A method, system and medium for defrosting disturbance denoising and temperature control quality baseline calculation

Abstract

The application discloses a defrosting disturbance denoising and temperature control quality baseline calculation method and system and a medium, and the method comprises the following steps: acquiring an original temperature sequence of a target space; identifying a defrosting event based on the original temperature sequence, and dividing a time axis into a disturbance section and a recovery section according to the defrosting event; performing constrained interpolation correction on temperature data in the disturbance section based on stable section temperature data before and after the disturbance section, to generate a denoised temperature sequence; calculating a temperature control quality index based on the denoised temperature sequence, and constructing a temperature control quality baseline. Through the organic combination of event identification, segmented modeling, disturbance correction and baseline statistics, the application realizes accurate denoising of interpretable disturbances such as defrosting, and constructs a scientific and quantifiable temperature control quality baseline system.

Description

Technical Field

[0001] This invention relates to the field of cold chain temperature control data processing and quality assessment technology, and in particular to a method, system and computer-readable storage medium for identifying and denoising temperature fluctuations caused by explainable disturbances such as defrosting, and calculating a temperature control quality baseline based on these events. Background Technology

[0002] This invention aims to address the problem in existing cold chain temperature control data processing where abnormal fluctuations in temperature curves caused by explainable disturbances such as defrosting affect the accuracy of temperature control quality assessment. Specifically, existing technologies often simply remove defrosting periods or apply low-pass filtering to temperature sequences, failing to accurately identify disturbance and recovery segments. This can easily lead to misjudgments of real anomalies or masking of equipment malfunctions, and lacks a practical and interpretable baseline calculation scheme for temperature control quality. Furthermore, reliably identifying defrosting events and achieving noise reduction when different project locations have incomplete configurations (e.g., lacking defrosting control signals) is also a pressing technical challenge. Summary of the Invention

[0003] The present invention aims to provide a method, system and medium for defrosting disturbance denoising and temperature control quality baseline calculation, so as to solve the technical problems in the prior art that it is difficult to distinguish between load changes and efficiency decay, difficult to locate faulty components, and lack of interpretability of diagnostic results.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] A method for defrosting disturbance denoising and temperature control quality baseline calculation includes the following steps:

[0006] Obtain the original temperature sequence of the target space;

[0007] Defrost events are identified based on the original temperature sequence, and the time axis is divided into a perturbation segment and a recovery segment according to the defrost events;

[0008] Based on the temperature data of the stable segment before and after the disturbance segment, the temperature data within the disturbance segment is subjected to constrained interpolation correction to generate a denoised temperature sequence.

[0009] The temperature control quality index is calculated based on the denoised temperature sequence, and a temperature control quality baseline is constructed.

[0010] In some embodiments, the step of identifying defrosting events includes:

[0011] Acquire device control signals or operating status data of the target space;

[0012] When at least one of a defrost command signal, a heating status signal, or a fan start / stop status signal is present, the start and end points of the defrost event are determined directly based on the rising and falling edges of the signal.

[0013] When the device control signal is missing, the start and end points of the defrosting event are inferred by combining the temperature slope criterion of the original temperature sequence with at least one auxiliary quantity, such as the humidity signal of the target space, the door switch signal, or the evaporator coil temperature signal.

[0014] In some embodiments, the step of dividing the disturbance segment and the recovery segment includes:

[0015] Mark the moment when defrosting begins or when the temperature slope first exceeds the temperature rise threshold as the start of the disturbance segment t0;

[0016] The moment when the temperature reaches its peak or the temperature rise slope falls back below the temperature rise threshold is marked as the end of the disturbance segment t1;

[0017] The moment when the temperature drops back to within the set allowable deviation band and remains stable, or enters a local stable range, is marked as the end point of the recovery segment, t2.

[0018] The period after t2 is marked as the stable period.

[0019] In some embodiments, the step of constrained interpolation correction of the temperature data within the disturbance segment includes:

[0020] Within the stable segment before the start of the disturbance segment t0, a first window is selected, and the trend prediction function g_pre(t) is obtained by fitting.

[0021] Within the recovery segment, a second window is selected, and the recovery dynamic parameters are fitted based on the first-order inertial model T(t)=T_inf + (T(t1)-T_inf)·exp(-(t-t1) / τ), where T(t1) is the end temperature of the disturbance segment, τ is the recovery time constant, and T_inf is the steady-state recovery temperature.

[0022] Based on the trend prediction function g_pre(t) and the recovery dynamics parameters, a continuous and consistent denoised temperature value T_hat(t) is constructed within the disturbance segment [t0, t1] that is consistent with the trend of the recovery segment.

[0023] In some embodiments, the temperature control quality indicators include at least one of the following: compliance rate, stability, over-limit integral, uniformity, or defrosting impact indicators;

[0024] The compliance rate is the percentage of time the temperature falls within the set allowable deviation band; the stability is the percentage or standard deviation of the temperature deviation from the set value; the over-limit integral is the integral of the magnitude and duration of the temperature deviation from the set allowable deviation band; the uniformity is the maximum difference or percentage difference of the temperature at multiple points in space at the same time; and the defrosting impact index includes the peak temperature of the disturbance segment, the duration, or the area of ​​the disturbance.

[0025] In some embodiments, the step of constructing a temperature control quality baseline includes:

[0026] The temperature control quality indicators are grouped and statistically analyzed according to spatial identification, season, month, or operating mode.

[0027] Calculate the median of each set of indicators and the preset percentage range, which will serve as the reference range for the temperature control quality baseline.

