Intelligent alarm system for abnormal storage and transportation state of hydrogen peroxide based on internet of things

Abstract

The present application relates to the field of alarm technology, in particular to a hydrogen peroxide storage and transportation state abnormal intelligent alarm system based on Internet of Things. The system comprises an Internet of Things collection interface, a working condition segmentation processor, a coupling deviation calculator, a decomposition potential determinator and a graded alarm output device. The Internet of Things collection interface receives the temperature, pressure, liquid level or concentration, environmental temperature and storage and transportation stage data of the hydrogen peroxide storage and transportation object. The working condition segmentation processor generates storage, loading and unloading, transportation, parking exposure and arrival for inspection working condition segments and state baseline. The coupling deviation calculator generates thermal lag deviation, thermal pressure residual error and material stability deviation. The decomposition potential determinator forms a decomposition potential index according to the same direction evolution relationship of the three types of deviation in the same time window. The graded alarm output device outputs abnormal source identification and alarm level. The present application can identify the abnormal decomposition trend of hydrogen peroxide before the fixed threshold is triggered, and distinguish external thermal exposure, collection disturbance and medium state abnormality.

Description

Technical Field

[0001] This invention relates to the field of alarm technology, and more specifically to an intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things. Background Technology

[0002] During storage, loading and unloading, road transportation, exposure during parking, and inspection upon arrival, the state of hydrogen peroxide is affected by factors such as ambient temperature, container pressure, filling volume, material concentration, transportation posture, and residence time. Existing hydrogen peroxide storage and transportation alarm systems typically rely on IoT interfaces to acquire data such as temperature, pressure, liquid level, concentration, or ambient temperature, and upload this data to an alarm platform. The platform then generates alarm information according to preset thresholds, change rate limits, or manually configured alarm rules. In conventional implementations, a temperature alarm is triggered when the temperature exceeds a set upper limit, a pressure alarm is triggered when the pressure exceeds a set upper limit, a material status alarm is triggered when the liquid level drops abnormally or the concentration changes abnormally, and an equipment alarm is triggered when communication is interrupted or data is missing. While this approach has a clear structure and low implementation cost, making it suitable for alerting users to single-state parameters exceeding limits, its judgment relies primarily on independent comparisons of single-point or single-channel data, failing to provide a unified modeling of the correlation between differences in storage and transportation stages, changes in external heat exposure, internal pressure response, and changes in material stability.

[0003] In more conventional technical solutions, to reduce false alarms at single-point thresholds, alarm platforms often incorporate data filtering, moving averages, continuous multi-point confirmation, tiered thresholds, and historical curve comparisons. Specifically, the platform sets continuous sampling windows for temperature, pressure, level, or concentration data, only outputting an alarm when multiple sampling points exceed the threshold or the rate of change exceeds a set range. Different threshold groups can be configured for different stages such as storage, transportation, and loading / unloading. For tanks or vehicles with abundant historical data, the platform can establish an empirical baseline based on curves from similar time periods, using the degree of deviation of the current curve from this baseline as an auxiliary judgment condition. While this approach can filter out some instantaneous jumps and differentiate between certain operating conditions, its core remains the separate limit judgment or empirical comparison for each data acquisition channel. The various state variables typically lack a coupling relationship constrained by the hydrogen peroxide decomposition process. Slow temperature rises caused by external high temperatures, pressure fluctuations caused by transportation disturbances, and level changes caused by loading / unloading switching may still be uniformly categorized as abnormal alarms.

[0004] The main technical problem with existing technologies is that abnormal hydrogen peroxide storage and transportation conditions are not caused by a single independent exceedance of temperature, pressure, or concentration indicators. Instead, they are a continuous coupled evolutionary process formed under specific storage and transportation conditions, involving lag-induced temperature changes, residual pressure accumulation, and gradual deviations in concentration or liquid level. Existing threshold alarms and independent curve comparison methods are insufficient to identify this coupled evolutionary process before traditional alarm thresholds are triggered. When the external environment heats up, the internal temperature will change with lag, but this does not necessarily correspond to the decomposition of internal anomalies. Transportation vibrations, altitude changes, or loading and unloading disturbances may cause short-term pressure fluctuations, but these do not necessarily correspond to media risks. Concentration or liquid level changes have slow response characteristics and, when used alone, are difficult to reflect early exothermic and oxygen release trends in a timely manner. When the alarm system cannot simultaneously distinguish the correspondence between operating condition segments, state baselines, thermal hysteresis deviations, thermo-pressure residuals, and material stability deviations, early anomalies will be weakened by ordinary filtering, and external disturbances may be mistakenly identified as media anomalies, leading to delayed alarm timing and unclear alarm sources. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things, which can solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An IoT-based intelligent alarm system for abnormal hydrogen peroxide storage and transportation status includes an IoT data acquisition interface, a condition segmentation processor, a coupling deviation calculator, a decomposition potential determiner, and a graded alarm output device. The IoT data acquisition interface receives data on the temperature, pressure, liquid level or concentration, ambient temperature, and storage and transportation stages of the hydrogen peroxide being stored and transported. The condition segmentation processor generates condition segments and corresponding state baselines based on storage, loading / unloading, transportation, parking exposure, and arrival-for-inspection status. The coupling deviation calculator generates thermal hysteresis deviation, thermo-pressure residual, and material stability deviation based on the state baselines. The decomposition potential determiner converts the co-evolution relationship of the thermal hysteresis deviation, thermo-pressure residual, and material stability deviation within the same time window into a decomposition potential index. The graded alarm output device outputs alarm information including anomaly source identification and alarm level based on the decomposition potential index.

[0007] Preferably, the operating condition segmentation processor constructs storage and transportation stage tags based on the continuity of the collected timestamps, the position change status, the speed status, the loading and unloading start and stop markers, and the duration of the shutdown. Data with adjacent tags that are consistent and whose state slopes meet the boundary conditions of the same type of operating condition are classified into the same operating condition segment, and a segment start and end identifier and a segment reliability identifier are added to each operating condition segment. The operating condition segmentation processor establishes an ambient temperature response curve, a pressure response curve, and a concentration or liquid level gradual change curve for each operating condition segment, and saves data with abrupt slope changes before and after segment switching as transition segments.

[0008] Preferably, the coupling deviation calculator generates a thermal hysteresis deviation based on the hysteresis difference between the ambient temperature response curve and the hydrogen peroxide temperature sequence within the same operating condition segment, generates a thermo-pressure residual based on the remaining amount between the measured pressure change and the predicted pressure change corresponding to the ambient temperature response curve, generates a material stability deviation based on the continuous deviation amount after removing the loading / unloading transition segment and the segment credibility identifier abnormal segment from the concentration or liquid level gradual change curve, and assigns the same segment number, time window number and data source identifier to the thermal hysteresis deviation, the thermo-pressure residual and the material stability deviation.

