A method and system for accounting for a dynamically adjusted regional water budget benchmark quota

Abstract

The application discloses a kind of dynamically regulated regional water budget benchmark quota accounting method and system, is specifically related to water resource measurement accounting field, including obtaining the measurement dataset of target region in target accounting period, resampling according to uniform time granularity to measurement dataset and aligning according to time sequence, linear interpolation is filled in according to its adjacent effective sampling point to missing sampling point, and output measurement processing set;According to the summary caliber of target region, the partition summary of measurement processing set is carried out, and the measurement candidate is obtained by accumulation. By constructing the consistency writing judgment of double-source candidate of measurement link and independent verification link comparison, and when deviating, the common offset is extracted based on the batch or operation and maintenance strategy grouping of measurement device, the measurement candidate is corrected, and the verification amplitude is limited, the continuous deviation of the same batch systematic measurement deviation is inhibited to the rolling update of regional water budget benchmark quota, and the recheckable verification is realized.

Description

Technical Field

[0001] This invention relates to the field of water resources metering and accounting technology, and more specifically, to a method and system for dynamically adjusting regional water budget benchmark quotas. Background Technology

[0002] In the dynamic accounting of regional water budget benchmark quotas, the goal of existing technologies is to enable management departments to obtain a benchmark quota that can be used for allocation and control in each rolling cycle. Therefore, the industry practice is to take the metering reports of each water intake as the main basis, first align the data of each point according to time and fill in the missing data, then remove or smooth abnormal fluctuations, and then summarize the processed data by administrative region or water supply zone to obtain the benchmark quota for the current period, and continue to update it in the same way in the next period. In actual deployment, edge computing nodes are also introduced to undertake on-site preprocessing and summary reporting. Taking the actual application of monthly rolling in a certain watershed area as an example, the metering devices in the area are mostly purchased in the same batch and maintained in the same way for a long time. The management side requires that the benchmark quota must be generated on time and can be verified. At the same time, it requires that the benchmark quota changes with the actual water consumption rather than being driven by a small amount of abnormal data. Furthermore, under the edge computing architecture, it also requires that the data processing and reporting at each water intake side remain continuous under the conditions of limited computing power and communication. Since the more common problem in this type of scenario is not that individual points suddenly become abnormal, but that the same batch of devices simultaneously exhibit measurement deviations in the same direction after running for a period of time, the data from each point appears more consistent statistically. Existing technologies have difficulty triggering the rules for detecting anomalies. Smoothing and weighted summarization also accumulate this consistency deviation as a real trend into the baseline quota. When the local summarization on the edge computing side uses similar rules, it will also pass this consistency deviation to the upper-level results. As a result, it can be observed that the baseline quota has a continuous unidirectional shift over several consecutive periods. Moreover, this shift is inconsistent with independently verifiable information such as engineering scheduling records or changes in flow at key sections. Ultimately, this results in a result that deviates more and more from the actual water consumption the more it is updated. The technical problem this application aims to solve is: how to calculate and adjust the regional water budget benchmark quota under edge computing conditions when measurement deviations occur in the same direction among the same batch of metering devices and the existing anomaly handling mechanism is difficult to identify, so that the benchmark quota will not be continuously pulled off by systematic deviations and can be verified. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for dynamically adjusting regional water budget benchmark quota calculation. By constructing a dual-source quota comparison and consistency writing judgment between the metering link and the independent verification link, and extracting common offsets based on metering device batches or operation and maintenance strategies when deviations occur, the metering quota is corrected and the verification range is limited. This suppresses the continuous deviation of the rolling update of the regional water budget benchmark quota due to systematic measurement deviations in the same batch and achieves verifiable verification. It also adapts to on-site calculation and reporting under edge computing conditions, thereby solving the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamically adjusting regional water budget benchmark quota calculation, comprising: S1. Obtain the measurement dataset of the target area within the target accounting period, resample the measurement dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points according to their adjacent valid sampling points, and output the measurement processing set. S2. Perform partitioned aggregation and summation on the metering processing set according to the aggregation caliber of the target area to obtain the metering quota, and use the metering direction as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude as the absolute value of the difference, and output the metering characteristics. S3. Obtain verification datasets from different sources on the acquisition link from the measurement dataset. Perform time alignment on the verification datasets according to the same time granularity as the measurement dataset and summarize them to obtain the verification queue. Use the verification direction as the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude as the absolute value of the difference. Output the verification features. S4. Execute the write judgment. When the measurement direction is the same as the verification direction, and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period. Otherwise, the verification mark is output.

[0005] It should be noted that the waiting quota in this scheme refers to the candidate benchmark quota calculated from data according to the prescribed summary caliber within the current accounting cycle, which has not yet been confirmed through the writing judgment or verification.

[0006] In a preferred embodiment, S5, when outputting the verification mark, the metering processing set is grouped according to the metering device batch or operation and maintenance strategy to obtain a metering group set. For each metering group, its group quota and its group range are calculated. The median of each group range is taken as the common range, and the difference between the common range and the verification range is used as the offset. The metering quota is corrected using the offset to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of the current period.

[0007] In a preferred embodiment, S5 further includes: S5-1. Taking the metering processing set as input, the metering device is grouped according to batch or operation and maintenance strategy to obtain the metering group set. The target accounting cycle is divided into multiple equal-length sub-time periods. For each metering group, partitioning and summarizing are performed in each sub-time period to obtain the group sub-time period sequence. The group sub-time period sequences are then arranged into a grouping matrix according to the grouping order, and the grouping matrix is ​​output. Among them, the operation and maintenance strategy grouping refers to classifying the corresponding devices into the same group for statistics and verification based on the consistency of operation and maintenance rules such as maintenance cycle, calibration method, fault handling rules and data reporting mechanism of the metering devices. S5-2. Taking the group matrix as input, perform singular value decomposition on the group matrix, take the left singular vector corresponding to the maximum singular value as the master mode vector, project the sub-time sequence of each measurement group onto the master mode vector to obtain the group projection value, and take the absolute value of the group projection value as the group amplitude of the measurement group, and output the group amplitude sequence. S5-3. Using the grouped amplitude sequence and the verification amplitude as input, the median of the grouped amplitude sequence is taken to obtain the common amplitude, and the difference between the common amplitude and the verification amplitude is used to determine the offset. The offset is used to correct the measurement reserve to obtain the corrected reserve. If the change of the corrected reserve relative to the measurement base of the previous period exceeds the verification amplitude, the corrected reserve is shrunk along the measurement base of the previous period towards the corrected reserve until the change is equal to the verification amplitude. The shrunk corrected reserve is output as the regional water budget base for this period.

[0008] In a preferred embodiment, S1 further includes: S1-1. Perform statistics on the adjacent time intervals of each measurement sampling point in the measurement dataset and take the median as the unified time granularity. Generate a target time axis covering the target accounting cycle based on the unified time granularity and output the target time axis. S1-2. Perform mapping alignment on the target time axis for the measurement dataset. For multiple measurement sampling points falling into the same target time point, sum them up by time distance and normalize them to obtain the aligned sampling value of the target time point. Output the aligned measurement sequence. S1-3. For missing time points in the aligned measurement sequence, take the previous valid sample value and the next valid sample value and perform linear interpolation according to the time ratio to generate a supplementary sample value. When the missing span is greater than twice the uniform time granularity, mark the corresponding supplementary sample value as a low confidence supplementary value and output the measurement processing set.

[0009] In a preferred embodiment, S2 further includes: S2-1. Determine the set of member measurement points for at least one partition according to the summary caliber of the measurement processing set, and perform time-by-time summation on the measurement sequence of each member measurement point in each partition at a uniform time granularity, and output the partition summary sequence. S2-2. Perform time integration on the summary sequence of each partition within the target accounting period and accumulate it to obtain the metering quota, and output the metering quota; S2-3. Calculate the difference between the metering quota and the metering base quota of the previous period, take its sign as the metering direction, take the absolute value of the difference as the metering amplitude, and output the metering characteristics.

[0010] In a preferred embodiment, S3 further includes: S3-1. Construct a verification time axis with the same measurement time granularity for each verification source data in the verification dataset. Under the constraint of maintaining the time order and minimizing the time shift of adjacent mappings, solve the time mapping relationship that minimizes the sum of squared deviations between the mapped sequence of the verification source data and the predicted sequence obtained by interpolation of its adjacent time. Output the verification alignment sequence of each verification source. S3-2. Initialize the weights of each verification source based on the verification alignment sequence of each verification source and generate the verification fusion sequence. Calculate the residual sequence of each verification source and update the corresponding weights with the reciprocal of the residual variance. Repeat the generation of the verification fusion sequence and the updating of weights until the sum of the weight changes in two adjacent rounds is less than the convergence tolerance or the number of iterations reaches the maximum number of iterations. Output the converged verification fusion sequence. S3-3. Perform time integration on the verification fusion sequence within the target accounting period and summarize to obtain the verification candidate quota. Then, use the verification direction as the sign of the difference between the verification candidate quota and the verification base quota of the previous period, and the verification amplitude as the absolute value of the difference, and output the verification feature.