[0028] In some embodiments, an alarm and work order closure step is also included:

[0029] When an over-temperature event is detected, it is determined whether the over-temperature event occurred within the identified defrosting disturbance segment;

[0030] If so, the over-temperature event is marked as an explainable disturbance event and recorded or notified according to the preset low-level strategy;

[0031] If not, the system will determine a genuine anomaly based on the door open / close status or equipment shutdown signal, triggering a high-level alarm and generating a work order.

[0032] In some embodiments, a multi-probe fault-tolerant processing step is also included:

[0033] When the difference in readings from multiple temperature probes in the same space exceeds a preset threshold, the data from the preferred probe is selected based on a preset confidence rule or the proximity to the return air location for the generation of the denoised temperature sequence and the calculation of the temperature control quality index.

[0034] This embodiment also provides a system for defrosting disturbance denoising and temperature control quality baseline calculation, including:

[0035] The data acquisition module is used to acquire the original temperature sequence and equipment status signals of the target space;

[0036] The event identification and segmentation module is used to execute any of the methods described in this article, identify defrosting events and divide them into disturbance segments and recovery segments;

[0037] A noise reduction module is used to execute the method described above and generate a noise-reduced temperature sequence.

[0038] The baseline calculation and evaluation module is used to execute methods, calculate temperature control quality indicators, and construct temperature control quality baselines.

[0039] In some embodiments, the computer program, when executed by a processor, implements the method as described in any one of claims 1 to 8. Attached Figure Description

[0040] Figure 1 This is a schematic diagram illustrating the defrosting event identification and segmentation provided in an embodiment of the present invention. The diagram shows a typical curve of cold storage temperature changing over time, and marks the disturbance segment (the area of ​​abnormal temperature increase), the recovery segment (the area of ​​temperature returning to a stable range), and the stable segment (the stable range after recovery).

[0041] Figure 2 This is a flowchart illustrating the process of generating a denoised temperature sequence according to an embodiment of the present invention. The flowchart shows the complete steps from the original temperature sequence, through pre-perturbation trend fitting, restoration of kinetic parameter fitting, and constrained interpolation construction, to finally generating a denoised temperature sequence.

[0042] Figure 3 This diagram illustrates the calculation of temperature control quality indicators and baselines provided in an embodiment of the present invention. It shows the process of calculating compliance rate, stability, and out-of-limit integrals on a denoised temperature sequence, and constructing a baseline reference range through grouped statistics.

[0043] Figure 4 This diagram illustrates the spatial uniformity assessment using multiple probes, as provided in an embodiment of the present invention. It shows the differences in temperature probe readings at different locations at the same time, and how spatial temperature uniformity can be assessed by calculating the maximum difference or percentile difference.

[0044] Figure 5 This diagram illustrates the system architecture and data flow provided in an embodiment of the present invention. It shows the connections and data flow between the data acquisition module, event recognition and segmentation module, noise reduction module, baseline calculation and evaluation module, alarm and work order module, and storage module. Detailed Implementation

[0045] Various exemplary embodiments, features, and aspects of this application will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0046] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, an indirect connection through an intermediate medium, or the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0047] See Figures 1-5 As shown, a method for defrosting disturbance denoising and temperature control quality baseline calculation includes the following steps: acquiring the original temperature sequence of the target space; identifying defrosting events based on the original temperature sequence, and dividing the time axis into a disturbance segment and a recovery segment according to the defrosting events; performing constrained interpolation correction on the temperature data within the disturbance segment based on the stable segment temperature data before and after the disturbance segment to generate a denoised temperature sequence; calculating the temperature control quality index based on the denoised temperature sequence, and constructing the temperature control quality baseline.

[0048] First, raw temperature data is collected. Then, the time interval of defrosting events is identified through an identification algorithm and subdivided into a disturbance segment (abnormal temperature rise) and a recovery segment (temperature drop). Next, using stable segment data before the start of the disturbance and after the end of the recovery, mathematical modeling and interpolation are performed on the disturbance segment to generate a "denoised" temperature curve that eliminates the impact of defrosting. Finally, based on this denoised sequence, compliance rate, stability and other indicators are calculated, and a long-term statistical baseline is formed.

[0049] By using defrost event identification and segmentation as a preliminary step, the perturbation range can be accurately located, avoiding misjudgments caused by simple elimination or smoothing. Constrained interpolation correction ensures the continuity between the denoised sequence and the true trend, eliminating defrost interference while retaining the characteristics of other real anomalies (such as doors that are always open). Finally, a baseline is constructed on the denoised sequence, making the temperature control quality assessment more objective and accurate, providing a reliable basis for subsequent alarm and energy-saving analysis. This invention combines "defrost event identification and segmentation" with "constrained interpolation correction," solving the problems of unclear perturbation range definition and uninterpretable correction methods in existing technologies, and achieving accurate assessment of temperature control capabilities while excluding interpretable perturbations.

[0050] The steps for identifying defrost events include: acquiring equipment control signals or operating status data of the target space; when at least one of a defrost command signal, heating status signal, or fan start / stop status signal is present, directly determining the start and end points of the defrost event based on the rising and falling edges of the signal; when the equipment control signal is missing, inferring the start and end points of the defrost event by combining at least one auxiliary quantity, such as the temperature slope criterion of the original temperature sequence, the humidity signal of the target space, the door switch signal, or the evaporator coil temperature signal, through a combined criterion.

[0051] This method provides two parallel identification paths: Path A (direct signal connection) uses readily available control signals to accurately pinpoint the start and end of defrosting; Path B (morphological inference) analyzes the slope of temperature change and combines data from other sensors (such as humidity and coil temperature) to make a comprehensive judgment on multiple features at points without signals. Only when multiple criteria are met (such as the temperature rise slope exceeding the standard and the fan stopping) is it determined to be a defrosting event.