[0009] Preferably, the decomposition potential determiner performs trend direction encoding and continuous window matching on the thermal hysteresis deviation, the thermo-pressure residual, and the material stability deviation to form a ternary evolution sequence; when the temperature hysteresis is enhanced and the thermo-pressure residual does not accumulate continuously, the ternary evolution sequence is written into the external thermal exposure channel; when the thermo-pressure residual has an isolated transition and the material stability deviation does not change synchronously, the ternary evolution sequence is written into the acquisition disturbance channel; when the thermal hysteresis deviation, the thermo-pressure residual, and the material stability deviation are all enhanced codes within a continuous time window, a decomposition potential index is formed and submitted to the graded alarm output device.

[0010] Preferably, when establishing a state baseline, the operating condition segmentation processor selects stable sub-windows from each operating condition segment that do not contain loading / unloading markers, communication missing markers, or attitude change markers. The stable sub-windows are combined with data from historical similar operating condition segments that have the same storage and transportation stage tags to form a baseline sample, and each baseline sample is associated with the corresponding segment's trusted identifier. When there is a seasonal shift in the ambient temperature response curve between the current operating condition segment and historical similar operating condition segments, the operating condition segmentation processor stores the baseline samples in layers according to the shift direction.

[0011] Preferably, the coupling deviation calculator performs time window alignment on the sampling sequences of the temperature, pressure, liquid level, or concentration, generates a synchronization segment for sequences with inconsistent sampling intervals according to the most recent valid timestamp, and retains the original sampling time difference in the synchronization segment; within the synchronization segment, the thermal hysteresis deviation is characterized by the displacement and amplitude difference between the peak ambient temperature and the peak hydrogen peroxide temperature, the thermo-pressure residual is characterized by the continuous residual amount after deducting the temperature-driven term from the pressure increment, and the material stability deviation is characterized by the directional consistency of the concentration or liquid level gradual change curve.

[0012] Preferably, the decomposition potential determiner is equipped with a candidate event queue. The candidate event queue records the ternary evolution sequence of the external heat exposure channel, the acquisition disturbance channel, and the decomposition potential channel according to the segment number. Each ternary evolution sequence is configured with an entry time window, an exit time window, and a channel maintenance identifier. The channel maintenance identifier is associated with and stored with the segment credibility identifier. For candidate events that cross transition segments, the decomposition potential determiner retains the trend direction codes before and after the transition segment separately, and merges them into the same decomposition potential index only when both the trend directions before and after the transition segment remain enhanced and the time window numbers are consecutive.

[0013] Preferably, the graded alarm output device is connected to the edge alarm processor and the platform calibration terminal; the edge alarm processor receives the decomposition potential index, candidate event queue, channel hold identifier and anomaly source identifier within the current operating condition segment, generates local alarm information according to individual objects, and writes the segment number, time window number and channel type into the local alarm information; the platform calibration terminal receives the status baseline and decomposition potential index of different batches, different vehicles or different storage tanks within the same operating condition segment, writes the cross-object differences into the baseline update record, and sends the versioned baseline to the edge alarm processor.

[0014] Preferably, the candidate event queue configures the event start point, fragment source, channel type, and number of reliable samples for each ternary evolution sequence; when an isolated transition in the acquisition perturbation channel coincides with a communication missing marker or attitude change marker, the candidate event queue marks the corresponding time window as an isolation window and blocks the thermo-pressure residual within the isolation window from entering the decomposition potential channel; when the thermal hysteresis deviation and material stability deviation on both sides of the isolation window continuously increase, the candidate event queue generates a supplementary verification item containing the verification fragment number and verification deviation type.

[0015] Preferably, the platform calibration terminal sets an object source index, a working condition segment index, and a deviation type index for the versioned baseline; when cross-object differences occur in the same batch or the same parking exposure segment, the platform calibration terminal only updates the ambient temperature response curve, pressure response curve, or concentration or liquid level gradual change curve under the corresponding index; after receiving the updated versioned baseline, the edge alarm processor retains the previous versioned baseline, writes the decomposition potential index obtained from the two versioned baselines within the same time window into the alarm comparison record, and associates the alarm comparison record with the supplementary review item.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention utilizes the collaborative processing of an IoT data acquisition interface, a segmented processing unit, a coupling deviation calculator, a decomposition potential determiner, and a tiered alarm output unit. Abnormal alarms in storage and transportation no longer rely on isolated out-of-range events in temperature, pressure, level, or concentration. Instead, they are based on segmented operating conditions such as storage, loading / unloading, transportation, parking exposure, and arrival for inspection. Environmental temperature response curves, pressure response curves, and gradual change curves for concentration or level are established separately. Then, within the same segment and time window, thermal hysteresis deviation, thermo-pressure residual, and material stability deviation are generated. Thermal hysteresis deviation characterizes the misalignment between environmental temperature changes and the hydrogen peroxide temperature response; thermo-pressure residual characterizes the residual accumulation after deducting temperature-driven factors from pressure changes; and material stability deviation characterizes continuous deviations during gradual changes in concentration or level. When these three types of deviations are configured with the same segment number, time window number, and data source identifier, a traceable ternary evolution sequence can be formed during the decomposition potential determination process. When temperature lags and increases but pressure residual does not accumulate continuously, alarm information can be attributed to external heat exposure sources; when pressure exhibits isolated transitions but material stability deviations do not change synchronously, alarm information can be attributed to data acquisition disturbance sources; only when all three types of deviations show enhanced coding within a continuous time window does a decomposition potential index form and enter the tiered alarm output. Therefore, the alarm triggering basis shifts from single-channel limits to multi-source coupled judgment constrained by storage and transportation conditions. This allows for the identification of the trend chain of abnormal hydrogen peroxide decomposition before temperature or pressure reaches fixed limits, reducing interference from external heating, transportation disturbances, and loading / unloading switching on media risk alarms, ensuring consistent data basis for alarm timing, anomaly sources, and alarm levels.