[0011] In a preferred embodiment, S4 further includes: S4-1. Taking the verification alignment sequence as input, perform repeated sampling on the verification alignment sequence and calculate the verification candidate and verification range corresponding to each sampling. Take the lower quantile and upper quantile values ​​of the verification range to form the verification range interval. Stop repeated sampling when the difference between the interval widths of two adjacent rounds of verification range intervals is less than one-hundredth of the interval width of the previous round. Output the converged verification range interval. S4-2. Taking the verification range interval and the verification range sequence of the most recent accounting periods as input, calculate the median absolute deviation of the verification range sequence and use three times it as the stable deviation. Expand the verification range interval to both sides according to the stable deviation to obtain the allowable range interval, and output the allowable range interval.

[0012] In a preferred embodiment, S4 further includes: S4-3. Taking the measurement direction, measurement range, verification direction, and allowable range range as input, if the measurement direction and verification direction are the same and the measurement range falls within the allowable range range, the measurement reserve quota is output as the regional water budget base quota for this period. Otherwise, under the constraint that the adjustment coefficient is greater than or equal to 0 and less than or equal to 1, the adjustment coefficient that minimizes the absolute value of the difference between the adjusted range and the verification range and makes the adjusted range fall within the allowable range range is calculated, and the verification mark carrying the adjustment coefficient is output. The adjusted range is the product of the adjustment coefficient and the measurement range.

[0013] A dynamically adjustable regional water budget benchmark quota accounting system includes: The metering preprocessing module is used to obtain the metering dataset of the target area within the target accounting period, resample the metering dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points with their adjacent valid sampling points, and output the metering processed set. The metering summary module is used to perform partitioned summarization of the metering processing set according to the summary caliber of the target area and accumulate it to obtain the metering quota. The metering direction is used as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude is the absolute value of the difference. The module outputs the metering characteristics. The verification and aggregation module is used to acquire verification datasets from different sources on the acquisition link than the measurement dataset. It performs time alignment on the verification datasets according to the same time granularity as the measurement dataset and aggregates them to obtain the verification queue. The verification direction is the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude is the absolute value of the difference. The module outputs the verification features. The write judgment module is used to perform write judgment. When the measurement direction is the same as the verification direction and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period; otherwise, a verification mark is output. The grouping correction module is used to group the metering processing set according to the metering device batch or operation and maintenance strategy to obtain the metering group set when outputting the verification mark. For each metering group, it calculates its group quota and its group range. The median of each group range is taken as the common range, and the difference between the common range and the verification range is used as the offset. The offset is used to correct the metering quota to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of the current period.

[0014] The technical effects and advantages of this invention are as follows: 1. The measurement link and verification link are obtained from separate sources and aligned and summarized separately. The measurement direction and verification direction and the range of the measurement are used to determine the writing, so that the benchmark amount is relatively less likely to be continuously deviated under the systematic deviation of the same batch and has a basis for review. It is also adapted to local accounting and reporting under edge computing conditions. 2. The measurement dataset is resampled with uniform time granularity and time alignment. Missing sampling points are filled by linear interpolation of adjacent valid sampling points, which reduces the drift of the summary caliber caused by uneven sampling and missing points, making the calculation of the measurement quota relatively stable and recalculated. 3. The measurement processing set is divided into zones according to the summary scope and accumulated to form the measurement reserve quota. The difference between the reserve quota and the measurement base quota of the previous period is used to obtain the measurement direction and measurement range, which strengthens the explicit quantification of cross-cycle changes and facilitates the consistency verification of subsequent writing judgment. 4. The verification dataset is aligned with the measurement granularity and summarized to obtain the verification waiting quota. The difference between the verification waiting quota of the previous period is then calculated to obtain the verification direction and verification range. The change reference from an independent source is introduced to relatively suppress single-source misjudgment caused by relying solely on measurement reporting. Furthermore, the alignment and summarization can be completed by edge computing nodes and then transmitted back. 5. The write judgment is based on the verification range and constructs an allowable range according to the proportional coefficient. The range of measurement is constrained to ensure that the write result is consistent with the verification change within a certain range, thereby reducing the probability of abnormal accumulation entering the benchmark limit. 6. When the verification mark is triggered, the metering device is grouped according to the batch or operation and maintenance strategy, and the median of the group amplitude is taken to form a common amplitude. The offset is calculated to correct the metering quota and the amplitude is limited by the verification amplitude. This can alleviate the systematic impact of the common offset of homogeneous groups on the benchmark quota, and support the output of results after triggering group correction on the edge computing side. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method steps of the present invention.

[0016] Figure 2 This is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1 The present invention provides a method for calculating a dynamically adjustable regional water budget benchmark quota, comprising: S1. Obtain the measurement dataset of the target area within the target accounting period, resample the measurement dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points according to their adjacent valid sampling points, and output the measurement processing set. This embodiment provides a specific implementation method for the measurement data preprocessing process in step S1. Its purpose is to convert the measurement dataset formed in the target region within the target accounting period into a continuous, recalculated, and confidence-labeled measurement processing set at a unified time granularity. This allows subsequent step S2 to directly perform partitioned summarization and time integration on the measurement processing set to obtain the measurement quota. The mechanism of step S1 is to first statistically obtain a unified time granularity from the measurement dataset and generate a target time axis covering the target accounting period accordingly. Then, the observation records in the measurement dataset are mapped to the target time axis to obtain an aligned measurement sequence. Finally, missing time points in the aligned measurement sequence are filled with imputed sample values ​​according to determined interpolation or preservation rules. Imputed sample values ​​with excessively large missing spans are marked as low-confidence imputed values, ensuring that the measurement processing set satisfies both continuity and retains uncertainty warning information. This implementation process includes the following steps: S1-1. The purpose of this step is to calculate a stable and unified time granularity from the actual sampling rhythm of the measurement dataset, and to generate a target time axis covering the target accounting cycle based on the unified time granularity, thereby providing a unique time reference for the mapping alignment in step S1-2. Specifically, taking the measurement dataset as input, the observation records of each measurement sampling point in the measurement dataset are read and sorted in ascending order by timestamp. The timestamp difference between two adjacent observation records of the measurement sampling point is calculated to obtain the adjacent time interval sequence. Abnormal interval elimination is performed on the adjacent time interval sequence. The abnormal interval elimination rule is to eliminate timestamp differences smaller than the minimum allowable interval and timestamp differences larger than the target accounting cycle length. The minimum allowable interval is given by the system configuration and is used to mask jitter and duplicate reporting. The adjacent time interval sequences of all measurement sampling points are merged into a global time interval set, and the median of the global time interval set is taken as the unified time granularity. When the number of elements in the global time interval set is even, the arithmetic mean of the two middle values ​​is used as the unified time granularity to ensure uniqueness. Based on the start time and end time of the target accounting cycle, a timestamp sequence is generated incrementally from the start time of the target accounting cycle according to the unified time granularity until the generated timestamp exceeds the end time of the target accounting cycle, thus obtaining the target time axis covering the target accounting cycle and outputting the target time axis. When the difference between the end time and the start time of the target accounting cycle is less than the unified time granularity, the target time axis is set to contain only the timestamp sequence of the start time and end time of the target accounting cycle and output as the target time axis, to ensure that subsequent mapping alignment can still be performed. S1-2. The purpose of this step is to map the observation records in the econometric dataset to the target time axis, thereby generating an aligned econometric sequence at a unified time granularity. This allows data from different econometric sampling points to be directly compared on the same time scale and can be used for missing data completion in step S1-3. Specifically, using the econometric dataset and the target time axis as input, the observation records sorted in ascending order by timestamp are read for each econometric sampling point in the econometric dataset. Mapping alignment processing is performed on each target time point in the target time axis. The mapping window for the mapping alignment processing is set to a time range that is expanded forward and backward by half of the unified time granularity centered on the target time point. The set of observation records whose timestamps fall within the mapping window is then selected from the observation records of that econometric sampling point. When the set of observation records contains only one observation record, the observation value of that observation record is determined as the aligned sampling value of the target time point and written into the aligned econometric sequence of that econometric sampling point. When the observation record set contains multiple observation records, calculate the time distance between the timestamp of each observation record and the target time point, calculate the weight based on the reciprocal of the time distance, and normalize the weight. Use the normalized weight to perform a weighted summation on the multiple observation values ​​to obtain the aligned sampled value of the target time point and write it into the aligned measurement sequence of the measurement sample point. When there is an observation record with a time distance of zero, reset the weight of the observation record with a time distance of zero to one and reset the weight of the remaining observation records to zero to avoid the reciprocal operation reaching infinity. When the observation record set is empty, mark the target time point as missing and write the missing identifier into the aligned measurement sequence of the measurement sample point. After completing the mapping and alignment processing of all target time points in the target time axis, output the aligned measurement sequence, which is indexed by the measurement sample point identifier and the target time point and contains the aligned sampled value or missing identifier for reading in steps S1-3. S1-3: The purpose of step S1-3 is to perform recalcible completion calculations on missing identifiers in the aligned measurement sequence and generate completed sample values. Simultaneously, completed sample values ​​with excessively large missing spans are marked as low-confidence completed values, thus outputting a measurement processing set with both continuity and confidence markings for step S2 to read. Specifically, using the aligned measurement sequence, the target time axis, and a unified time granularity as input, the aligned measurement sequence for each measurement sampling point is traversed sequentially along the target time axis. When the missing time point corresponding to a missing identifier is reached, the nearest previous valid sample value and its previous valid time point are searched forward along the target time axis, and the nearest next valid sample value and its next valid time point are searched backward along the target time axis. A valid time point is defined. When both the preceding and following valid sample values ​​are found simultaneously, the time ratio between the missing time point and the preceding and following valid time points is calculated. Using this time ratio, a padded sample value is generated for the preceding and following valid sample values ​​according to linear interpolation rules. This padded sample value is then written to the missing time point to replace the missing identifier. When only the preceding valid sample value is found and the following valid sample value is not found, the preceding valid sample value is used as the padded value and written to the missing time point to replace the missing identifier. When only the following valid sample value is found and the preceding valid sample value is not found, the following valid sample value is used as the padded value and written to the missing time point to replace the missing identifier. When neither a previous nor a subsequent valid sample value is found, the measurement sampling point is marked as having no valid data within the target accounting period, and a null-aligned measurement sequence is output for the upper-level process to perform elimination or weight reduction processing on the measurement sampling point. Further, for each measurement sampling point, a continuous missing interval consisting of consecutive missing identifiers is identified. The missing span of the continuous missing interval is calculated as the number of consecutive missing time points contained in the continuous missing interval multiplied by a uniform time granularity. When the missing span is greater than twice the uniform time granularity, the continuous missing interval is obtained by linear interpolation or by maintaining the rules. Each completed sampled value is marked as a low-confidence complete value, and the low-confidence complete value mark is written into the measurement processing set along with the corresponding completed sampled value. When the missing span is no more than twice the uniform time granularity, the completed sampled value is marked as a high-confidence complete value and written into the measurement processing set. Finally, the measurement processing set is output, which includes the measurement sample point identifier, the target time point, the completed sampled value, and the low-confidence complete value mark or the high-confidence complete value mark, so that step S2 can select to perform elimination, deweighting or trigger verification mark according to the mark when performing partition summary and time integration. Through steps S1-1 to S1-3, the measurement dataset can automatically generate a unified time granularity and a target time axis without relying on manual specification of the sampling period. The measurement dataset can complete the mapping and alignment on the target time axis and form an aligned measurement sequence. The aligned measurement sequence can generate supplementary sampling values ​​according to determined linear interpolation or preservation rules and output a measurement processing set with confidence labels, thereby improving the recalculation and verifiability of the measurement processing set in subsequent measurement quota calculations and reducing the impact of sequence breaks caused by communication interruption or reporting jitter on the regional water budget benchmark quota calculation results. In practical applications: Taking the monthly target accounting cycle of the target area as an example, the measurement dataset contains observation records of multiple measurement sampling points, and the sampling interval of the observation records simultaneously exists in three cases: five minutes, ten minutes, and irregular intervals. Step S1-1 merges the adjacent time interval sequences of each measurement sampling point and takes the median to obtain ten minutes as a unified time granularity and generates a target time axis covering the start and end times of the month. Step S1-2 calculates the aligned sample value by weighting the time distance in the mapping window of each target time point for each measurement sampling point, and writes the missing identifier to form an aligned measurement sequence when the mapping window is empty. Step S1-3 performs linear interpolation to generate the supplementary sample value for the missing identifier in the aligned measurement sequence when there are valid sample values ​​on both sides of the missing interval, and performs the maintenance supplementation to generate the supplementary sample value when there are only valid sample values ​​on one side of the missing interval. At the same time, for continuous missing intervals with a missing span of more than twenty minutes, the corresponding supplementary sample value is marked as a low-confidence supplementary value and written into the measurement processing set. Finally, the output is a measurement processing set that can be read and executed by step S2 for partition summary and time integration.