[0052] By combining direct signal connection with morphological inference, this invention can adapt to the hardware configurations of different projects, reliably identifying defrosting events even at locations without defrosting feedback signals, greatly improving the method's versatility and engineering feasibility. Simultaneously, introducing auxiliary parameters such as humidity and door opening / closing status effectively distinguishes between defrosting disturbances and loading / unloading disturbances, avoiding misjudgments and improving identification accuracy.

[0053] The steps for dividing the disturbance segment and the recovery segment include: marking the moment when defrosting begins or the temperature slope first exceeds the heating threshold as the start of the disturbance segment t0; marking the moment when the temperature reaches its peak or the heating slope falls back below the heating threshold as the end of the disturbance segment t1; marking the moment when the temperature falls back to within the set allowable deviation band and remains stable, or enters a local stable range as the end of the recovery segment t2; and marking the period after t2 as the stable segment.

[0054] This method refines the defrosting influence range based on the dynamic characteristics of the temperature curve: the starting point of the temperature rise is taken as the start of the disturbance, the peak point or the end of the temperature rise is taken as the end of the disturbance, and the temperature stabilization is taken as the end of the recovery, thus dividing the entire defrosting process into three segments, providing clear boundaries for subsequent interpolation correction.

[0055] By introducing quantitative indicators such as heating threshold, peak detection, and allowable deviation band, adaptive division of the disturbance and recovery segments is achieved, avoiding errors caused by fixed-duration elimination (e.g., the defrosting effect may continue after the command ends). This division can accurately capture the entire range of defrosting's influence on temperature, ensuring complete denoising coverage, while preserving the true dynamic characteristics of the recovery segment, providing a basis for subsequent modeling.

[0056] The steps for constrained interpolation correction of temperature data within the disturbance segment include: selecting a first window within the stable segment before the start of the disturbance segment t0, and fitting a trend prediction function g_pre(t); selecting a second window within the recovery segment, and fitting recovery dynamic parameters based on the first-order inertial model T(t)=T_inf + (T(t1)-T_inf)·exp(-(t-t1) / τ), where T(t1) is the end temperature of the disturbance segment, τ is the recovery time constant, and T_inf is the steady-state recovery temperature; and constructing a continuous and consistent denoised temperature value T_hat(t) within the disturbance segment [t0, t1] based on the trend prediction function g_pre(t) and the recovery dynamic parameters.

[0057] This method uses the trend of the stable segment before the disturbance (such as steady or slow change) to predict the temperature trend that should occur if defrosting does not occur. At the same time, it uses the data of the recovery segment to fit the dynamic characteristics of the system recovery (first-order inertial model). Then, it constructs an interpolation curve in the disturbance segment that connects the preceding trend and smoothly transitions to the recovery segment at the end point, thereby achieving the denoising reconstruction of the entire disturbance range.

[0058] Employing a physically based recovery model (first-order inertia) instead of purely mathematical smoothing ensures the interpretability of the interpolation results, reflecting the actual refrigeration recovery process in the cold storage. Simultaneously, using the pre-perturbation trend as a constraint guarantees that the denoised sequence does not introduce artificial jumps and remains consistent with reality. This constrained interpolation method effectively eliminates interference from defrosting temperature rise without masking genuine anomalies such as doors remaining open (because doors remaining open often lack this recovery pattern or are accompanied by other characteristics), achieving a balance between accurate denoising and anomaly preservation.

[0059] The temperature control quality indicators include at least one of the following: compliance rate, stability, over-limit integral, uniformity, or defrosting impact indicator; the compliance rate is the percentage of time the temperature falls within the set allowable deviation range; the stability is the percentage or standard deviation of the temperature deviation from the set value; the over-limit integral is the integral of the magnitude and duration of the temperature deviation from the set allowable deviation range; the uniformity is the maximum difference or percentage difference of the temperature at multiple points in space at the same time; the defrosting impact indicator includes the peak temperature of the disturbance segment, the duration, or the area of ​​the disturbance.

[0060] The above indicators are calculated on the denoised temperature series, where compliance rate reflects the overall compliance status, stability reflects the degree of fluctuation, over-limit integral measures the severity of over-temperature, uniformity assesses the spatial temperature difference, and the defrosting impact indicator is specifically used to evaluate the merits of defrosting strategies.

[0061] By defining a multi-dimensional temperature control quality index system, the temperature control performance of cold storage can be comprehensively characterized. This includes not only conventional pass rates and stability, but also more refined over-limit integrals and uniformity. A defrosting impact index is also specifically introduced to optimize defrosting strategies. These indicators are all calculated based on denoised sequences, eliminating defrosting interference, thus more accurately reflecting the equipment's temperature control capabilities and operating status, providing rich data support for operation and maintenance decisions.

[0062] The steps for constructing a temperature control quality baseline include: grouping and statistically analyzing the temperature control quality indicators according to spatial identification, season, month, or operating mode; calculating the median and preset percentage range of each group of indicators as the reference range for the temperature control quality baseline.

[0063] This method groups historical denoised sequences according to different dimensions (such as different cold storage facilities, different seasons, and different operating conditions), and calculates the median and, for example, the 90th percentile of each indicator to form a dynamic baseline range for subsequent real-time monitoring and comparison.

[0064] By using grouped statistics, the impact of different spaces and environmental conditions on temperature control was considered, making the baseline more targeted and adaptable. Using the median and percentiles instead of simple mean and standard deviation helps to resist interference from individual outliers, improving the robustness of the baseline. This baseline can serve as the basis for setting subsequent alarm thresholds, enabling adaptive early warning.