[0017] 2. This invention, through operating condition segmentation, state baseline, candidate event queue, and edge platform collaborative calibration, also enables a more stable data processing chain in the alarm process. The operating condition segmentation processor classifies data with adjacent tags that are consistent and whose state slope meets the boundary conditions of the same type of operating condition into the same operating condition segment, and saves data with abrupt slope changes before and after segment switching as transition segments, avoiding direct mixing of loading / unloading start / stop, parking restart, and transportation state switching into stable segments. When establishing the baseline, stable sub-windows and historical similar operating condition segments together constitute the baseline sample, and the environmental temperature response curves with seasonal shifts are stored in layers, so that the deviation calculation under similar operating conditions has a reference closer to the current storage and transportation conditions. The candidate event queue records the ternary evolution sequence of the external heat exposure channel, the acquisition disturbance channel, and the decomposition potential channel according to the segment number, and retains the previous and subsequent trend direction encoding for events crossing transition segments. Only when the trend direction continues to enhance and the time window number is continuous are they merged into the same decomposition potential index, reducing the erroneous merging caused by segment boundaries. For isolated transitions that coincide with communication missing markers or attitude change markers, the candidate event queue marks the corresponding time window as an isolation window, blocking the thermo-pressure residual within the isolation window from entering the decomposition potential channel. Simultaneously, supplementary verification terms are generated when the deviation on adjacent sides continues to increase. The edge alarm processor is responsible for generating local alarm information. The platform calibration end forms versioned baselines based on similar operating condition data from different batches, vehicles, or storage tanks, and retains alarm comparison records between the old and new baseline versions, ensuring that subsequent alarm calculations maintain the correspondence between data source, segment number, time window number, and baseline version. Attached Figure Description

[0018] Figure 1 This is an overall alarm flowchart of the intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things of the present invention; Figure 2 This is a flowchart of the working condition segmentation, state baseline construction, and coupling deviation calculation of the present invention; Figure 3 This is a flowchart of the candidate event queue, isolation window processing, and versioned baseline calibration of the present invention. Detailed Implementation

[0019] refer to Figure 1In one embodiment, an IoT-based intelligent alarm system for abnormal hydrogen peroxide storage and transportation status is deployed on the existing storage and transportation data acquisition link between the storage and transportation alarm platform and the edge alarm processing terminal. The data acquisition targets are limited to those directly related to the judgment of abnormal hydrogen peroxide storage and transportation status. The system consists of an IoT acquisition interface, a condition segmentation processor, a coupling deviation calculator, a decomposition potential determiner, and a hierarchical alarm output device. The IoT acquisition interface receives temperature, pressure, liquid level or concentration, ambient temperature, and storage and transportation stage data of the hydrogen peroxide storage and transportation objects during storage, loading and unloading, transportation, parking exposure, and arrival for inspection. It writes a collection timestamp, object identifier, batch identifier, and data source identifier to each acquisition record. The condition segmentation processor does not directly use fixed temperature or pressure thresholds to trigger alarms. Instead of issuing a warning, this embodiment divides continuous data of the same storage and transportation object into operating condition segments with consistent operating states, and generates a state baseline based on each operating condition segment. The coupling deviation calculator extracts thermal hysteresis deviation, thermo-pressure residual, and material stability deviation under the constraints of the state baseline. The decomposition potential arbiter encodes the trend direction and persistence relationship of the three types of deviations within the same time window. Only when the three types of deviations form a continuous evolution relationship of mutual enhancement is a decomposition potential index generated. The graded alarm output device generates alarm information with anomaly source identifier, alarm level, segment number, and time window number based on the decomposition potential index. The advantage of this embodiment is that it transforms the judgment of hydrogen peroxide storage and transportation anomalies from a single state quantity out-of-bounds condition into an alarm judgment process constrained by operating condition segments, state baselines, and multi-source deviations.

[0020] In this embodiment, the IoT acquisition interface interfaces with data. Temperature data includes the temperature of hydrogen peroxide medium or the temperature inside the container; pressure data includes the internal pressure of the container or the process pressure equivalent to the container pressure; liquid level or concentration data is used to represent the quantity and stability changes of the medium; ambient temperature is used to characterize external heat exposure conditions; and storage and transportation stage data is used to determine the storage, loading and unloading, transportation, parking exposure, and arrival inspection status. The IoT acquisition interface performs timestamp normalization processing on data from different sources, marking data with missing timestamps as data that cannot be directly used for trend determination; retaining the latest collected value and recording the overwrite mark for data with duplicate timestamps; and writing the delay mark for data whose communication delay exceeds the normal interval of the sampling sequence of the same object. All marks are not directly used as the basis for hazard alarms, but are used for subsequent segment credibility identification, candidate event isolation, and decomposition potential determination. The advantage of this embodiment is that the alarm input data already has a unified expression of object, time, source, and quality marks before entering the working condition segment, avoiding subsequent algorithms from directly equating abnormal communication status with hydrogen peroxide medium risk.

[0021] refer to Figure 2In one embodiment, after receiving continuous storage and transportation data output from the IoT acquisition interface, the operating condition segmentation processor generates storage and transportation stage tags according to the storage and transportation stage data, location change status, speed status, loading / unloading start / stop markers, and shutdown duration. When the storage and transportation stage tags of the same storage and transportation object are consistent between adjacent sampling points and the temperature slope, pressure slope, liquid level, or concentration slope meets the boundary conditions of the same type of operating condition, the relevant sampling points are assigned to the same operating condition segment. When loading / unloading start / stop, speed status change, shutdown start, arrival for inspection, or timestamp interruption occur at the segment boundary, the operating condition segmentation processor generates a new operating condition segment. The data before and after the segment switch is stored. Data with abrupt slope changes are not incorporated into stable segments but are instead saved as transitional segments. Each operating condition segment is written with a segment start and end identifier, a storage and transportation stage label, a segment credibility identifier, and an object identifier. The segment credibility identifier is jointly determined by communication missing measurement markers, attitude change markers, loading and unloading markers, and sampling continuity. Segments with abnormal segment credibility identifiers are not deleted but have their participation reduced in subsequent baseline construction and deviation calculation. The advantage of this embodiment is that data under storage, loading and unloading, transportation, parking exposure, and arrival inspection states will not be mixed with the same threshold rule, and pseudo-anomalies introduced by segment boundaries are retained as identifiable transitional data.

[0022] Table 1 shows an example of field configuration for operating condition segments, illustrating how storage and transportation data are organized before deviation calculation.

[0023] Table 1 Example of working condition segment field configuration

[0024] In this embodiment, the operating condition segmentation processor generates a state baseline for each operating condition segment. The state baseline includes an ambient temperature response curve, a pressure response curve, and a concentration or liquid level gradual change curve. The ambient temperature response curve is used to characterize the normal driving relationship between external temperature changes and hydrogen peroxide temperature sequence under similar operating conditions. The pressure response curve is used to characterize the normal response range of pressure with temperature, liquid level, or concentration changes under similar operating conditions. The concentration or liquid level gradual change curve is used to characterize the material stability change trajectory after excluding loading and unloading transition segments. The state baseline does not use a globally fixed curve, but selects stable sub-windows without loading and unloading markers, communication missing measurement markers, and attitude change markers from the current operating condition segment. These stable sub-windows are combined with data consistent with the storage and transportation stage labels in historical similar operating condition segments to form a baseline sample. When there is a seasonal offset between the ambient temperature response curve of the current operating condition segment and historical similar operating condition segments, the baseline sample is stored in layers according to the offset direction. Subsequent deviation calculations read reference curves from samples of the same operating condition, the same offset layer, and the same object type. The advantage of this embodiment is that the deviation calculation has a reference basis corresponding to the current storage and transportation operating condition, avoiding the use of static storage baselines to interpret loading and unloading transition data or shutdown exposure data.