[0019] S2. Perform partitioned aggregation and summation on the metering processing set according to the aggregation caliber of the target area to obtain the metering quota, and use the metering direction as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude as the absolute value of the difference, and output the metering characteristics. This embodiment provides a specific implementation method for the metering quota calculation process in step S2. Its purpose is to divide and summarize the metering processing set output in step S1 according to the summary caliber of the target area, and form a recalcible metering quota within the target accounting cycle. At the same time, the metering quota is further compared with the metering base quota of the previous cycle, and the metering features that can be directly written and read in step S4 are output. The mechanism of step S2 is as follows: First, a definite mapping is established between the metering sampling points and the partitions using a summary caliber. Then, the metering sequences within the partitions are aggregated into a partition summary sequence at a uniform time granularity. Next, the partition summary sequence is integrated over time within the target accounting period and accumulated to obtain the metering reserve quota. Finally, the difference between the metering reserve quota and the metering base quota of the previous period is used to calculate the metering direction and metering amplitude to form metering characteristics. This implementation process includes the following steps: S2-1, the purpose of step S2-1 is to convert the metering processing set into a partition-computable summary sequence according to the summary caliber, so that the subsequent step S2-2 can perform time integration at the partition level without being affected by differences in the number of metering sampling points and sampling rhythm. Specifically, taking the metering processing set, summary caliber, unified time granularity, and target time axis as inputs, the partition set and member metering point set are first determined according to the summary caliber. The summary caliber is given by a preset configuration and includes at least a correspondence table between partition identifiers and metering sampling point identifiers. When a metering sampling point undergoes a partition change within the target accounting period, the summary caliber further includes the effective time of the partition change and is used to segment and determine the member metering point set on the target time axis. Subsequently, the metering sequence of the member metering point set on the target time axis is read for each partition, and... For each target time point on the target time axis, a time-by-time summation is performed. The time-by-time summation rule is to sum the sampled values ​​of the member meter set of the partition at the target time point to obtain the partition summary value, forming a partition summary sequence sorted by the target time axis and outputting the partition summary sequence. During the time-by-time summation process, if the sampled value of a member meter at the target time point is marked as a low-confidence padding value, a partition contribution coefficient is applied to the sampled value according to the rule constraint before it participates in the summation. The partition contribution coefficient is given by a preset configuration and its value range is greater than or equal to zero and less than or equal to one, so as to avoid unreasonable pulling of long span padding on the partition summary sequence. If a member meter is marked as having no valid data within the target accounting period, the member meter is removed from the member meter set of the partition and the removal log is recorded for traceability. S2-2. The purpose of this step is to convert the zonal summary sequence into a metering quota within the target accounting period, so that the metering quota can represent the candidate water use benchmark quota for the target area with a unified caliber. Specifically, taking the zonal summary sequence, unified time granularity, and the boundary of the target accounting period as input, the zonal summary sequence of each zonal is first integrated with time within the target accounting period to obtain the zonal quota. The time integration rule is executed in two categories according to the physical meaning of the zonal summary sequence and the type is specified by the preset configuration: when the zonal summary sequence is an instantaneous flow sequence, the rectangular integration rule is adopted, multiplying the zonal summary value at each target time point by the duration corresponding to the unified time granularity and accumulating it within the target accounting period to obtain the zonal quota. When the partition summary sequence is a cumulative sequence, the differential accumulation rule is adopted. The partition summary value corresponding to the end time of the target accounting period is subtracted from the partition summary value corresponding to the start time of the target accounting period. When there is a return to zero or a rollback, the values ​​are segmented according to the return point and accumulated separately to obtain the partition quota. After obtaining the partition quota of each partition, the quota of all partitions is accumulated to obtain the measurement quota and output the measurement quota. When the partition summary sequence is missing the sample value corresponding to the target time point at the boundary of the target accounting period, the most recent valid sample value is selected as the boundary sample value in the mapping window of the boundary of the target accounting period. If there is no valid sample value in the mapping window of the start time and the end time of the target accounting period, the partition is marked as the boundary non-integrable and the quota of the partition is set to zero. At the same time, the boundary non-integrable mark is written into the composition details of the measurement quota for subsequent verification. S2-3 aims to convert the metering quota into a metering feature that can be directly used in the judgment, so that step S4 can use the metering direction and metering amplitude to make a consistency judgment with the verification features. Specifically, taking the metering quota and the metering base of the previous period as input, the difference is calculated as the metering quota minus the metering base of the previous period. The metering base of the previous period is synchronously fixed and stored by the system when the regional water budget base of the current period is output in the previous period and can be retrieved according to the target area identifier. The sign of the difference is determined as the metering direction. When the difference is greater than zero, the metering direction is determined as the positive direction. When the difference is less than zero, the measurement direction is determined to be negative; when the difference is equal to zero, the measurement direction is determined to be zero and will only match the zero direction in subsequent write judgments according to the principle of direction consistency. The absolute value of the difference is determined as the measurement amplitude, and the measurement direction, measurement amplitude and measurement quota are written together into the measurement feature and the measurement feature is output. When the measurement base of the previous period is missing or cannot be retrieved, the measurement base of the previous period is set as the measurement quota, the measurement direction is set to zero and the measurement amplitude is set to zero, and a base missing mark is written to ensure that the process can continue to be executed and to provide traceable basis for subsequent verification. Through steps S2-1 to S2-3, the metering processing set can be stably mapped to the partition according to the summary caliber and form a partition summary sequence. The partition summary sequence can obtain a recalcible metering quota within the target accounting cycle according to the determined time integration rules. The metering quota can form a difference with the metering base of the previous cycle and output the metering direction and metering magnitude, so that the metering characteristics have a clear value source, clear operation rules and clear anomaly handling path, and reduce the uncontrollable impact of low confidence fill values ​​and boundary missing values ​​on the metering quota and metering characteristics. In practical applications: Taking the monthly target accounting cycle of the target area as an example, the summary caliber is given by the system configuration and the metering sampling points are mapped to the two-level structure of administrative division and water supply division. Step S2-1 performs time-by-time summation on the set of member metering points of each division at a unified time granularity to obtain the division summary sequence, and applies a division contribution coefficient to the sampled values ​​marked as low confidence fill values ​​to reduce the impact of long span fill. Step S2-2 performs rectangular integration on the division summary sequence of instantaneous flow type at a unified time granularity to obtain the division reserve quota, and accumulates all division reserve quotas to obtain the metering reserve quota. At the same time, the division summary sequence of cumulative amount type obtains the division reserve quota according to the differential accumulation rule to avoid duplicate accumulation. Step S2-3 calculates the difference between the metering reserve quota and the metering base quota of the previous period to obtain the difference value, and determines the metering direction and metering amplitude accordingly. The final output metering feature is directly read by step S4 and consistent with the verification feature, thus providing a verifiable basis for whether to trigger the verification mark in the future.