[0065] In some embodiments, the alarm and work order closed-loop steps are also included: when an over-temperature event is detected, it is determined whether the over-temperature event occurs within the identified defrosting disturbance segment; if so, the over-temperature event is marked as an interpretable disturbance event and recorded or notified according to a preset low-level strategy; if not, it is determined as a real abnormality by combining the door open / close status or equipment shutdown signal, triggering a high-level alarm and generating a handling work order.

[0066] In real-time monitoring, when the temperature exceeds the allowable range, the system first checks whether the current moment falls within the identified defrosting disturbance segment. If so, it is considered a normal fluctuation caused by defrosting, and only a record or low-level prompt is made. If not, it further combines access control, equipment status, etc. to determine whether it is a real fault, thereby triggering alarms and work orders of the corresponding level.

[0067] This closed-loop mechanism effectively distinguishes between explainable disturbances and genuine anomalies, significantly reducing false alarms caused by defrosting and enabling maintenance personnel to focus on actual equipment failures or operational issues. Furthermore, if long-term statistical defrosting impact indicators deviate from the baseline, it can generate suggested work orders for optimizing defrosting strategies, achieving closed-loop management from data to action.

[0068] In some embodiments, a multi-probe fault-tolerant processing step is also included: when the difference in readings of multiple temperature probes in the same space exceeds a preset threshold, based on a preset confidence rule or the proximity to the return air location, the data of the preferred probe is selected for the generation of the denoised temperature sequence and the calculation of the temperature control quality index.

[0069] In a cold storage facility with multiple temperature probes, if the temperature difference between probes is too large (which may be due to location, malfunction, etc.), the system selects data from one or a group of representative probes as the main data source according to preset rules (such as selecting the probe closest to the return air vent, or assigning different weights based on historical reliability) for subsequent noise reduction and index calculation.

[0070] This fault-tolerance mechanism avoids evaluation bias caused by individual probe malfunctions or improper installation locations, improving system robustness and data quality, and ensuring the reliability of baseline calculations. Automatic probe selection also reduces the burden of manual judgment.

[0071] This embodiment also provides a system for defrosting disturbance denoising and temperature control quality baseline calculation, including: a data acquisition module for acquiring the original temperature sequence and equipment status signal of the target space; an event recognition and segmentation module for recognizing defrosting events and dividing them into disturbance segments and recovery segments; a denoising processing module for generating a denoised temperature sequence; and a baseline calculation and evaluation module for calculating temperature control quality indicators and constructing a temperature control quality baseline.

[0072] The data acquisition module is responsible for collecting raw data; the event identification and segmentation module preprocesses the data, identifies defrosting events, and divides them into time periods; the denoising module interpolates and corrects disturbed segments to generate denoised sequences; and the baseline calculation and evaluation module calculates indicators and establishes baselines based on the denoised sequences. These modules work together to achieve fully automated processing from raw data to quality baselines.

[0073] This system modularizes the methods described, forming a complete product-level solution that can be deployed independently or integrated into existing cold chain monitoring platforms. Each module has clearly defined responsibilities, facilitating maintenance and upgrades. Furthermore, data flow between modules ensures the orderly execution of the methods, improving system efficiency and reliability.

[0074] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the methods described herein.

[0075] The storage medium contains program code that implements the method of the present invention, and when run on a computing device, it can perform a series of steps such as defrosting recognition, noise reduction correction, and baseline calculation.

[0076] By using storage media, the method of this invention can be easily disseminated, deployed, and updated, enabling existing cold chain monitoring systems to acquire the functionality of this invention through software upgrades, thereby reducing implementation costs and improving the accessibility of the technology.

[0077] Example 2: Calculation of Baseline for Defrosting Disturbance Reduction and Temperature Control Quality in Cold Chain Cold Storage

[0078] This embodiment uses a typical cold chain cold storage facility as an application scenario to describe the implementation process of the present invention in detail. The cold storage facility is set to a temperature of -18℃, with an allowable deviation range of ±2℃ (i.e., [-20℃, -16℃]). Temperature probes are installed inside the storage facility, including return air vent temperature probes and cargo area temperature probes, with a sampling frequency of 1 minute. The PLC system provides defrosting command signals, fan start / stop status signals, and optionally provides door open / close signals and evaporator coil temperature signals. This embodiment will combine... Figures 1 to 5 The implementation details of the method and system of this invention are fully demonstrated.

[0079] I. System Architecture and Data Flow

[0080] First, such as Figure 5 As shown, the present invention provides a defrosting disturbance denoising and temperature control quality baseline calculation system, including a data acquisition module, an event recognition and segmentation module, a denoising processing module, a baseline calculation and evaluation module, an alarm and work order module, and a storage module. The connection relationships and data flow between the modules are as follows: Figure 5 As shown:

[0081] The data acquisition module connects to temperature sensors, PLC controllers, door magnetic sensors, and other devices at the cold storage site to collect raw temperature sequences and equipment status signals in real time, and performs preprocessing (time alignment, defect marking, and abnormal jump filtering).

[0082] The preprocessed data is transmitted to the event recognition and segmentation module, which performs defrosting event detection and interval division, and outputs the recognition results (disturbance segment, recovery segment, and stable segment).

[0083] The denoising module receives the event segmentation results, performs constrained interpolation correction on the disturbance segments, generates a denoised temperature sequence, and stores the results in the storage module.

[0084] The baseline calculation and evaluation module periodically reads the denoised sequence from the storage module, calculates the temperature control quality index and constructs the baseline, and writes the results back to the storage module.

[0085] The alarm and work order module monitors the temperature in real time, and, based on the event identification results and baseline range, triggers alarms or work orders of the corresponding level and stores the records in the storage module.