[0025] In one embodiment, the operating condition segmentation processor quantifies the matching degree of operating condition segments, using storage and transportation stage tags, state slope, and sampling continuity to jointly determine segment attribution. The operating condition segment matching quantity is denoted as... The expression is: ; in, Indicates the first The matching quantity of each sampling point and the previous sampling point belonging to the same working condition segment. Indicates the first Labels for the storage and transportation stages of each sampling point. This indicates that the value is 1 when the labels are the same and 0 when the labels are different. Indicates the first The position change or velocity state of each sampling point Indicates the first The overall state slope of each sampling point is composed of temperature, pressure, liquid level, or concentration slope. and This represents a normalized reference value for similar operating conditions. , and Indicates the type of working condition The label weight, motion state weight, and slope weight are given, and satisfy the following conditions: For example, taking samples under transportation conditions. , , If adjacent sampling points have the same label, the velocity state difference is 2, and the reference quantity is... The overall slope difference is 0.3, and the reference quantity is... ,but When the boundary conditions of the same working condition require the matching quantity to be no less than the segment belonging threshold, adjacent sampling points are merged into the same working condition segment. The advantage of this embodiment is that the segment division retains the stage label, motion state and state slope information at the same time, reducing the missegmentation caused by the change of a single label.

[0026] In one embodiment, the coupling deviation calculator aligns the sampling sequence within the same operating condition segment using time windows. When the sampling intervals for temperature, pressure, liquid level, or concentration are inconsistent, the coupling deviation calculator generates a synchronization segment based on the most recent valid timestamp and retains the original sampling time difference within the synchronization segment. Data exceeding the acceptable time difference is marked as late data and temporarily excluded from the same-direction evolution determination within the same time window. The thermal hysteresis deviation is generated based on the peak displacement and amplitude difference between the ambient temperature response curve and the hydrogen peroxide temperature sequence. The thermo-pressure residual is generated based on the remaining amount after deducting the temperature-driven term and liquid level change term from the measured pressure change. The material stability deviation is generated based on the continuous deviation amount after removing the loading / unloading transition segment and the segment credibility identifier abnormal segment from the concentration or liquid level gradual change curve. All three types of deviations are configured with the same segment number, time window number, and data source identifier. Subsequent decomposition potential determination only compares deviation data with the same segment number and the same time window number. The advantage of this embodiment is that the state quantities formed under different sampling frequencies are unified within the same time window framework, while retaining the original time difference for anomaly source identification.

[0027] In this embodiment, the thermal hysteresis deviation is denoted as This is used to characterize the degree of misalignment between the ambient temperature response curve and the hydrogen peroxide temperature sequence within the same operating condition segment. The formula for calculating the thermal hysteresis deviation is: ; in, Indicates the first Thermal hysteresis deviation corresponding to each synchronous sampling point This represents the measured temperature of the hydrogen peroxide solution. This represents the reference value for hydrogen peroxide temperature obtained from the ambient temperature response curve under the same operating condition baseline. Indicates the type of working condition The discrete quantity of the lower temperature reference curve. This represents the observed displacement from the peak ambient temperature to the peak hydrogen peroxide temperature. Indicates the type of working condition The baseline displacement below, Indicates the magnitude deviation weight. This indicates a stable term to prevent the denominator from being zero. The deviation direction is not used in this formula but is preserved by the difference direction, for example, in parking exposure segments. , , , , , , hour, This indicates that the current temperature response has a positive hysteresis deviation relative to the baseline. The advantage of this embodiment is that the thermal hysteresis deviation includes both temperature amplitude deviation and thermal response time deviation, which can distinguish between the normal temperature rise caused by external heat exposure and the medium temperature response that does not conform to the baseline.

[0028] In one embodiment, the hot-pressing residual is denoted as This is used to characterize the residual accumulation in pressure changes that cannot be explained by temperature and level changes, and is calculated as follows:

[0029] in, Indicates the first Thermo-pressure residuals corresponding to each synchronous sampling point Indicates the increase in pressure. This represents the measured pressure value. This indicates the temperature increment of hydrogen peroxide. Indicates the liquid level increment. Indicates the type of working condition The baseline coefficient of temperature change versus pressure change. Indicates the type of working condition The baseline coefficient of pressure change for liquid level change; the minus sign indicates that the pressure driving term that can be explained by temperature and liquid level changes is subtracted from the measured pressure increment, such as under storage conditions. , , , , hour, This indicates that there is a positive residual in the pressure change, which is significant in the transport vibration segment. Short-term jumps occur with adjacent time windows If the positive trend cannot be maintained continuously, the decomposition potential detector will not directly classify it into the decomposition potential channel. The advantage of this embodiment is that the pressure alarm is no longer based on the absolute value of the pressure or a single increment, but on the remaining change after excluding the normal influence of temperature and liquid level.

[0030] In one embodiment, the material stability deviation is denoted as... When concentration data is available, the concentration gradient curve is used as the primary data source; when concentration data is unavailable but liquid level data is continuous, the liquid level gradient curve is used as the primary data source; when both types of data are available, they are weighted according to their reliability. The calculation formula is as follows: ; in, Indicates the first Material stability deviation corresponding to each synchronous sampling point This represents the measured concentration value. This represents the concentration reference value under the same operating condition baseline. This represents the measured liquid level. This represents the liquid level reference value under the same operating condition baseline. This represents the dispersion of the concentration reference curve. This represents the discrete value of the liquid level reference curve. This represents the confidence weight of the concentration channel and its value ranges from 0 to 1. Indicates the confidence weight of the liquid level channel, for example, within the segment to be inspected at the arrival station. , , , , , hour, This indicates that the material stability deviates negatively from the baseline. The decomposition potential determiner retains the continuity of this negative deviation when encoding the trend direction. The advantage of this embodiment is that when the concentration response period is long or the liquid level is affected by loading and unloading, the material stability can still be kept in a computable state through the reliable weight and the gradual change curve.

[0031] In one embodiment, after receiving thermal hysteresis deviation, thermo-pressure residual, and material stability deviation, the decomposition potential determiner performs trend direction encoding on the deviation value within each time window. The trend direction encoding includes enhancement encoding, weakening encoding, stabilization encoding, and invalid encoding. The enhancement encoding is jointly determined by the directional change of the current deviation value relative to the deviation value of the previous time window, the duration of the continuous window, and the segment credibility identifier. The decomposition potential determiner forms a ternary evolution sequence of thermal hysteresis deviation, thermo-pressure residual, and material stability deviation with the same segment number, the same time window number, and the same traceable data source link. When the temperature hysteresis is enhanced and the thermo-pressure residual does not accumulate continuously, the ternary evolution sequence is written into the external thermal exposure channel. When the thermo-pressure residual has an isolated transition and the material stability deviation does not change synchronously, the ternary evolution sequence is written into the acquisition disturbance channel. When the thermal hysteresis deviation, thermo-pressure residual, and material stability deviation are all enhanced codes within a continuous time window, the ternary evolution sequence is written into the decomposition potential channel and forms a decomposition potential index. The advantage of this embodiment is that the alarm source is determined by the continuous relationship between deviations, and external thermal exposure, acquisition disturbance, and media decomposition trend are assigned to different channels.