[0020] S3. Obtain verification datasets from different sources on the acquisition link from the measurement dataset. Perform time alignment on the verification datasets according to the same time granularity as the measurement dataset and summarize them to obtain the verification queue. Use the verification direction as the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude as the absolute value of the difference. Output the verification features. This embodiment provides a specific implementation method for the verification feature generation process in step S3. Its purpose is to convert the verification dataset, which is from a different source than the measurement dataset collection link, into a verification feature that can be compared with the measurement feature in the same dimension, so that step S4 can complete the writing judgment at the same time granularity and the same quota caliber. The mechanism of step S3 is as follows: First, establish a verification time axis consistent with the measurement time granularity for each verification source data in the verification dataset. Then, under the constraints of maintaining the time sequence and smoothing the mapping time shift, solve the time mapping relationship to obtain the verification alignment sequence. Next, use the verification alignment sequence to construct the verification fusion sequence and achieve self-suppression of abnormal verification sources through residual variance-driven weight iteration. Finally, perform integral summarization consistent with the measurement quota on the verification fusion sequence within the target accounting period to obtain the verification quota. Compare it with the verification base quota of the previous period to output the verification direction and verification magnitude to form verification features. The implementation process includes the following steps: S3-1. The purpose of this step is to align the verification source data to the measurement time granularity in the time dimension, so that different verification source data can form a comparable verification alignment sequence on a unified time axis, thereby providing the same index space for the fusion calculation in step S3-2. Specifically, taking the verification dataset, measurement time granularity, target accounting cycle start time, target accounting cycle end time, and target time axis as input, the following processing is performed on each verification source data in the verification dataset: First, the verification time axis is generated by incrementally increasing the measurement time granularity from the target accounting cycle start time until it does not exceed the target accounting cycle end time, thus obtaining a verification time axis consistent with the measurement time granularity; Subsequently, a time mapping relationship is constructed using the original timestamp sequence of the verification source data and the verification time axis as objects. The time mapping relationship is represented by a monotonically increasing mapping index sequence. Each element of the mapping index sequence is used to indicate which observation record in the original timestamp sequence of the verification source data corresponds to a time point on the verification time axis, or the interpolation position between two observation records. When solving the mapping index sequence, two types of constraints are satisfied simultaneously. The first type of constraint is the time order preservation constraint, which requires that the mapping index sequence increases with the time point of the verification time axis and is not allowed to decrease, so as to avoid time reversal. The second type of constraint is the minimum change in adjacent mapping time shift, which requires that the change in mapping time deviation corresponding to two adjacent verification time points be as small as possible. The mapping time deviation is defined as the difference between the verification time point and its corresponding original timestamp of the verification source, and the change in adjacent mapping time shift is measured by the square of the difference in adjacent mapping time deviation and accumulated over the entire sequence. Furthermore, the objective is constructed with minimizing the sum of squared deviations as the goal. The deviation for minimizing the sum of squared deviations is defined as the difference between the mapped sequence of the verification source data and the predicted sequence. The predicted sequence is obtained by linear interpolation of adjacent time points of the mapped sequence. Specifically, the predicted value for each verification time point is obtained by interpolating the mapped values ​​of the previous and subsequent verification time points according to the time ratio. Dynamic programming is used to solve the mapping index sequence, minimizing the sum of squared deviations between the mapped and predicted sequences while satisfying the constraints of maintaining time order and minimizing the time shift change between adjacent mappings, thus obtaining the time mapping relationship. Finally... Based on the time mapping relationship, the verification source data is resampled to the verification time axis. The resampling rule is that when the mapping index sequence points to a single observation record, the observation value of that observation record is directly taken. When the mapping index sequence points to two observation records, the two observation values ​​are linearly interpolated according to the time ratio to obtain the observation value, forming a verification alignment sequence and outputting the verification alignment sequence. When the number of valid observation records of the verification source data in the target accounting period is less than two, it is impossible to construct a prediction sequence and an interpolation sequence. The verification source data is marked as an unalignable verification source and an empty verification alignment sequence is output for the weight reduction processing to be performed in step S3-2. S3-2. The purpose of this step S3-2 is to generate a stable verification fusion sequence in the case of noise, missing data, or local anomalies in multiple verification sources. This verification fusion sequence can represent the comprehensive verification result of the verification dataset and automatically reduce the impact of abnormal verification sources. Specifically, taking the verification alignment sequence, convergence tolerance, and maximum number of iterations of each verification source as input, initial verification source weights are first assigned to all alignable verification sources. The initial verification source weights are initialized with equal weights and normalized according to the number of verification sources to obtain an initial verification source weight set with a weight sum of one. For non-alignable verification sources, the initial verification source weights are reset to zero and kept in the weight set for traceability. Subsequently, the verification alignment sequences of each verification source are weighted and summed at each time step according to the verification source weights to obtain the verification fusion sequence, and the verification fusion sequence is written to the fusion cache for residual calculation. For each verification source, a residual sequence is calculated. The residual sequence is defined as the observation value of the verification alignment sequence of the verification source at each verification time point minus the observation value of the verification fusion sequence at the same verification time point. After obtaining the residual sequence, the residual variance is calculated within the target calculation period. The residual variance calculation rule is the square mean of the residual sequence minus the square mean of the residual sequence. When the residual variance is less than the minimum variance lower limit, the residual variance is corrected to the minimum variance lower limit to avoid the weights being infinitely amplified. The minimum variance lower limit is given by a preset configuration. The verification source weights are updated according to the reciprocal of the residual variance. The verification source weight update rule is to set the unnormalized weight of each verification source to the reciprocal of the residual variance of that verification source, and perform normalization on all unnormalized weights to obtain the updated set of verification source weights. The process of generating the verification fusion sequence, calculating the residual sequence, calculating the residual variance, and updating the verification source weights is repeated until a stopping condition is met. The stopping condition is that the sum of the weight changes of the weight sets of two adjacent rounds of verification sources is less than the convergence tolerance or the number of iterations reaches the maximum number of iterations. The convergence tolerance is given by a preset configuration and is set to the upper limit of the sum of the weight changes. The maximum number of iterations is given by a preset configuration to ensure that the processing delay is controllable. When the stopping condition is met, the converged verification fusion sequence is output. Furthermore, when a verification source is missing observations at multiple consecutive verification time points, the residual sequence calculation for the verification source at the missing verification time points is skipped and the residual value at the missing verification time point is recorded as zero. At the same time, the residual variance of the verification source is unbiasedly corrected according to the number of effective residual samples to avoid the residual variance being underestimated due to missing data. S3-3 The purpose of this step S3-3 is to convert the verification fusion sequence into a verification queue comparable to the metering queue, and further output the verification direction and verification amplitude to form verification features, so that step S4 can directly read the verification features to complete the writing judgment; Specifically, with the verification fusion sequence, the target accounting cycle boundary, the metering time granularity, and the verification queue type configuration as inputs, the verification fusion sequence is first integrated and summarized within the target accounting cycle to obtain the verification queue. The verification queue type configuration is given by the preset configuration and is used to indicate the physical meaning of the verification fusion sequence. When the verification fusion sequence is an instantaneous flow type, the rectangular integration rule is adopted to multiply the verification fusion value of each verification time point by the duration corresponding to the metering time granularity and accumulate it within the target accounting cycle to obtain the verification queue. When the verification fusion sequence is of the cumulative type, the differential accumulation rule is adopted. The verification fusion value corresponding to the end time of the target accounting cycle is subtracted from the verification fusion value corresponding to the start time of the target accounting cycle. When a zero return or rollback occurs, the values ​​are segmented according to the zero return point and then accumulated to obtain the verification candidate quota. After outputting the verification candidate quota, the verification base quota of the previous cycle is read from the verification base storage. The verification base quota of the previous cycle is synchronously fixed and stored by the system when the regional water budget base quota is finally output in the previous cycle and can be retrieved according to the target region identifier. The verification difference is calculated as the verification candidate quota minus the verification base quota of the previous cycle. The sign of the verification difference is determined as the verification direction and the absolute value of the verification difference is determined as the verification amplitude. When the verification difference is equal to zero, the verification direction is determined as the zero direction and only matches the zero direction in the direction consistency judgment in step S4. The verification candidate quota, verification direction and verification amplitude are written into the verification features and the verification features are output. When the verification base amount of the previous cycle is missing or cannot be retrieved, the verification base amount of the previous cycle is set as the verification waiting amount, the verification direction is set to zero, and the verification range is set to zero. At the same time, a verification base amount missing mark is written to ensure that the verification features can still be output and provide traceable basis for subsequent verification. Through steps S3-1 to S3-3, the verification dataset can form a verification alignment sequence at the measurement time granularity and generate a converged verification fusion sequence through weighted iterative fusion. The verification fusion sequence can obtain verification candidates and form verification direction and verification magnitude within the target accounting cycle according to the determined integral summarization rules. This gives the verification features clear value sources, clear solution constraints, clear convergence stopping conditions, and clear missing anomaly handling paths, and reduces the risk of misleading write decisions due to single verification source anomalies, verification source clock drift, and verification source missingness. In practical applications: taking the monthly target accounting cycle of the target area as an example, the verification dataset includes two types of verification source data: net inbound and outbound flow records from the scheduling system and storage capacity change records from the telemetry system. Step S3-1 generates and measures the time granularity of each verification source data. The verification time axis is consistent with the time order, and the time mapping relationship is solved by dynamic programming under the constraint of minimizing the time shift change of adjacent mappings, to obtain the verification alignment sequence aligned to the verification time axis respectively; Step S3-2 initializes the verification source weights equally to the verification alignment sequence to generate the verification fusion sequence, and iteratively updates the verification source weights according to the inverse of the residual variance until the sum of the weight changes is less than the convergence tolerance or the number of iterations reaches the maximum number of iterations, to obtain the converged verification fusion sequence; Step S3-3 obtains the verification quota according to the rectangular integral rule for the verification fusion sequence of instantaneous flow type, and calculates the difference with the verification base quota of the previous period to output the verification direction and verification amplitude. The final output verification feature is directly read by Step S4 and is consistent with the measurement feature, so that the verifiable verification feature can still be output even if there is a local missing verification source or clock drift.