[0086] The following will be arranged in the order of data flow, combined with Figures 1 to 4 Provide a detailed explanation of the implementation details for each step.

[0087] II. Data Acquisition and Preprocessing

[0088] The data acquisition module obtains the following data from the field equipment:

[0089] Temperature probe sequence: includes minute-level temperature data from at least the return air vent probe, denoted as T_raw(t).

[0090] Equipment status signals: defrost command signal defrost_cmd(t) (0 indicates off, 1 indicates on), fan operating status fan_status(t) (0 indicates off, 1 indicates running).

[0091] Optional auxiliary signals: door switch signal door_status(t) (0 indicates closed, 1 indicates open), evaporator coil temperature T_coil(t).

[0092] The preprocessing steps include:

[0093] Time alignment: Align all signals with a uniform timestamp to ensure that data at the same time corresponds.

[0094] Defect marking: If temperature data is missing at a certain moment, that point is marked as a defect. If consecutive defects exceed 10 minutes, that period is skipped in subsequent processing; if the defect duration is short (e.g., less than 5 minutes), linear interpolation is used to complete it. However, if the defect covers the critical boundary of the defrosting event (e.g., the start or end of the disturbance segment), the event is marked as "cannot be denoised" according to the fault tolerance rules.

[0095] Abnormal jump filtering: A sliding median filter is used to eliminate instantaneous spikes in the sensor. Specifically, for each sampling point, the temperature values ​​of 5 points (2 points before and 2 points after it) are taken, and the median is calculated. If the difference between the current temperature and the median exceeds 3℃, it is considered a spike and the median is used instead.

[0096] III. Identification and Segmentation of Defrosting Events

[0097] like Figure 1 As shown, defrosting events cause the temperature curve to exhibit a typical "rise-peak-fall" pattern. This embodiment employs two identification paths (automatically selected based on the point configuration) and divides the time axis into a disturbance segment, a recovery segment, and a stable segment.

[0098] (a) Identification of defrosting events

[0099] Path A (Direct Signal Connection): Since this cold storage facility has a defrost command signal, the system directly uses the rising edge of defrost_cmd(t) as the start time t_start of the defrost event and the falling edge as the end time t_end of the event. To avoid signal jitter, the defrost command is de-jittered: the status is only considered valid if it lasts for more than 2 sampling points (i.e., 2 minutes).

[0100] Path B (Morphological Inference): In other cold storage facilities lacking defrost command signals, the system uses morphological inference. For example, continuously monitoring the temperature slope ΔT / Δt, if the slope is greater than the temperature rise threshold k1 = 0.5℃ / min for three consecutive sampling points, and the fan is off (if available), and the coil temperature rises rapidly, then a defrost event is determined to have begun. The event ends when the temperature slope turns negative and the fan resumes operation.

[0101] In this embodiment, due to the presence of a signal, path A is used to obtain the start and end times of the defrost event: t_start=10:00, t_end=10:12 (the defrost command lasts for 12 minutes). However, the actual temperature response is as follows: Figure 1 As shown, the temperature started to rise at 10:00, reached a peak of -13℃ at 10:12, and then continued to drop until it stabilized at -18℃ at 10:30.

[0102] (II) Division of Disturbance Segment and Recovery Segment

[0103] Based on the characteristics of the temperature curve, the system automatically divides the system into three intervals (see...). Figure 1 ):

[0104] Disturbance segment: The starting point t0 is taken as the defrosting start time of 10:00 (or the moment when the temperature rise slope first exceeds the threshold, which is consistent here); the ending point t1 is taken as the moment when the temperature reaches the peak value of 10:12.

[0105] Recovery phase: The endpoint t2 is the moment when the temperature drops back to within the allowable deviation range [-20℃, -16℃] and remains stable (e.g., the temperature is within the deviation range for 5 consecutive minutes with fluctuations of less than 0.5℃), which is 10:30 in this case.

[0106] Stable period: The time period after t2.

[0107] The specific division method is as follows:

[0108] The starting point of the disturbance segment t0: If path A is used, t_start is taken directly; if path B is used, the moment when the temperature rise slope first exceeds the threshold is taken.

[0109] The end point of the disturbance segment t1: the moment corresponding to the maximum temperature after t0, or the moment when the temperature rise slope first turns from positive to negative and remains below the threshold.

[0110] Recovery segment endpoint t2: After t1, the moment when the search temperature first enters the set deviation band and remains there for n consecutive points (e.g., n=5); if it never enters the deviation band, the moment when the temperature change rate is less than a certain threshold (e.g., 0.1℃ / min) is taken.

[0111] IV. Noise Reduction Temperature Sequence Generation

[0112] The core of this step is to perform constrained interpolation on the perturbation segment [t0, t1] to generate a denoised temperature T_hat(t), ensuring that the corrected sequence eliminates the abnormal temperature rise caused by defrosting without contradicting the actual recovery process. The specific process is as follows: Figure 2 As shown, it includes the following sub-steps:

[0113] Step S1: Fit the trend before the perturbation

[0114] Before time t0, select a first window W_pre with a length of 30 minutes (configurable from 20 to 60 minutes). Use the stable temperature data within the window (with defects removed) to fit a trend prediction function g_pre(t). Since the stable temperature usually fluctuates slightly around the set value, this embodiment uses constant value fitting, i.e., taking the average temperature within the window as the predicted value: g_pre(t) = T_avg_pre = -18.2℃. If linear drift exists, linear fitting can also be used.