[0032] In this embodiment, the decomposition potential index is defined as follows: , used to indicate the first The degree of anomalous evolution formed by the three types of deviations within a time window is calculated using the following formula: ; in, Indicates the first Decomposition potential indicators for each time window Indicates the first The set of synchronous sampling points within a time window Represents a set Number of sampling points , and These represent the combined weights of thermal hysteresis bias, thermocompression residual, and material stability bias, respectively, with the sum of the three being 1. Represents the compression function. This indicates that the absolute value of the material stability deviation is taken, and the deviation range is retained. This indicates a co-evolutionary indicator, when the three types of bias are at the sampling point. The value is 1 if all values ​​within the continuous window are enhanced codes, and 0 otherwise. For example, within a certain time window... Both points satisfy the same direction of evolution, the first point , , Second point , , Weighting , , The first contribution is The second contribution is ,get If any point does not satisfy the same direction of evolution, then the corresponding Taking 0 does not contribute to the decomposition potential index. The advantage of this embodiment is that the decomposition potential index is formed by the deviation amplitude and the same direction evolution condition, which avoids directly triggering the medium risk alarm due to a single excessive deviation.

[0033] In one embodiment, the graded alarm outputter generates alarm information based on the decomposition potential index, anomaly source identifier, channel type, and segment credibility identifier. The alarm information includes at least the object identifier, batch identifier, operating condition segment number, time window number, channel type, decomposition potential index, alarm level, and anomaly source identifier. The alarm level is not determined solely by a fixed temperature or pressure value, but is jointly determined by the continuous holding status of the decomposition potential channel, the duration of the external heat exposure channel, and the isolation status of the acquisition disturbance channel. Local alarm information is generated by the edge alarm processor and stored in the event record corresponding to the storage and transportation object. The platform synchronously receives the alarm information, as well as the deviation sequence and baseline version number used for recalculation. When outputting an alarm, candidate events in the external heat exposure channel or acquisition disturbance channel are not deleted; instead, the relevant candidate events are stored as evidence fields for the alarm source identifier. The advantage of this embodiment is that the alarm level and anomaly source can be traced back to the segment number, time window number, deviation sequence, and channel type, facilitating subsequent verification of whether the same alarm originates from media risk or data disturbance.

[0034] In one embodiment, the establishment of the state baseline is jointly accomplished by stable sub-window selection, historical segment matching, and seasonal offset stratification. When selecting stable sub-windows, time windows with loading / unloading markers, communication missing measurement markers, and attitude change markers are excluded, and it is required that the adjacent change directions of temperature, pressure, liquid level, or concentration do not frequently reverse within the stable sub-window. When matching historical segments, only data with consistent labels during storage and transportation, consistent object types, associative batch attributes, and normal segment reliable identification are selected. When stratifying seasonal offsets, the original collected values ​​are not changed, and only the reference layer of the environmental temperature response curve is grouped, so that reference curves can be established separately for summer high temperature exposure, nighttime parking exposure, and ordinary storage segments. The platform writes an object source index, operating condition segment index, and deviation type index for each state baseline. When the edge calls the state baseline, it must match all three types of indexes at the same time. The advantage of this embodiment is that the baseline update will not mix data from different operating conditions, different environmental exposure layers, and different deviation types into the same reference.

[0035] In a preferred embodiment, when the amount of data in a current operating condition segment is insufficient to independently establish a stable baseline, the operating condition segment processor reads data consistent with the current storage and transportation stage label from historical similar operating condition segments as supplementary baseline samples, and uses the confirmed reliable short-term stable sub-windows within the current segment as correction samples. The supplementary baseline samples and correction samples are stored separately. The coupling deviation calculator records the source ratio of the samples used when calculating thermal hysteresis deviation, thermo-pressure residual, and material stability deviation. If the subsequent data of the current segment recovers to stability, the operating condition segment processor adds the new stable sub-window to the baseline sample pool of the current segment, while retaining the version identifier of the original supplementary baseline sample. After receiving baseline samples from multiple storage and transportation objects, the platform only updates the corresponding baseline under similar operating condition segments, similar ambient temperature response layers, and similar deviation types. The advantage of this embodiment is that short data segments can also enter the coupling deviation calculation process, and the unclear baseline source is avoided by using the sample source ratio and version identifier.

[0036] In one embodiment, the coupling deviation calculator performs directional consistency processing on the data after time window alignment. When a late data occurs in a channel of temperature, pressure, liquid level, or concentration, the late data does not participate in the ternary evolution determination of the current time window, but can enter the candidate event queue as supplementary data in the next time window. When a channel is missing, the corresponding items of thermal hysteresis deviation, thermo-pressure residual, or material stability deviation are marked as invalid codes. When the decomposition potential determiner encounters an invalid code in the same time window, it does not generate a decomposition potential index, but only writes the existing deviation into the candidate event queue. If subsequent consecutive time windows fill in the data and the trend direction before and after remains enhanced, the candidate event queue can merge the trend direction codes before and after the invalid code into a verification event. The merging process retains the original sampling time difference and the late mark. The advantage of this embodiment is that the alarm system will not discard all abnormal evolution information due to a single communication delay, nor will it construct a false co-evolution relationship with missing data.

[0037] refer to Figure 3 In this embodiment, the candidate event queue records the ternary evolution sequence of the external heat exposure channel, the acquisition disturbance channel, and the decomposition potential channel according to the segment number. Each ternary evolution sequence is configured with an entry time window, an exit time window, a channel holding identifier, a segment credibility identifier, an event start point, a channel type, and a number of credibility samples. The external heat exposure channel saves events where the temperature hysteresis increases but the thermo-pressure residual does not accumulate continuously. The acquisition disturbance channel saves events where the thermo-pressure residual transitions in isolation but the material stability deviation does not change synchronously. The decomposition potential channel saves events where the three types of deviations increase continuously. For candidate events that cross transition segments, the candidate event queue retains the trend direction encoding before and after the transition segment. When the trend direction before and after both segments increases and the time window numbers are continuous, they are merged into the same decomposition potential index. If the trend direction before and after the transition segment is inconsistent, they are retained as two independent candidate events and written into different channels. The advantage of this embodiment is that the segment boundary generated near the loading / unloading switch or shutdown restart will not be directly interpreted as continuous medium risk.