[0021] S4. Execute write judgment. When the measurement direction is the same as the verification direction, and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period; otherwise, the verification mark is output. This embodiment provides a clear and specific implementation method for the writing judgment and verification mark output process in step S4. Its purpose is to establish a verifiable allowable range between the measurement characteristics and the verification characteristics, and based on this, determine whether the measurement quota can be directly written into the regional water budget base for the current period. Simultaneously, when the writing conditions are not met, a recalcible adjustment coefficient is calculated, and a verification mark carrying the adjustment coefficient is output for subsequent verification processes. The mechanism of step S4 is as follows: first, a stable verification range is formed based on the verification alignment sequence through repeated sampling; then, a stable deviation is calculated by combining the verification range sequences of several recent accounting periods, and the verification range is expanded to obtain the allowable range; finally, a consistency judgment is performed between the measurement direction and the verification direction, and a range inclusion judgment is performed between the measurement range and the allowable range. If the conditions are met, the measurement quota is written; if the conditions are not met, under the constraint of the adjustment coefficient, an adjustment coefficient is calculated to make the adjusted range as close as possible to the verification range and fall within the allowable range, and a verification mark is output. This implementation process includes the following steps: S4-1. The purpose of this step, S4-1, is to construct a verifiable verification range from the uncertainty of the verification alignment sequence, ensuring that the allowable range is based on the statistical distribution of the verification alignment sequence rather than being arbitrarily set. Specifically, taking the verification alignment sequence, the upper limit of repeated sampling times, the lower quantile configuration, the upper quantile configuration, and the convergence criterion configuration as inputs, repeated sampling is performed on the verification alignment sequence. The repeated sampling uses a block sampling rule to maintain the temporal relevance of the verification alignment sequence. The block sampling rule is to place the target accounting period within the verification... The time axis is divided into several consecutive sampling blocks, and sampling with replacement is performed with the sampling block as the smallest sampling unit. The length of the sampling block is given by a preset configuration and is an integer multiple of the measurement time granularity. In each repeated sampling, the sampled blocks are spliced ​​together in their original order on the verification time axis to form a sampling verification sequence. The sampling verification sequence is then subjected to the same integral summarization rule as step S3-3 to obtain the sampling verification queue. The integral summarization rule of the sampling verification queue is specified by the verification queue type configuration and is consistent with the verification queue calculation. The sampling verification difference is calculated based on the sampling verification quota and the verification base quota of the previous cycle, and the absolute value of the sampling verification difference is taken as the sampling verification range. The sampling verification range obtained from each repeated sampling is written into the verification range sample set. When the verification range sample set accumulates to the current repeated sampling round, the lower quantile and upper quantile values ​​of the verification range sample set are calculated according to the lower quantile configuration and the upper quantile configuration respectively to form the verification range interval. The lower quantile configuration and the upper quantile configuration are given by the preset configuration and are used to indicate the position of the quantile point respectively. The interval width of the verification range interval is calculated as the upper bound of the verification range interval minus the lower bound of the verification range interval. The current round verification is compared after each round of repeated sampling. The difference between the width of the amplitude interval and the width of the amplitude interval of the previous round of verification is used to determine whether the convergence stopping condition is met and the repeated sampling is stopped when the difference is less than one percent of the width of the previous round, or when the number of repeated sampling rounds reaches the upper limit of the number of repeated samplings, the repeated sampling is forcibly stopped; the converged verification amplitude interval is output. When there are missing verification time points in the verification alignment sequence, the missing verification time points are filled in by linear interpolation of adjacent valid verification time points before constructing the sampling verification sequence to ensure that the integral summarization can be executed. The verification time points obtained by filling in the missing time points are marked as low confidence verification time points and the proportion of low confidence is recorded in the verification amplitude sample set for subsequent quality assessment. S4-2. The purpose of this step S4-2 is to combine the verification amplitude range with the historical verification amplitude fluctuation range to form a stable deviation, thereby obtaining an allowable amplitude range that is tolerant of short-term anomalies and constrains long-term drift. Specifically, taking the verification amplitude range, the verification amplitude sequence of the most recent accounting periods, the historical window length configuration, and the stable deviation multiple configuration as inputs, the verification amplitudes of the most recent accounting periods are read from the verification amplitude storage according to the target area identifier and accounting period order to form a verification amplitude sequence. The historical window length configuration is given by a preset configuration and is used to determine the number of accounting periods of the most recent accounting periods. The median of the verification amplitude sequence is calculated as the verification amplitude benchmark value, and the absolute deviation of each verification amplitude in the verification amplitude sequence relative to the verification amplitude benchmark value is calculated. The median absolute deviation is obtained by taking the median of all absolute deviations. The stable deviation is obtained by multiplying the fixed deviation multiple configuration by the median absolute deviation. The stable deviation multiple configuration is given by a preset configuration and is set to three in this embodiment to cover common fluctuations in the verification amplitude sequence. The lower bound of the verification amplitude interval is obtained by subtracting the stable deviation from the lower bound of the verification amplitude interval. The upper bound of the verification amplitude interval is obtained by adding the stable deviation to the upper bound of the verification amplitude interval. The lower bound of the allowable amplitude interval is limited to not less than zero to avoid negative amplitude intervals. The allowable amplitude interval is then output. When the length of the verification amplitude sequence is less than the number of calculation cycles required by the historical window length configuration, the insufficient length of the verification amplitude sequence is marked in the output result of the allowable amplitude interval. The stable deviation is amplified and corrected according to the actual length of the verification amplitude sequence. The amplification and correction rule is to multiply the stable deviation by the ratio of the historical window length configuration to the actual length to improve the tolerance for insufficient historical samples. S4-3. The purpose of this step S4-3 is to convert the consistency between the measurement direction and the verification direction, and the range satisfaction of the measurement range, into an executable write judgment. Simultaneously, when the write judgment is not satisfied, a verification flag with an adjustment coefficient is output to provide a retrievable subsequent verification entry point. Specifically, taking the measurement direction, measurement range, verification direction, allowable range range, measurement quota, verification range, and adjustment coefficient constraint configuration as input, a direction consistency judgment is first performed. The direction consistency judgment rule is that if the measurement direction and the verification direction have the same value, the direction is considered consistent; otherwise, the direction is considered inconsistent. If the direction is consistent... Under the premise of [condition], the amplitude range judgment is performed. The amplitude range judgment rule is that if the lower limit of the allowable amplitude range is not greater than the measurement amplitude and the measurement amplitude is not greater than the upper limit of the allowable amplitude range, the measurement amplitude is determined to fall within the allowable amplitude range; otherwise, the measurement amplitude is determined not to fall within the allowable amplitude range. When the direction is consistent and the measurement amplitude falls within the allowable amplitude range, the measurement reserve is written into the regional water budget base amount for this cycle and the regional water budget base amount for this cycle is output. When the direction is inconsistent or the measurement amplitude does not fall within the allowable amplitude range, the adjustment coefficient solution process is entered and a verification mark carrying the adjustment coefficient is output. The adjustment coefficient solution process is as follows: Under the constraint that the adjustment coefficient ranges from greater than or equal to zero to less than or equal to one, a candidate set of adjustment coefficients is calculated. This candidate set is determined by the lower bound coefficient obtained by dividing the lower bound of the allowable range by the measurement range, and the upper bound coefficient obtained by dividing the upper bound of the allowable range by the measurement range. The lower and upper bound coefficients are then trimmed to the constraint range of zero to one to obtain a feasible coefficient range. When the measurement range is zero, the adjustment coefficient is set to zero, and a check flag carrying the adjustment coefficient is directly output. When the feasible coefficient range is empty, the adjustment coefficient is set to zero, and a check flag carrying the adjustment coefficient is output. When the feasible coefficient interval is not empty, select the adjustment coefficient within the feasible coefficient interval that minimizes the absolute value of the difference between the adjusted amplitude and the verification amplitude, and output the verification flag carrying the adjustment coefficient. The adjusted amplitude is the product of the adjustment coefficient and the measurement amplitude. When multiple adjustment coefficients simultaneously satisfy the minimum difference, select the adjustment coefficient with the smallest value to ensure uniqueness. Write the adjustment coefficient, the direction consistency judgment result, and the amplitude interval judgment result together into the verification flag and output the verification flag so that the subsequent step S5 can read the verification flag when outputting the verification flag and trigger the measurement processing set grouping correction process. Through steps S4-1 to S4-3, the verification alignment sequence can form a convergent verification range interval through block sampling and integral summarization. The verification range interval can form an allowable range interval by combining the median absolute deviation of the verification range sequence. The allowable range interval, together with the measurement direction, verification direction, and measurement range, can complete the verifiable writing judgment. When the writing judgment is not met, an adjustment coefficient with a unique value rule and a verification mark are output. Thus, the writing basis of the regional water budget quota for this period has a clear statistical source, a clear interval construction rule, a clear stopping condition, and a clear anomaly handling path, and reduces the risk of erroneous writing caused by verification data noise, verification alignment error, or short-term measurement anomalies. In practical applications: taking the monthly target accounting cycle of the target region as an example, the verification alignment sequence is output by step S3 and aligned according to the measurement time granularity. Step S4-1 performs block sampling on the verification alignment sequence with a sampling block length given by the preset configuration to form multiple sets. The sampling verification sequence is sampled and verified, and the sampling verification candidate quota and sampling verification range are calculated separately. The lower quantile and upper quantile of the sampling verification range are used as the verification range interval, and sampling is stopped after the interval width converges. Step S4-2 reads the verification range of the past few accounting cycles from the verification range storage to form a verification range sequence and calculates the median absolute deviation. The stable deviation is obtained by using three times the median absolute deviation and the verification range interval is expanded to obtain the allowable range interval. Step S4-3 judges the consistency between the measurement direction and the verification direction and judges the interval between the measurement range and the allowable range interval. If the condition is met, the measurement candidate quota is written into the regional water budget base quota for this cycle. If the condition is not met, the feasible coefficient interval is calculated under the constraint of the adjustment coefficient from zero to one, and the adjustment coefficient that makes the adjusted range closest to the verification range is selected. The output carries the verification mark of the adjustment coefficient. The subsequent step S5 reads the verification mark and performs further verification and correction on the measurement processing set according to the batch of measurement devices or operation and maintenance strategy.