[0115] Step S2: Fit and recover dynamic parameters

[0116] Within the recovery segment [t1, t2], a second window W_rec is selected, with a window length encompassing all data from t1 to t2 (if t2-t1 is too long, the maximum length can be limited, for example, to 60 minutes). A first-order inertial model is used to fit the recovery process:

[0117] T(t) = T_inf + (T(t1) - T_inf) · exp(-(t-t1) / τ)

[0118] in:

[0119] t is the time variable within the recovery segment, and t ≥ t1;

[0120] T(t) is the fitted temperature value at time t within the recovery segment;

[0121] T(t1) is the measured temperature at the end of the disturbance segment. In this example, T(t1) = -13℃.

[0122] T_inf is the steady-state recovery temperature to be fitted;

[0123] τ is the recovery time constant to be fitted.

[0124] Nonlinear regression was performed on the measured data of the recovery segment using the least squares method, yielding T_inf = -18.1℃ and τ = 8.5 minutes. The goodness of fit R² = 0.98, indicating a good model fit.

[0125] Step S3: Construct the noise reduction temperature of the disturbance section

[0126] The denoised temperature T_hat(t) at any time t within the disturbance segment [t0, t1] should satisfy:

[0127] Connect with the trend before the disturbance at t0: T_hat(t0) = g_pre(t0).

[0128] Connect to the recovery model at t1: T_hat(t1) = T(t1) (i.e., the measured peak temperature, to ensure continuity).

[0129] The intermediate transition is smooth, and either linear interpolation or spline interpolation can be used. This embodiment uses linear interpolation, that is:

[0130] T_hat(t) = g_pre(t0) + (T(t1) - g_pre(t0)) * (t - t0) / (t1 - t0)

[0131] The calculated T_hat(t) changes linearly from -18.2℃ at t0 to -13℃ at t1.

[0132] Step S4: Generate a complete denoised temperature sequence

[0133] For t ∈ [t0, t1], the original temperature is replaced by T_hat(t); for t > t1, this embodiment chooses to directly use the original temperature (because the measured temperature in the recovery segment already reflects the actual cooling process and does not need correction; however, if there is a slight perturbation in the recovery segment, a slight correction can be made, such as smoothing according to the recovery model). The final denoised temperature sequence T_denoised(t) is obtained, as follows: Figure 2 As shown.

[0134] V. Calculation of Temperature Control Quality Indicators

[0135] Various temperature control quality indicators are calculated on the denoised temperature series, and the statistical period can be set to days, weeks, months, etc. This example uses one month as an example; the calculation results are as follows: Figure 3 As shown.

[0136] Compliance rate: The percentage of points in the denoised sequence whose temperature falls within [-20℃, -16℃]. Calculated as 98.5%.

[0137] Stability: Calculate the absolute value of the temperature deviation from the set value by -18℃, and take the 95th percentile (P95). The calculated value is 1.2℃.

[0138] Over-limit integral: For each over-limit event, the integral is ∫ max(0, |T-set value|-Δ) dt, where Δ=2℃. The summation yields a monthly over-limit integral of 3.5℃·h.

[0139] Uniformity: If multiple probes exist, such as Figure 4 As shown, the maximum temperature difference between different probes at the same time is calculated, and P95 is taken. In this embodiment, the P95 difference between the return air vent and the cargo area probe is 1.8℃.

[0140] Defrosting impact indicators: For each defrosting event, the peak temperature of the disturbance segment (-13℃), the duration of the disturbance (12 minutes), and the disturbance area (integral temperature rise) were statistically analyzed. During this month, the average peak defrosting temperature was -12.8℃, and the average disturbance area was 4.2℃·h.

[0141] VI. Establishment of Temperature Control Quality Baseline

[0142] The above indicators are grouped and statistically analyzed by space (different cold storage facilities) and season (or month). For example, for this cold storage facility, the indicators for each month of the past 12 months are statistically analyzed, and the median and 90th percentile range are calculated as the baseline (see [link to relevant documentation]). Figure 3 ).For example:

[0143] Compliance baseline: [97%, 99%] (median 98%, lower limit of P90 97%, upper limit of P90 99%).

[0144] Stability P95 baseline: [1.0℃, 1.5℃].

[0145] Overlimit integral baseline: [2.0℃·h, 5.0℃·h].

[0146] Uniformity baseline: [1.5℃, 2.2℃].

[0147] Defrosting peak baseline: [-13.5℃, -12.5℃].

[0148] The baseline is used for setting alarm thresholds for subsequent real-time monitoring and for long-term trend analysis.

[0149] VII. Alarm and Work Order Closed Loop

[0150] The system monitors the temperature in real time, and when an over-temperature event is detected (e.g., the temperature is below -20℃ or above -16℃ at a certain moment), it issues an intelligent alarm based on the event identification results.

[0151] If the over-temperature event occurs within the identified defrosting disturbance segment, it will be recorded as a "defrosting-related event," and only a low-level notification will be sent without triggering a work order.

[0152] If the overheating occurs during the non-defrosting period, further check the door opening / closing signal: if the door remains open for more than 5 minutes, it is determined to be a "door opening disturbance," recorded but an emergency work order may not be triggered; if the door is not open and the equipment is operating abnormally, it is determined to be a "real anomaly," triggering a high-level alarm and generating a maintenance work order.

[0153] Meanwhile, when defrosting-related indicators (such as peak values) exceed the baseline range multiple times consecutively, the system automatically generates a "Defrosting Strategy Optimization Suggestion" work order to remind maintenance personnel to adjust defrosting parameters.

[0154] This closed-loop process enables the differentiation between explainable disturbances and real anomalies, significantly reducing false alarms and improving operational efficiency.

[0155] VIII. Exception and Fault Tolerance Handling

[0156] To address common anomalies encountered in engineering practice, this invention designs the following fault-tolerance mechanism to ensure system robustness:

[0157] Handling defects: If there is missing data (more than 10%) near the start point t0 or end point t1 of a defrosting event, the event is marked as "cannot be denoised", the original temperature is used directly to calculate the index, and a prompt is made in the report.