[0038] Table 2 shows an example of channel records for the candidate event queue, illustrating the diversion method of the deviation sequence before alarm output.

[0039] Table 2. Example of channel records for the candidate event queue.

[0040] In one embodiment, the candidate event queue performs isolation window processing on isolated transitions in the acquisition disturbance channel. When an isolated transition of thermo-pressure residual is located in the same time window as a communication missing measurement marker, a communication delay marker, or a change in attitude marker, the candidate event queue marks this time window as an isolation window. The thermo-pressure residual within the isolation window does not enter the decomposition potential channel. The isolation window does not delete the original pressure data, but instead writes the pressure source identifier, communication status marker, and attitude status marker into the acquisition disturbance channel. If the thermal hysteresis deviation and material stability deviation on both sides of the isolation window continue to increase, the candidate event queue generates a supplementary verification item. The supplementary verification item includes the verification segment number, verification deviation type, isolation window position, and adjacent trend direction code. When the edge alarm processor receives the supplementary verification item, it does not directly change the original alarm level, but saves the supplementary verification item and the local alarm information side by side. The advantage of this embodiment is that the acquisition disturbance will not be directly superimposed into the decomposition potential, and the abnormal evolution that continues to exist on both sides of the isolation window will not be filtered out.

[0041] In one embodiment, the hierarchical alarm output device connects the edge alarm processor and the platform calibration end. The edge alarm processor is deployed on the data processing side close to the storage and transportation object. It is used to receive the decomposition potential index, candidate event queue, channel holding identifier, and anomaly source identifier within the current operating condition segment, and generate local alarm information for each individual object. The local alarm information is written with the segment number, time window number, channel type, decomposition potential index, and data source identifier. The platform calibration end receives the status baseline and decomposition potential index of different batches, different vehicles, or different storage tanks within the same operating condition segment, and writes the cross-object differences into the baseline update record. After the platform calibration end forms a versioned baseline, it sends it to the edge alarm processor. After receiving the versioned baseline, the edge alarm processor does not overwrite the local old baseline, but retains the previous version baseline for alarm comparison within the same time window. The advantage of this embodiment is that local alarms can be completed at the edge, and the platform end can maintain the consistency of reference for the same operating condition through cross-object baseline calibration.

[0042] In a preferred embodiment, the platform calibration terminal sets an object source index, a working condition segment index, and a deviation type index for the versioned baseline. The object source index is used to distinguish different storage tanks, different vehicles, or different containers. The working condition segment index is used to distinguish storage, loading and unloading, transportation, parking exposure, and arrival inspection segments. The deviation type index is used to distinguish ambient temperature response curves, pressure response curves, and concentration or liquid level gradual change curves. When cross-object differences occur concentrated in the same batch, the platform calibration terminal only updates the curves under the corresponding batch and the corresponding deviation type index. When cross-object differences occur concentrated in the same parking exposure segment, the platform calibration terminal only updates the curves under the corresponding parking exposure segment and the ambient temperature response curve index. When the difference sources are not associated with the same batch or the same segment, the platform calibration terminal only saves the difference record and does not issue the versioned baseline. The advantage of this embodiment is that the baseline update range is limited to the indexes with data association, avoiding abnormal fluctuations of a single object from polluting other working condition references.

[0043] In one embodiment, after receiving the updated versioned baseline, the edge alarm processor calculates thermal hysteresis deviation, thermocompression residual, material stability deviation, and decomposition potential index for data within the same time window using both the previous and new version baselines. The decomposition potential indexes under both versions are then written into the alarm comparison record. The alarm comparison record includes the object identifier, operating condition segment number, time window number, old baseline version number, new baseline version number, two sets of decomposition potential indexes, channel type, and supplementary review item identifier. If a decomposition potential channel is formed under both the old and new version baselines, the edge alarm processor retains the alarm source identifier for the decomposition potential channel. If a decomposition potential channel is formed only under the new version baseline, the edge alarm processor associates the alarm comparison record with the supplementary review item. If a decomposition potential channel is formed only under the old version baseline, the edge alarm processor retains the old baseline calculation result and reports it to the platform calibration end for review. The advantage of this embodiment is that the versioned baseline update does not cause a break in the alarm record, and the alarm calculation process can retain evidence of differences between the old and new references.

[0044] In one embodiment, the system separately identifies external heat exposure sources. Events in the external heat exposure channel require the ambient temperature response curve to show a continuous upward trend or remain at a high level, the hydrogen peroxide temperature sequence to show a baseline-compliant hysteresis change relative to the ambient temperature, the thermo-pressure residual not to show continuous positive accumulation, and the material stability deviation not to show a continuous increase in the same direction as the thermal hysteresis deviation. When generating external heat exposure alarm information, the graded alarm output device writes the ambient temperature response curve level, the thermal hysteresis deviation sequence, the thermo-pressure residual continuity marker, and the material stability deviation status. If the thermo-pressure residual begins to accumulate continuously in a subsequent time window and the material stability deviation increases synchronously, the candidate event queue copies the event from the external heat exposure channel to the decomposition potential channel and retains the original external heat exposure event starting point. The advantage of this embodiment is that the external heating process can be continuously tracked, and the medium decomposition alarm will not be directly triggered when there is a lack of pressure residual accumulation and material stability deviation.

[0045] In one embodiment, the system separately identifies the sources of acquisition disturbances. Events in the acquisition disturbance channel include isolated pressure jumps, delayed timestamps, communication loss, sudden attitude changes, or short-term liquid level changes near the loading / unloading boundary. When processing the acquisition disturbance channel, the decomposition potential determiner does not use simple filtering to delete the original data, but retains the original sampled values, delayed markers, isolation window identifiers, and forward and backward trend direction codes. Thermo-pressure residuals are blocked from entering the decomposition potential channel within the isolation window, but can be used as the basis for acquisition link verification and written into alarm information. When the liquid level or concentration changes abruptly in the loading / unloading transition segment, the material stability deviation is not included in the current stable segment, but is stored separately as transition segment data. If the liquid level or concentration gradual change curve continues to deviate from the baseline after the transition segment ends, the coupling deviation calculator recalculates the material stability deviation in the new stable segment. The advantage of this embodiment is that acquisition link anomalies and medium state anomalies are carried by different data channels, avoiding ordinary filtering from weakening the true trend or single-point jumps triggering high-level alarms.