[0022] S5. When outputting the verification mark, the metering processing set is grouped according to the metering device batch or operation and maintenance strategy to obtain the metering group set. For each metering group, its group quota and its group range are calculated. The median of each group range is taken as the common range. The difference between the common range and the verification range is used as the offset. The metering quota is corrected using the offset to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of this period. This embodiment provides a specific implementation method for the group verification and correction process in step S5. Its purpose is to, after outputting the verification mark in step S4, utilize the homogeneous grouping structure formed within the metering processing set according to metering device batches or operation and maintenance strategies to identify common offsets caused by the same batch or operation and maintenance strategy, and perform recalcible corrections on the metering quota under verification range constraints, thereby outputting the regional water budget base quota for the current period that meets the verification constraints. The working mechanism of step S5 is to first divide the metering processing set into metering group sets according to metering device batches or operation and maintenance strategies, and then summarize the sub-time periods of each metering group within the target accounting period according to equal-length sub-time periods to obtain the group sub-time period sequence. Then, the group sub-time periods... The time-period series are assembled into a grouping matrix in grouping order; then, singular value decomposition is performed on the grouping matrix to obtain the master pattern vector, and the sub-time-period series of each group are projected onto the master pattern vector to obtain the group projection value. The absolute value of the group projection value forms the group amplitude sequence; finally, the median of the group amplitude sequence is taken to obtain the common amplitude, which is compared with the verification amplitude to obtain the offset. The offset is used to correct the measurement reserve to obtain the corrected reserve. When the change of the corrected reserve relative to the measurement base of the previous period exceeds the verification amplitude, the corrected reserve is shrunk to the point where the change equals the verification amplitude according to the determined shrinkage rule, thereby outputting the regional water budget base for the current period. This implementation process includes the following steps: S5-1. The purpose of step S5-1 is to reconstruct the metering processing set into a grouping matrix that can reflect the common changes in the same batch or the same operation and maintenance strategy, so that step S5-2 can extract the main pattern at the matrix level and quantify the change magnitude of each group. Specifically, taking the metering processing set, metering device batch field, operation and maintenance strategy field, target accounting cycle start time, target accounting cycle end time, unified time granularity, summary caliber, and sub-period length configuration as input, the metering processing set is first grouped according to the metering device batch or operation and maintenance strategy to obtain the metering group set. The metering device batch field is provided by the metering device ledger and is mapped one-to-one with the metering sampling point identifier. The operation and maintenance strategy field is provided by the operation and maintenance system and includes at least the maintenance cycle identifier, calibration method identifier, and fault handling. The rule identifier and data reporting mechanism identify four types of fields. The operation and maintenance strategy group is defined as a set of metering sampling points that are completely consistent with the four types of fields of the operation and maintenance strategy field. When the metering device batch field and the operation and maintenance strategy field are available at the same time, the grouping is based on the metering device batch field first. When the metering device batch field is missing, the grouping is based on the operation and maintenance strategy field. When both are missing, the corresponding metering sampling points are assigned to the missing group and the missing group mark is recorded for traceability. Then, the target accounting period is divided into multiple equal-length sub-periods. The length of the equal-length sub-period is given by the sub-period length configuration and is an integer multiple of the uniform time granularity. The equal-length sub-periods are generated in a manner that increases sequentially from the start time of the target accounting period and covers up to no more than the end time of the target accounting period. For each measurement group, the set of member measurement sampling points contained in that measurement group is read, and the same partitioning and summarizing rule as in step S2 is applied to the set of member measurement sampling points in each equal-length sub-period to obtain the group summary value of that measurement group in that equal-length sub-period. The partitioning and summarizing rule is to sum the sampled values ​​of the member measurement sampling points at each time point on the target time axis covered by the equal-length sub-period, and then accumulate or integrate them over all time points in the equal-length sub-period to obtain the group summary value of that equal-length sub-period. When the sampled values ​​in the measurement processing set carry low-confidence padding value markers, the sampled values ​​are first multiplied by the low-confidence contribution coefficient according to the low-confidence contribution coefficient configuration before participating in the time-by-time summation to reduce the impact of long-span padding on the group summary value; the distribution of each measurement group in all equal-length sub-periods is then calculated. The group summary values ​​are concatenated in the order of equal-length sub-time periods to form a group sub-time period sequence, and the group sub-time period sequence is output. Further, the group sub-time period sequences of all metering groups are combined into a grouping matrix according to the metering group sorting rules, where the metering group sorting rules are first sorted by the lexicographical order of the metering device batch field and then by the lexicographical order of the operation and maintenance strategy field. The row index of the grouping matrix corresponds to the metering group identifier, the column index of the grouping matrix corresponds to the equal-length sub-time period number, and the matrix elements of the grouping matrix are the group summary values ​​of the corresponding metering group in the corresponding equal-length sub-time period, and the grouping matrix is ​​output. When all member metering sampling points of a metering group have no valid data or all have missing identifiers in a certain equal-length sub-time period, the matrix elements of the equal-length sub-time period are set to zero and a sub-time period missing marker is written for subsequent reliability assessment reading. S5-2. The purpose of this step is to extract the main pattern vector representing the common change pattern of most measurement groups from the grouping matrix, and to quantify the magnitude of each measurement group relative to the common change pattern by projection, so as to output a grouping magnitude sequence that can resist individual group anomalies. Specifically, taking the grouping matrix as input, the first step is to perform standardization on the grouping matrix to avoid the main pattern being dominated by a single high-volume group due to differences in absolute magnitude. The standardization rule is to center each row of the grouping matrix by the mean of that row and normalize it by the standard deviation of that row. If the standard deviation of that row is less than the minimum standard deviation lower limit, the standard deviation of that row is corrected to the minimum standard deviation lower limit, where the minimum standard deviation lower limit is given by a preset configuration. Singular value decomposition is performed on the standardized grouping matrix to obtain the left singular vector corresponding to the largest singular value, and this left singular vector is determined as the main pattern vector. The main pattern vector is a column vector and its dimension is consistent with the number of rows in the grouping matrix. Each element of the main pattern vector corresponds to the weight contribution of a measurement group in the main pattern. When the overall sign of the main pattern vector obtained by singular value decomposition can be positive or negative, resulting in non-uniqueness of the subsequent projection sign, a sign consistency rule is used to perform sign orientation on the main pattern vector. The sign consistency rule is to calculate the arithmetic sum of the elements of the main pattern vector. If the arithmetic sum is less than zero, the main pattern vector is multiplied by negative one to make the arithmetic non-negative, thereby ensuring that the main pattern vector has a unique value. Subsequently, the group projection value is calculated for each measurement group. The calculation rule for the group projection value is to perform an inner product operation between the row vector corresponding to the measurement group in the standardized group matrix and the main pattern vector to obtain the group projection value. The inner product operation alignment is performed according to the consistency of the measurement group index. The absolute value of the group projection value is determined as the group amplitude of the measurement group. The reason for taking the absolute value of the group amplitude is that the group amplitude is used to characterize the deviation strength of the measurement group relative to the main pattern rather than the deviation direction. The deviation direction is uniformly controlled by the measurement direction and the verification direction in step S4. The group amplitudes of all measurement groups are connected according to the measurement group sorting rule to form a group amplitude sequence and output the group amplitude sequence. When the number of rows or columns of the group matrix is ​​less than two, causing the singular value decomposition to fail to form a stable main pattern vector, the group amplitude sequence is set to a sequence composed of the absolute value of the difference between the sum of the sub-period sequences of each measurement group in the target accounting period and the corresponding group quota of the previous period, and a low-dimensional substitution mark is written for subsequent traceability. S5-3. The purpose of this step S5-3 is to use the median of the group amplitude sequence to form a robust estimate of the common changes of most measurement groups, and to compare the common amplitude with the verification amplitude to obtain the offset to correct the measurement quota. At the same time, the verification amplitude is used to apply a hard constraint to the corrected change amplitude to ensure that the output results are consistent with the verification evidence. Specifically, taking the grouped amplitude sequence, verification amplitude, metering quota, previous period's metering base, and metering direction as input, the common amplitude is first obtained by taking the median of the grouped amplitude sequence. When the number of elements in the grouped amplitude sequence is odd, the common amplitude is taken from the middle element after sorting; when the number of elements in the grouped amplitude sequence is even, the common amplitude is taken from the arithmetic mean of the two middle elements after sorting to ensure uniqueness. The offset is calculated by subtracting the verification amplitude from the common amplitude, and the offset is written to the correction cache for metering quota correction. Subsequently, the metering quota is corrected to obtain the corrected quota. The calculation rule is as follows: when the measurement direction is consistent with the verification direction and the measurement direction is positive, the correction candidate amount is determined as the measurement candidate amount minus the offset; when the measurement direction is consistent with the verification direction and the measurement direction is negative, the correction candidate amount is determined as the measurement candidate amount plus the offset; when the measurement direction is zero, the correction candidate amount is determined as the measurement candidate amount and the offset is set to zero, so that the direction of the offset is determined by the direction consistency constraint and the direction loss caused by taking the absolute value of the grouping amplitude is avoided; when the correction candidate amount is less than zero, the correction candidate amount is limited to zero to meet the non-negative candidate amount constraint and written into the non-negative clipping mark. Furthermore, the change in the corrected ante amount relative to the previous period's measurement baseline is calculated. The change is calculated as the absolute value of the difference between the corrected ante amount and the previous period's measurement baseline. When the change does not exceed the verification range, the corrected ante amount is directly output as the regional water budget baseline for this period. When the change exceeds the verification range, the shrunken corrected ante amount is output as the regional water budget baseline for this period according to the shrinkage rule. The shrinkage rule is to first determine the shrinkage direction as from the previous period's measurement baseline to the corrected ante amount, and then trim the change in that direction to the verification range. Specifically, when the corrected ante amount is greater than the previous period's measurement baseline, the shrunken corrected ante amount is determined as the previous period's measurement baseline plus the verification range; when the corrected ante amount is less than the previous period's measurement baseline, the shrunken corrected ante amount is determined as the previous period's measurement baseline minus the verification range; when the corrected ante amount is equal to the previous period's measurement baseline, the shrunken corrected ante amount is determined as the previous period's measurement baseline. The shrunken corrected ante amount, offset, and common range are written into the output record for audit traceability and output of the shrunken corrected ante amount. Through steps S5-1 to S5-3, the metering processing set can form metering group sets according to the batch of metering devices or operation and maintenance strategies, and construct a grouping matrix that reflects the summary form of equal-length sub-periods. The grouping matrix can extract the main pattern vector through singular value decomposition and form a grouping amplitude sequence by projection. The grouping amplitude sequence can obtain the common amplitude through the median and form an offset with the verification amplitude, thereby correcting the direction of metering quota execution in a controlled manner. At the same time, the corrected quota can obtain the final output through a determined shrinkage rule under the hard constraint of the verification amplitude. Thus, the group verification and correction in step S5 has a clear input source, a clear matrix construction caliber, clear decomposition and projection rules, clear offset correction rules, and clear amplitude constraint rules, and reduces the risk of misleading the regional water budget benchmark quota calculation results due to systematic offsets caused by the same batch or the same operation and maintenance strategy. In practical applications: Taking the monthly target accounting cycle of the target area as an example, after outputting the verification mark in step S4, the system reads the metering processing set and obtains the metering device batch field from the metering device ledger, and obtains the maintenance cycle identifier, calibration method identifier, fault handling rule identifier, and data reporting mechanism identifier from the operation and maintenance system. Step S5-1 first divides the metering sampling points into multiple metering groups according to the metering device batch field and divides the target accounting cycle into multiple equal-length sub-periods according to the sub-period length configuration. Then, for each metering group, partitioning and summarizing are performed according to the summary caliber within each equal-length sub-period to obtain the group sub-period sequence and assemble it into a grouping matrix. Step S5-2 performs row centering and row scale normalization on the grouping matrix. After transformation, singular value decomposition is performed to obtain the master pattern vector. The row vector of each measurement group is then multiplied by the master pattern vector to obtain the group projection value, and its absolute value is used to form the group amplitude sequence. In step S5-3, the median of the group amplitude sequence is taken to obtain the common amplitude, which is then used to form an offset with the verification amplitude. The offset is then used to determine the correction direction of the measurement candidate quota according to the measurement direction to obtain the correction candidate quota. When the change of the correction candidate quota relative to the measurement base quota of the previous period exceeds the verification amplitude, the correction candidate quota is trimmed to the point where the change is equal to the verification amplitude according to the shrinkage rule. Finally, the shrunken correction candidate quota is output as the regional water budget base quota for this period, and the common amplitude, offset, and shrinkage result are written into the traceability record to support subsequent review and audit.