[0158] Inconsistency among multiple probes: such as Figure 4 As shown, when the temperature difference between the two probes exceeds a threshold (e.g., 3°C), the system prioritizes using the return air vent probe data (because it is closer to the evaporator return air and reflects the overall temperature) for noise reduction and baseline calculation, while recording the anomaly for manual verification. If the probe difference continues to exceed the limit, a probe calibration work order is triggered.

[0159] Sensor drift detection: By comparing the long-term average value with similar probes, if the temperature of a certain probe deviates by more than 1.5℃ over a long period of time, a calibration work order will be triggered.

[0160] Defrost signal jitter: As mentioned above, defrost commands are de-jittered (the status only takes effect after more than 2 sampling points) to avoid misjudgment caused by signal jitter.

[0161] Sampling frequency changes: When the sampling frequency changes due to the accident mode (e.g., from 1 minute to 5 seconds), all window lengths are calculated based on the time length (e.g., 30 minutes corresponds to 1800 seconds), rather than by the number of points, to ensure the adaptability of the algorithm.

[0162] IX. System Implementation Results

[0163] After processing one month's worth of data from the cold storage using the method described in this embodiment, the metrics of the original sequence and the denoised sequence were compared:

[0164] The original sequence compliance rate was 93%, which increased to 98.5% after denoising, indicating that defrosting disturbances have a significant impact on the compliance rate.

[0165] The original sequence had an overlimit integral of up to 15℃·h, which was reduced to 3.5℃·h after denoising. Most of the overlimit was caused by defrosting.

[0166] The number of alarms has decreased from 30 per month to 5, the false alarm rate has dropped significantly, and the operational efficiency has been significantly improved.

[0167] 10. Other Implementation Methods

[0168] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art can make various improvements and modifications without departing from the spirit and principles of the invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. For example:

[0169] The first-order inertial model can also be replaced by other dynamic models, such as a second-order model or a model with time delay, to adapt to the recovery characteristics of different systems.

[0170] The constrained interpolation method can also employ state estimation methods such as Kalman filtering or particle filtering, combining the observations before and after the disturbance to perform optimal estimation.

[0171] The temperature control quality indicators can be added or removed according to the specific application scenario. For example, for pharmaceutical cold chain, indicators such as the number of temperature deviations and the longest continuous temperature exceedance can be added.

[0172] The baseline construction method can also employ machine learning methods, such as training a regression model based on historical data to predict the expected range of each indicator.

[0173] The present invention provides a method, system, and medium for defrosting disturbance denoising and temperature control quality baseline calculation, which has the following advantages compared with the prior art:

[0174] Precise noise reduction while preserving real anomalies: By using dual-constraint interpolation of "pre-disturbance trend + recovery dynamics", abnormal temperature rise caused by defrosting is effectively eliminated, while avoiding simple smoothing or elimination that may mask real anomalies such as doors being open for extended periods or equipment malfunctions, thus ensuring the physical interpretability of the denoised sequence.

[0175] Adaptable to various engineering scenarios: By adopting a dual-path identification strategy of "direct signal connection + morphological inference", defrosting events can be reliably identified at different project locations with or without defrosting control signals, greatly improving the universality and engineering implementation capability of the method.

[0176] Automatic identification of the recovery phase, complete coverage of the disturbance range: Through quantitative indicators such as temperature slope, peak detection, and stability determination, the impact range of defrosting is automatically extended from the disturbance phase to the recovery phase, avoiding the common problem of missing the impact of the recovery period when only the defrosting command period is removed.

[0177] A multi-dimensional quality indicator system provides a comprehensive and objective evaluation: A multi-dimensional indicator system, including compliance rate, stability, over-limit integral, uniformity, and defrosting impact indicators, is constructed on the denoised sequence. This system can comprehensively evaluate temperature control performance from different dimensions, providing data support for refined management and optimization.

[0178] Dynamic baseline, adaptive alarm: The baseline is built based on historical data grouping and statistics, and can be automatically adjusted with changes such as seasons and equipment aging, making the alarm threshold more scientific and reasonable, and avoiding false alarms or missed alarms caused by fixed thresholds.

[0179] Significantly reduce false alarm rate and improve operation and maintenance efficiency: By distinguishing between explainable disturbances (defrosting, door opening) and real anomalies, the number of alarms can be reduced by more than 70% (e.g., from 30 times / month to 5 times / month in the example), enabling operation and maintenance personnel to focus on the real problems that need to be addressed.

[0180] More realistic compliance rate assessment: The compliance rate is calculated on the denoised sequence, eliminating the influence of defrosting disturbances, so that the assessment results can better reflect the temperature control capability of the equipment itself, and avoid the falsely low pass rate caused by defrosting (such as the increase from 93% to 98.5% in the example).

[0181] Guiding defrosting strategy optimization: Through long-term monitoring and baseline comparison of defrosting impact indicators (peak value, duration, disturbance area), the advantages and disadvantages of defrosting strategies can be quantitatively evaluated, providing data basis for optimizing parameters such as defrosting interval and defrosting duration, which helps to save energy and extend equipment life.

[0182] Facilitates acceptance and benchmarking: Standardized quality baseline definitions make temperature control capabilities in different spaces and at different times comparable and traceable, facilitating project acceptance, operational assessment, and continuous improvement.

[0183] Modular design, easy to integrate: The system architecture is clear, and the responsibilities of each module are well-defined. It can be deployed independently or integrated into the existing cold chain monitoring platform, reducing the cost of upgrades and modifications.