[0046] In one embodiment, the alarm processing in the arrival-to-inspection scenario adopts the continuous relationship between the parking exposure segment and the arrival-to-inspection segment. The working condition segment processor generates the arrival-to-inspection segment after the vehicle arrives at the station and establishes an adjacency relationship between the end time window of the parking exposure segment and the start time window of the arrival-to-inspection segment. If there is an external heat exposure channel candidate event in the parking exposure segment, the thermal hysteresis deviation, thermo-pressure residual, and material stability deviation in the arrival-to-inspection segment are still calculated according to the new state baseline. The candidate event queue retains the starting point of the previous segment event and the trend direction code of the new segment. Only when the three types of deviations continue to increase in the new segment and the time window numbers are continuous, the decomposition potential determiner merges them into the same decomposition potential index. If the thermo-pressure residual in the arrival-to-inspection segment fades or the material stability deviation recovers to near the baseline, the previous segment event remains the source of external heat exposure. The advantage of this embodiment is that cross-segment risks will not be directly interrupted due to segment switching, nor will the alarm level be automatically increased in the arrival-to-inspection stage due to the existence of heat exposure records in the previous segment.

[0047] In one embodiment, the alarm processing in the transportation scenario jointly constrains the speed state and attitude change marker. The working condition segment processor divides continuous driving, low-speed congestion, parking exposure, and loading / unloading transition into different working condition segments. When transportation vibration causes a short-term pressure jump, the coupling deviation calculator still calculates the thermo-pressure residual, but the candidate event queue marks the time window that coincides with the attitude change marker as an isolation window. If the thermo-pressure residual no longer maintains positive accumulation after the isolation window, the decomposition potential arbiter leaves the event in the acquisition disturbance channel. If the thermal hysteresis deviation and material stability deviation before and after the isolation window maintain the same direction of enhancement within the continuous time window, the candidate event queue generates a supplementary verification item and submits it to the edge alarm processor. The edge alarm processor writes the verification segment number, verification deviation type, and channel type into the local alarm information. The advantage of this embodiment is that the pressure spike caused by transportation vibration will not directly trigger the decomposition potential alarm, while the continuous abnormal trend near the pressure spike is still recorded.

[0048] In one embodiment, alarm processing in the storage scenario primarily relies on stable segments and historical baselines of similar operating conditions. The operating condition segmentation processor forms storage segments when the storage is continuously static and the positional change state is stable. The state baseline is read from data of the same storage area, the same ambient temperature response layer, and the same object type. The coupling deviation calculator focuses on calculating the continuous relationship between thermal hysteresis deviation and thermo-pressure residual in the storage segment. If the ambient temperature response curve is stable but the hydrogen peroxide temperature sequence rises and the pressure residual continuously accumulates, the decomposition potential determiner writes the ternary evolution sequence into the decomposition potential channel. If the ambient temperature response curve rises with diurnal variation and the pressure residual does not accumulate, the ternary evolution sequence is written into the external heat exposure channel. If the liquid level slow change curve shows continuous deviation in the absence of loading / unloading marks, the material stability deviation enters the ternary evolution sequence to participate in subsequent determination. The advantage of this embodiment is that the slow changes during static storage can be expressed through baseline deviation, rather than being identified only after the temperature or pressure reaches a fixed alarm limit.

[0049] In a preferred embodiment, the loading and unloading process is not used as the main segment for determining stable decomposition potential, but is stored separately as a transition segment. The operating condition segment processor generates a loading and unloading transition segment when it detects loading and unloading start / stop markers, rapid changes in liquid level, and short-term pressure fluctuations. The coupling deviation calculator only saves the original direction and duration window of changes in liquid level or concentration in the loading and unloading transition segment, and does not include them in the stability deviation calculation of the concentration or liquid level gradual change curve. When entering a new storage segment, transportation segment, or arrival inspection segment after loading and unloading, the state baseline is recalculated from the starting point of the new operating condition segment. If the liquid level or concentration continues to deviate from the baseline after the loading and unloading transition segment ends, the material stability deviation is generated from the new stable segment and participates in the ternary evolution sequence. The candidate event queue saves the loading and unloading transition segment number as the source field. The advantage of this embodiment is that the necessary liquid level changes caused by loading and unloading will not directly enter the medium decomposition determination, and the abnormal deviations that continue to exist after loading and unloading can still be included in the subsequent alarm chain.

[0050] In one embodiment, alarm information generation adopts a hierarchical output structure. Local alarm information for individual objects is generated by the edge alarm processor according to object identifier, segment number, time window number, and channel type. Alarm information corresponding to the external thermal exposure channel records the thermal hysteresis deviation and ambient temperature response curve hierarchy. Alarm information corresponding to the acquisition disturbance channel records the isolation window, communication missing measurement marker, attitude change marker, and pressure source identifier. Alarm information corresponding to the decomposition potential channel records the trend direction encoding of three types of deviations, decomposition potential index, and baseline version number. After receiving local alarm information, the platform calibration end searches for similar deviation sequences of the same type of object based on the batch identifier, object source index, and operating condition segment index. If similar deviation sequences exist, the platform calibration end writes the cross-object differences into the baseline update record. If similar deviation sequences do not exist, the platform calibration end only saves the individual object alarm history. The advantage of this embodiment is that the alarm output content can simultaneously cover the data fields required for individual object processing and platform baseline calibration.

[0051] In one embodiment, the system's working principle can be summarized as follows: using operating condition segments as data interpretation boundaries, state baselines as deviation references, thermal hysteresis deviation, thermo-pressure residuals, and material stability deviations as anomaly characteristics, ternary evolution sequences as the basis for determining decomposition potential, candidate event queues as the diversion carriers for external thermal exposure, acquisition disturbances, and decomposition potential, and edge alarm processors and platform calibration terminals as the collaborative processing terminals for alarm output and baseline updates. After data enters from the IoT acquisition interface, it does not directly trigger a danger alarm, but instead forms a closed loop through segment division, baseline generation, deviation calculation, trend encoding, channel attribution, and hierarchical output. Any channel anomaly can retain the original sampling source and calculation source, and any alarm information can be traced back to the object identifier, batch identifier, segment number, time window number, deviation type, and baseline version. The advantage of this embodiment is that early anomaly decomposition trends, external thermal exposure, and acquisition disturbances are processed separately within the same data processing framework, and the alarm results are supported by a continuous data chain.

[0052] In one embodiment, the system maintains an alarm sample library during long-term operation. The alarm sample library stores historical operating condition segments, state baselines, deviation sequences, ternary evolution sequences, candidate event queues, decomposition potential indicators, alarm comparison records, and labels of results after manual review. The manual review results do not directly overwrite the original alarm calculation results, but serve as additional constraints when the platform calibration end updates the versioned baseline. If the review results confirm that a certain batch or a certain parking exposure segment has the same type of deviation, the platform calibration end updates the corresponding curves according to the object source index, operating condition segment index, and deviation type index. If the review results confirm that a certain alarm comes from communication missing measurement or attitude change, the platform calibration end adds isolation window identification records for the same type of acquisition disturbance channel. When the edge alarm processor receives the updated versioned baseline, it retains the previous version and generates alarm comparison records. The advantage of this embodiment is that historical alarm data can serve as a constraint source for baseline calibration and disturbance identification, while not destroying the traceability of the original alarm data.