[0023] Further, please refer to the appendix to the instruction manual. Figure 2 Based on a dynamically adjustable regional water budget benchmark quota calculation method, this scheme also includes a dynamically adjustable regional water budget benchmark quota calculation system, which includes: The metering preprocessing module is used to obtain the metering dataset of the target area within the target accounting period, resample the metering dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points with their adjacent valid sampling points, and output the metering processed set. The metering summary module is used to perform partitioned summarization of the metering processing set according to the summary caliber of the target area and accumulate it to obtain the metering quota. The metering direction is used as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude is the absolute value of the difference. The module outputs the metering characteristics. The verification and aggregation module is used to acquire verification datasets from different sources on the acquisition link than the measurement dataset. It performs time alignment on the verification datasets according to the same time granularity as the measurement dataset and aggregates them to obtain the verification queue. The verification direction is the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude is the absolute value of the difference. The module outputs the verification features. The write judgment module is used to perform write judgment. When the measurement direction is the same as the verification direction and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period; otherwise, a verification mark is output. The grouping correction module is used to group the metering processing set according to the metering device batch or operation and maintenance strategy to obtain the metering group set when outputting the verification mark. For each metering group, it calculates its group quota and its group range. The median of each group range is taken as the common range, and the difference between the common range and the verification range is used as the offset. The offset is used to correct the metering quota to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of the current period.