[0184] Highly fault-tolerant and robust: Built-in defect handling, multi-probe fault tolerance, and signal jitter reduction mechanisms ensure stable operation under adverse conditions such as sensor failure and communication anomalies.

[0185] Closed-loop management and data-driven decision-making: The entire process from data collection to alarm work order is automated, realizing closed-loop management of "monitoring-analysis-alarm-handling" and transforming data value into actual operation and maintenance actions.

[0186] In summary, this invention effectively solves the problem of defrosting disturbances interfering with temperature control assessment through innovative noise reduction methods and a complete index system. It provides a complete technical solution for accurate monitoring, scientific assessment, and intelligent operation and maintenance of cold chain temperature control, demonstrating significant technological advancements and broad practical value.

[0187] The above embodiments illustrate in detail the specific implementation of the technical solution of the present invention, the logical and connection relationships of each component, and the complete working process. Those skilled in the art will understand that various changes and modifications can be made to the above embodiments without departing from the principles and spirit of the present invention, and all such changes and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for defrosting disturbance denoising and temperature control quality baseline calculation, characterized in that, Includes the following steps: Obtain the original temperature sequence of the target space; Defrost events are identified based on the original temperature sequence, and the time axis is divided into a perturbation segment and a recovery segment according to the defrost events; Based on the temperature data of the stable segment before and after the disturbance segment, the temperature data within the disturbance segment is subjected to constrained interpolation correction to generate a denoised temperature sequence. The temperature control quality index is calculated based on the denoised temperature sequence, and a temperature control quality baseline is constructed.

2. The method according to claim 1, characterized in that, The steps for identifying defrosting events include: Acquire device control signals or operating status data of the target space; When at least one of a defrost command signal, a heating status signal, or a fan start / stop status signal is present, the start and end points of the defrost event are determined directly based on the rising and falling edges of the signal. When the device control signal is missing, the start and end points of the defrosting event are inferred by combining the temperature slope criterion of the original temperature sequence with at least one auxiliary quantity, such as the humidity signal of the target space, the door switch signal, or the evaporator coil temperature signal.

3. The method according to claim 1, characterized in that, The steps of dividing the disturbance segment and the recovery segment include: Mark the moment when defrosting begins or when the temperature slope first exceeds the temperature rise threshold as the start of the disturbance segment t0; The moment when the temperature reaches its peak or the temperature rise slope falls back below the temperature rise threshold is marked as the end of the disturbance segment t1; The moment when the temperature drops back to within the set allowable deviation band and remains stable, or enters a local stable range, is marked as the end point of the recovery segment, t2. The period after t2 is marked as the stable period.

4. The method according to claim 1, characterized in that, The step of performing constrained interpolation correction on the temperature data within the disturbance segment includes: Within the stable segment before the start of the disturbance segment t0, a first window is selected, and the trend prediction function g_pre(t) is obtained by fitting. Within the recovery segment, a second window is selected, and the recovery dynamic parameters are fitted based on the first-order inertial model T(t)=T_inf + (T(t1)-T_inf)·exp(-(t-t1) / τ), where T(t1) is the end temperature of the disturbance segment, τ is the recovery time constant, and T_inf is the steady-state recovery temperature. Based on the trend prediction function g_pre(t) and the recovery dynamics parameters, a continuous and consistent denoised temperature value T_hat(t) is constructed within the disturbance segment [t0, t1] that is consistent with the trend of the recovery segment.

5. The method according to claim 1, characterized in that, The temperature control quality indicators include at least one of the following: compliance rate, stability, over-limit integral, uniformity, or defrosting impact indicators; The compliance rate is the percentage of time the temperature falls within the set allowable deviation band; the stability is the percentage or standard deviation of the temperature deviation from the set value; the over-limit integral is the integral of the magnitude and duration of the temperature deviation from the set allowable deviation band; the uniformity is the maximum difference or percentage difference of the temperature at multiple points in space at the same time; and the defrosting impact index includes the peak temperature of the disturbance segment, the duration, or the area of ​​the disturbance.

6. The method according to claim 1, characterized in that, The steps for establishing a temperature-controlled quality baseline include: The temperature control quality indicators are grouped and statistically analyzed according to spatial identification, season, month, or operating mode. Calculate the median of each set of indicators and the preset percentage range, which will serve as the reference range for the temperature control quality baseline.

7. The method according to claim 1, characterized in that, It also includes alarm and work order closure steps: When an over-temperature event is detected, it is determined whether the over-temperature event occurred within the identified defrosting disturbance segment; If so, the over-temperature event is marked as an explainable disturbance event and recorded or notified according to the preset low-level strategy; If not, the system will determine a genuine anomaly based on the door open / close status or equipment shutdown signal, triggering a high-level alarm and generating a work order.

8. The method according to claim 1, characterized in that, It also includes multi-probe fault tolerance processing steps: When the difference in readings from multiple temperature probes in the same space exceeds a preset threshold, the data from the preferred probe is selected based on a preset confidence rule or the proximity to the return air location for the generation of the denoised temperature sequence and the calculation of the temperature control quality index.

9. A system for defrosting disturbance denoising and temperature control quality baseline calculation, characterized in that, include: The data acquisition module is used to acquire the original temperature sequence and equipment status signals of the target space; An event identification and segmentation module is used to execute the method as described in any one of claims 1 to 3, to identify defrosting events and divide them into disturbance segments and recovery segments; A noise reduction module is configured to perform the method as described in claim 1 or 4 to generate a noise-reduced temperature sequence; The baseline calculation and evaluation module is used to perform the method as described in claim 1, 5 or 6, calculate the temperature control quality index and construct the temperature control quality baseline.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 8.

Images