of the air cooler and the load wave of the refrigeration area, divide the air cooler into three states of immediate defrosting, delayed observation, and no defrosting; S5, execute differentiated defrosting regulation process according to the judgment label corresponding to the air cooler, and use the data generated by each air cooler in the current round to update the heat exchange state space-time perception map, so as to realize defrosting control optimization under dynamic regulation.

2. The distributed dynamic defrost control method for a cold air machine according to claim 1, wherein, S1 specifically comprises the following steps: S101, obtain the spatial position of each air cooler by deploying a space recognition unit in the refrigeration environment, obtain the wind direction and flow information of the air cooler under different operating conditions in combination with the wind direction and speed detection components configured at the outlet of the air cooler, and identify whether the adjacent air cooler is in a defrost exhaust state based on the air cooler operation control logic and heating start signal; S102, establish a spatial coordinate relationship according to the spatial position and wind direction and flow information of the air cooler, take the air cooler in the defrost exhaust state as a hot and humid disturbance source point, call a three-dimensional disturbance diffusion prediction model to calculate its influence radius and directional propagation path, and then construct a heat exchange state space-time perception map containing the air flow interaction relationship between the air coolers; S103, identify the interference area in the heat exchange state space-time perception map by analyzing the spatial angle between the air cooler and the disturbance source point, the air flow overlapping area, and the disturbance intensity mapping relationship, and determine whether the air cooler is disturbed by the defrost air flow of the adjacent air cooler according to the preset interference judgment rule.

3. The distributed dynamic defrost control method for a cold air machine according to claim 2, wherein, S103 specifically comprises: Calculate the spatial angle between the air cooler and the disturbance source point, take the coincidence degree of the angle range and the main direction of the air flow as the first interference index, and select the potential interference direction through the spatial path with an angle less than the set interference threshold; Extract the spatial volume of the air flow overlapping area, construct a disturbance intensity mapping relationship between the volume ratio and the air flow velocity superposition strength as the second interference index, and mark the overlapping area of the air flow propagation track and intensity distribution in the heat exchange state space-time perception map to identify the interference area meeting the interference index weight threshold; The preset interference judgment rules include an included angle threshold judgment, a disturbance intensity coverage ratio judgment, and an airflow interference duration judgment.

4. The distributed dynamic defrost control method for a cold air machine according to claim 1, wherein, S2 specifically includes the following steps: S201, through the temperature sensor, humidity sensor, current sensor, and air volume detection assembly arranged on the internal heat exchanger surface and the air inlet and outlet path of the air cooler, real-time collection of temperature, humidity, current, and air volume data of the air cooler determined to be interfered by the defrost airflow of the adjacent air cooler in the current operation period, and time alignment and data missing repair processing of different types of sensing data to ensure the continuity and consistency of the input data; S202, after the processed temperature, humidity, current, and air volume data are integrated in time sequence, the heat exchange behavior modeling engine is input, heat exchange characteristic factors are refined through a multivariate feature extraction algorithm, and a heat exchange characteristic curve of the target air cooler in the current period is constructed under the same time dimension to fully reflect the heat load change and operation state linkage relationship; S203, residual fitting calculation is performed on the heat exchange characteristic curve of the target air cooler in the current period and the standard heat exchange curve in the non-defrost state selected from the historical data, the offset amplitude, change trend, and duration of the residual interval are analyzed, the local heat exchange parameter distortion caused by the defrost airflow interference of the adjacent air cooler is extracted, and is used for subsequent defrosting judgment logic input.

5. The distributed dynamic defrost control method for a cold air machine according to claim 4, wherein, S202 specifically includes: The temperature, humidity, current, and air volume data are aligned in sequence on the same time axis, the sliding window mechanism is used to extract the sensing data segments of the continuous time period at equal intervals, the missing or abnormal fluctuation data is completed and corrected by using the bidirectional interpolation method, and the synchronization and data integrity of the multi-source data in the time sequence at each time point are ensured; The data aligned in time sequence is input into the heat exchange behavior modeling engine, heat exchange characteristic factors are extracted from the temperature, humidity, current, and air volume data by using the joint principal component analysis algorithm and clustering decomposition algorithm, and redundant and low correlation factors are removed to form a characteristic vector set for describing the heat exchange behavior state of the target air cooler; The characteristic vector set is mapped in time sequence to construct a heat exchange characteristic curve of the target air cooler in the current period, wherein the vertical axis is the refined heat exchange characteristic factor group, the horizontal axis is the time axis consistent with the sensing data, and the response law of the current heat exchange capacity and operation state of the target air cooler under the dynamic change of the heat load is reflected through the heat exchange characteristic curve.

6. The distributed dynamic defrost control method for a cold air machine according to claim 4, wherein, S203 specifically includes: In the historical operation data, the time period closest to the current refrigeration environment load state and operation condition characteristics is screened out, the standard heat exchange curve of the target cold air fan in the time period is extracted, and the standard curve is ensured to be free of data interference during defrosting, serving as a reference for the current heat exchange characteristic curve; The heat exchange characteristic curve of the target cold air fan in the current period is compared with the standard heat exchange curve point by point along the time axis, residual fitting calculation is performed in a sliding residual window mode, the numerical deviation and directional change of the characteristic factor are calculated in each window, a residual time series is obtained, and abnormal peak disturbance values are removed; The residual time series is subjected to dynamic clustering and segmentation analysis, residual intervals with a deviation amplitude exceeding a warning interval threshold are extracted, and the unidirectionality and duration characteristics of the deviation trend in these interval segments are combined to determine that the local heat exchange parameter distortion caused by the defrosting airflow interference of the adjacent cold air fan represents the cold air fan, and the distortion is passed to the subsequent defrosting judgment logic.