Claims

1. An intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things, characterized in that, This includes an IoT data acquisition interface, a working condition segmentation processor, a coupling deviation calculator, a decomposition potential determiner, and a graded alarm output device. The IoT data acquisition interface receives data on the temperature, pressure, liquid level or concentration, ambient temperature, and storage and transportation stage of the hydrogen peroxide storage and transportation object. The operating condition segmentation processor generates operating condition segments and corresponding status baselines based on storage, loading and unloading, transportation, parking exposure, and arrival inspection status. The coupling deviation calculator generates thermal hysteresis deviation, thermocompression residual, and material stability deviation based on the state baseline; The decomposition potential determiner converts the co-evolution relationship of the thermal hysteresis deviation, the thermocompression residual, and the material stability deviation within the same time window into a decomposition potential index. The graded alarm output device outputs alarm information containing anomaly source identification and alarm level based on the decomposition potential index.

2. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 1, characterized in that, The working condition segmentation processor constructs storage and transportation stage labels based on the continuity of the collected timestamps, the position change status, the speed status, the loading and unloading start and stop markers, and the duration of the stop. It classifies data with adjacent labels that are consistent and whose state slopes meet the boundary conditions of the same type of working condition into the same working condition segment, and adds segment start and end identifiers and segment trust identifiers to each working condition segment. The operating condition segmentation processor establishes an ambient temperature response curve, a pressure response curve, and a concentration or liquid level gradual change curve for each operating condition segment, and saves data with abrupt slope changes before and after segment switching as transition segments.

3. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 2, characterized in that, The coupling deviation calculator generates a thermal hysteresis deviation based on the hysteresis difference between the ambient temperature response curve and the hydrogen peroxide temperature sequence within the same operating condition segment. It generates a thermo-pressure residual based on the remaining amount between the measured pressure change and the predicted pressure change corresponding to the ambient temperature response curve. It generates a material stability deviation based on the continuous deviation amount after removing the loading / unloading transition segment and the segment credibility identifier abnormal segment from the concentration or liquid level gradual change curve. The thermal hysteresis deviation, the thermo-pressure residual, and the material stability deviation are assigned the same segment number, time window number, and data source identifier.

4. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 3, characterized in that, The decomposition potential determiner performs trend direction encoding and persistence window matching on the thermal hysteresis deviation, the thermocompression residual, and the material stability deviation to form a ternary evolution sequence. When the temperature hysteresis increases and the thermal pressure residual does not accumulate continuously, the ternary evolution sequence is written into the external thermal exposure channel; When the hot-pressing residual has an isolated transition and the material stability deviation does not change synchronously, the ternary evolution sequence is written into the acquisition disturbance channel; When the thermal hysteresis deviation, the thermocompression residual, and the material stability deviation are all enhanced codes within a continuous time window, a decomposition potential index is formed and submitted to the graded alarm output device.

5. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 2, characterized in that, When establishing a state baseline, the working condition segment processor selects stable sub-windows from each working condition segment that do not contain loading / unloading markers, communication missing markers, or attitude change markers. The stable sub-windows are combined with data from historical similar working condition segments that are consistent with the storage and transportation stage tags to form a baseline sample. Each baseline sample is then associated with the corresponding segment's trusted identifier. When there is a seasonal shift in the ambient temperature response curve between the current operating condition segment and the historical operating condition segment of the same type, the operating condition segmentation processor stores the baseline sample in layers according to the shift direction.

6. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 3, characterized in that, The coupling deviation calculator performs time window alignment on the sampling sequences of the temperature, pressure, liquid level or concentration, generates synchronization segments for sequences with inconsistent sampling intervals according to the most recent valid timestamp, and retains the original sampling time difference in the synchronization segments; Within the synchronization segment, the thermal hysteresis deviation is characterized by the displacement and amplitude difference between the peak ambient temperature and the peak hydrogen peroxide temperature; the thermo-pressure residual is characterized by the continuous residual amount after deducting the temperature-driven term from the pressure increment; and the material stability deviation is characterized by the directional consistency of the concentration or liquid level gradual change curve.

7. The IoT-based intelligent alarm system for abnormal hydrogen peroxide storage and transportation status as described in claim 4, characterized in that, The decomposition potential determiner is equipped with a candidate event queue. The candidate event queue records the ternary evolution sequence of the external heat exposure channel, the acquisition disturbance channel and the decomposition potential channel according to the segment number. Each ternary evolution sequence is configured with an entry time window, an exit time window and a channel holding identifier. The channel holding identifier is associated with the segment trust identifier and stored. For candidate events that span transition segments, the decomposition potential determiner retains the trend direction codes before and after the transition segment separately, and merges them into the same decomposition potential index only when both the trend directions before and after the transition segment remain enhanced and the time window numbers are consecutive.

8. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 7, characterized in that, The graded alarm output device is connected to the edge alarm processor and the platform calibration terminal; The edge alarm processor receives the decomposition potential index, candidate event queue, channel holding identifier and anomaly source identifier within the current operating condition segment, generates local alarm information according to individual objects, and writes the segment number, time window number and channel type into the local alarm information; The platform calibration terminal receives the status baselines and decomposition potential indicators of different batches, different vehicles or different storage tanks within the same operating condition segment, writes the cross-object differences into the baseline update record, and sends the versioned baseline to the edge alarm processor.

9. The intelligent alarm system for abnormal storage and transportation status of hydrogen peroxide based on the Internet of Things as described in claim 8, characterized in that, The candidate event queue configures the event start point, fragment source, channel type, and number of reliable samples for each ternary evolution sequence; When an isolated transition in the acquisition disturbance channel coincides with a communication missing measurement marker or attitude change marker, the candidate event queue marks the corresponding time window as an isolation window and blocks the thermo-pressure residual in the isolation window from entering the decomposition potential channel. When the thermal hysteresis deviation and material stability deviation on both sides of the isolation window continuously increase, the candidate event queue generates supplementary review items containing review segment numbers and review deviation types.

10. The IoT-based intelligent alarm system for abnormal hydrogen peroxide storage and transportation status as described in claim 9, characterized in that, The platform calibration terminal sets the object source index, operating condition segment index, and deviation type index for the versioned baseline; When cross-object differences occur in the same batch or the same parking exposure segment, the platform calibration terminal only updates the ambient temperature response curve, pressure response curve, or concentration or liquid level gradual change curve under the corresponding index. After receiving the updated versioned baseline, the edge alarm processor retains the previous versioned baseline, writes the decomposed potential index obtained from the two versioned baselines within the same time window into the alarm comparison record, and associates the alarm comparison record with the supplementary review item.

Images