7. The distributed dynamic defrost control method for a cold air machine according to claim 1, wherein, S3 specifically includes the following steps: S301, the local heat exchange parameter distortion caused by the defrosting airflow interference of the adjacent cold air fan is combined with the real-time heat exchange data collected by the cold air fan in the current period according to the time dimension, the features in the same dimension are numerically superimposed by using feature vector weighted operation during the fusion process, and fusion heat exchange feature data capable of representing the heat exchange state of the cold air fan is formed; S302, according to the spatial position relationship between the cold air fan and the disturbance source point, the airflow superposition intensity and the interference coverage ratio, the interference weight is calculated and applied to the feature dimensions selected by the interference index weight threshold in the fusion heat exchange feature data, and the weighted correction is performed on each feature dimension, and the modified fusion data after interference compensation processing is output; S303, the modified fusion data is input into the defrosting credibility evaluation model, and the credible classification algorithm and the heat exchange performance deviation identification algorithm trained in the evaluation model are used to generate defrosting credibility evaluation parameters, wherein the defrosting credibility evaluation parameters include heat exchange performance degradation degree, interference offset factor and duration index; according to whether the heat exchange performance degradation degree continuously exceeds the normal operation interval threshold, whether the interference offset factor exceeds the interference source influence threshold, and whether the duration index exceeds the preset time threshold, it is judged that the cold air fan meeting at least two index conditions is the cold air fan that really needs to perform defrosting operation.

8. The distributed dynamic defrost control method for a cold air machine according to claim 7, wherein, S302 specifically includes: The spatial position relationship parameters between the cold air fan and the disturbance source point in the current period are obtained, the angle between the cold air fan outlet direction and the disturbance source propagation direction is calculated, the airflow superposition intensity parameters are determined by combining the spatial intersection proportion of the airflow superposition area and the wind speed information of the disturbance source point propagation path, and the interference area volume proportion parameters covered by the cold air fan heat exchange surface are collected, forming an input data set for constructing the interference weight; A weighted function is constructed based on the spatial position included angle, air flow superposition strength parameter and interference coverage ratio parameter to generate an interference weight value representing the interference degree. The interference weight value is applied to each feature dimension in the fused heat exchange feature data to extract feature dimensions with interference sensitivity coefficients exceeding an interference index weight threshold to form a set of feature dimensions to be corrected. A weighted correction operation is performed on each feature dimension in the set of feature dimensions to be corrected. The interference weight value and the original value of the corresponding feature dimension are weighted and superimposed to generate corrected feature data for replacing the original value. Finally, the corrected fusion data after interference compensation processing is output as the input data source of the subsequent defrosting reliable evaluation model.

9. The distributed dynamic defrost control method for a cold air machine according to claim 1, wherein, S4 specifically is: Based on the joint state of the heat exchange performance degradation degree, the interference offset factor and the duration index in the defrosting reliable evaluation parameters, the numerical features of each heat exchange parameter of the cold air fan in the historical operation cycle are extracted to construct a historical operation feature vector including the heat exchange efficiency decline amplitude, the current load response amplitude and the humidity and heat fluctuation stability, which is used to describe the heat exchange behavior of the cold air fan under different load conditions. The load fluctuation data of the current refrigeration area are collected, and the feature factors including the spatial temperature and humidity distribution, the goods flow frequency and the transient cold load change rate are extracted. The historical operation feature vector and the current load fluctuation factor are jointly input into the defrosting state judgment engine to generate a current defrosting judgment label of the cold air fan through the trained judgment model. According to the judgment label output result, the cold air fan is divided into three states of needing immediate defrosting, needing delayed observation and not needing defrosting. The state of needing immediate defrosting satisfies that the heat exchange performance degradation degree continuously exceeds the normal operation interval threshold, the interference offset factor exceeds the interference source influence threshold and the duration index exceeds the preset time threshold. The state of needing delayed observation corresponds to that the heat exchange performance degradation degree is within the upper limit interval of the normal operation interval threshold, and the interference offset factor or the duration index satisfies one of them. The state of not needing defrosting corresponds to that the heat exchange performance degradation degree, the interference offset factor and the duration index all do not exceed the respective limited threshold.

10. The distributed dynamic defrost control method for a cold air machine according to claim 1, wherein, S5 specifically is: According to the judgment label corresponding to the cold air fan, a differentiated defrosting regulation process is performed. The cold air fan with the judgment label of needing immediate defrosting is included in the immediate defrosting control sequence and sends a defrosting execution signal. The cold air fan with the judgment label of needing delayed observation is marked as a key monitoring object and the heat exchange data collection frequency is improved. The cold air fan with the judgment label of not needing defrosting maintains the original operation parameters and retains the observation task record. The data generated by each cold air fan in the current round, including the heat exchange feature data, the interference identification index and the operation state parameter, are input into the heat exchange state space perception graph update engine in a time index and space position mapping manner. The cold air fan node state label, adjacent air flow interference relationship edge weight and spatial heat load distribution result in the graph are updated synchronously. Based on the updated heat exchange state space-time perception atlas, the heat exchange behavior correlation network of each cold air blower in the current operation cycle is reconstructed, the heat load distribution state, defrosting interference path and control strategy adaptability among the cold air blowers are comprehensively analyzed, the defrosting control priority ranking result for the next round is generated, and the defrosting control optimization under dynamic regulation is realized.

Images