[0024] Working principle: This scheme uses two independent links, measurement and verification, to calculate and verify the regional water budget baseline for the same accounting period. The overall process can be implemented collaboratively by edge computing nodes and the central side. The system first unifies the time granularity of the measurement dataset within the target accounting period and aligns it to the target time axis. For missing sampling points, it fills in the missing data by linear interpolation of adjacent valid sampling points to obtain the measurement processing set. Then, it summarizes and integrates the data according to the summary caliber to obtain the measurement candidate quota. At the same time, it compares the data with the measurement baseline of the previous period to obtain the measurement direction and measurement magnitude to form measurement features. In parallel, the system acquires verification datasets from different sources, aligns each verification source to the same time granularity, and iteratively updates the weights according to the inverse of the residual variance to generate a verification fusion sequence. It integrates and summarizes the data to obtain the verification candidate quota, and compares it with the verification baseline of the previous period to obtain the verification direction and verification magnitude to form verification features. Subsequently, the system obtains the converged verification amplitude range based on repeated sampling of the verification alignment sequence, and calculates the stable deviation amount by combining it with the historical verification amplitude sequence to expand it into the allowable amplitude range. When the measurement direction is consistent with the verification direction and the measurement amplitude falls within the allowable amplitude range, the measurement candidate amount is directly written as the regional water budget base amount for this period. Otherwise, a verification mark is output and group correction is initiated. The measurement processing set is grouped according to the batch of metering devices or the operation and maintenance strategy to construct a group matrix. The main pattern is extracted using singular value decomposition and projected to obtain the group amplitude. The median of the group amplitude is taken to obtain the common amplitude and the difference is calculated with the verification amplitude to obtain the offset correction measurement candidate amount. Finally, the change of the correction result relative to the measurement base amount of the previous period is limited to not exceeding the verification amplitude before the regional water budget base amount for this period is output. This writing judgment and group correction process can be triggered locally by the edge computing side and the results are reported. In practical applications, for example, when calculating the regional water use benchmark on a monthly basis in a certain region, there are instances of data loss and replenishment at metering sampling points. The metering side first completes time alignment and replenishment, and then summarizes and integrates the data to obtain the metering reserve and metering range. On the verification side, multiple verification sources from independent systems have clock deviations and local noise. The system first aligns and then iteratively merges these sources to obtain the verification reserve and verification range, and then forms an allowable range interval. If the metering range is within the allowable range interval and the direction is consistent, the metering reserve is directly used as the regional water budget benchmark for that month. If not, a verification flag is triggered. The system groups the metering sampling points according to batches or maintenance rules, uses the main mode projection to identify the common change intensity of most groups, calculates the common amplitude, and compares it with the verification amplitude to obtain the offset correction metering reserve. At the same time, the corrected change amplitude is controlled within the verification amplitude, thereby obtaining a writable and interpretable regional water budget benchmark for that month. Moreover, the above calculations can be completed locally at the edge computing node and transmitted back periodically.

[0025] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for calculating the benchmark amount of a dynamically adjusted regional water budget, characterized in that, include: S1. Obtain the measurement dataset of the target area within the target accounting period, resample the measurement dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points according to their adjacent valid sampling points, and output the measurement processing set. S2. Perform partitioned aggregation and summation on the metering processing set according to the aggregation caliber of the target area to obtain the metering quota, and use the metering direction as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude as the absolute value of the difference, and output the metering characteristics. S3. Obtain verification datasets from different sources on the acquisition link from the measurement dataset. Perform time alignment on the verification datasets according to the same time granularity as the measurement dataset and summarize them to obtain the verification queue. Use the verification direction as the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude as the absolute value of the difference. Output the verification features. S4. Execute the write judgment. When the measurement direction is the same as the verification direction, and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period. Otherwise, the verification mark is output.

2. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 1, characterized in that: S5. When outputting the verification mark, the metering processing set is grouped according to the metering device batch or operation and maintenance strategy to obtain the metering group set. For each metering group, its group quota and its group range are calculated. The median of each group range is taken as the common range, and the difference between the common range and the verification range is used as the offset. The metering quota is corrected using the offset to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of the current period.

3. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 2, characterized in that: The S5 also includes: S5-1. Taking the metering processing set as input, group the metering devices by batch or operation and maintenance strategy to obtain the metering group set, divide the target accounting cycle into multiple equal-length sub-periods, perform partitioning and summarizing for each metering group in each sub-period to obtain the group sub-period sequence, and form a grouping matrix according to the grouping order of each group sub-period sequence, and output the grouping matrix. S5-2. Taking the group matrix as input, perform singular value decomposition on the group matrix, take the left singular vector corresponding to the maximum singular value as the master mode vector, project the sub-time sequence of each measurement group onto the master mode vector to obtain the group projection value, and take the absolute value of the group projection value as the group amplitude of the measurement group, and output the group amplitude sequence. S5-3. Using the grouped amplitude sequence and the verification amplitude as input, the median of the grouped amplitude sequence is taken to obtain the common amplitude, and the difference between the common amplitude and the verification amplitude is used to determine the offset. The offset is used to correct the measurement reserve to obtain the corrected reserve. If the change of the corrected reserve relative to the measurement base of the previous period exceeds the verification amplitude, the corrected reserve is shrunk along the measurement base of the previous period towards the corrected reserve until the change is equal to the verification amplitude. The shrunk corrected reserve is output as the regional water budget base for this period.

4. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 3, characterized in that: The S1 also includes: S1-1. Perform statistics on the adjacent time intervals of each measurement sampling point in the measurement dataset and take the median as the unified time granularity. Generate a target time axis covering the target accounting cycle based on the unified time granularity and output the target time axis. S1-2. Perform mapping alignment on the target time axis for the measurement dataset. For multiple measurement sampling points falling into the same target time point, sum them up by time distance and normalize them to obtain the aligned sampling value of the target time point. Output the aligned measurement sequence. S1-3. For missing time points in the aligned measurement sequence, take the previous valid sample value and the next valid sample value and perform linear interpolation according to the time ratio to generate a supplementary sample value. When the missing span is greater than twice the uniform time granularity, mark the corresponding supplementary sample value as a low confidence supplementary value and output the measurement processing set.

5. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 4, characterized in that: The S2 also includes: S2-1. Determine the set of member measurement points for at least one partition according to the summary caliber of the measurement processing set, and perform time-by-time summation on the measurement sequence of each member measurement point in each partition at a uniform time granularity, and output the partition summary sequence. S2-2. Perform time integration on the summary sequence of each partition within the target accounting period and accumulate it to obtain the metering quota, and output the metering quota; S2-3. Calculate the difference between the metering quota and the metering base quota of the previous period, take its sign as the metering direction, take the absolute value of the difference as the metering amplitude, and output the metering characteristics.

6. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 5, characterized in that: The S3 also includes: S3-1. Construct a verification time axis with the same measurement time granularity for each verification source data in the verification dataset. Under the constraint of maintaining the time order and minimizing the time shift of adjacent mappings, solve the time mapping relationship that minimizes the sum of squared deviations between the mapped sequence of the verification source data and the predicted sequence obtained by interpolation of its adjacent time. Output the verification alignment sequence of each verification source. S3-2. Initialize the weights of each verification source based on the verification alignment sequence of each verification source and generate the verification fusion sequence. Calculate the residual sequence of each verification source and update the corresponding weights with the reciprocal of the residual variance. Repeat the generation of the verification fusion sequence and the updating of weights until the sum of the weight changes in two adjacent rounds is less than the convergence tolerance or the number of iterations reaches the maximum number of iterations. Output the converged verification fusion sequence. S3-3. Perform time integration on the verification fusion sequence within the target accounting period and summarize to obtain the verification candidate quota. Then, use the verification direction as the sign of the difference between the verification candidate quota and the verification base quota of the previous period, and the verification amplitude as the absolute value of the difference, and output the verification feature.

7. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 6, characterized in that: The S4 also includes: S4-1. Taking the verification alignment sequence as input, perform repeated sampling on the verification alignment sequence and calculate the verification candidate and verification range corresponding to each sampling. Take the lower quantile and upper quantile values ​​of the verification range to form the verification range interval. Stop repeated sampling when the difference between the interval widths of two adjacent rounds of verification range intervals is less than one-hundredth of the interval width of the previous round. Output the converged verification range interval. S4-2. Taking the verification range interval and the verification range sequence of the most recent accounting periods as input, calculate the median absolute deviation of the verification range sequence and use three times it as the stable deviation. Expand the verification range interval to both sides according to the stable deviation to obtain the allowable range interval, and output the allowable range interval.

8. The method for calculating the benchmark amount of a dynamically adjusted regional water budget according to claim 7, characterized in that: The S4 also includes: S4-3. Taking the measurement direction, measurement range, verification direction, and allowable range range as input, if the measurement direction and verification direction are the same and the measurement range falls within the allowable range range, the measurement reserve quota is output as the regional water budget base quota for this period. Otherwise, under the constraint that the adjustment coefficient is greater than or equal to 0 and less than or equal to 1, the adjustment coefficient that minimizes the absolute value of the difference between the adjusted range and the verification range and makes the adjusted range fall within the allowable range range is calculated, and the verification mark carrying the adjustment coefficient is output. The adjusted range is the product of the adjustment coefficient and the measurement range.

9. A dynamically adjustable regional water budget benchmark quota calculation system, characterized in that, include: The metering preprocessing module is used to obtain the metering dataset of the target area within the target accounting period, resample the metering dataset at a uniform time granularity and align it in time order, perform linear interpolation to fill in missing sampling points with their adjacent valid sampling points, and output the metering processed set. The metering summary module is used to perform partitioned summarization of the metering processing set according to the summary caliber of the target area and accumulate it to obtain the metering quota. The metering direction is used as the sign of the difference between the metering quota and the metering base of the previous period, and the metering amplitude is the absolute value of the difference. The module outputs the metering characteristics. The verification and aggregation module is used to acquire verification datasets from different sources on the acquisition link than the measurement dataset. It performs time alignment on the verification datasets according to the same time granularity as the measurement dataset and aggregates them to obtain the verification queue. The verification direction is the sign of the difference between the verification queue and the verification queue of the previous period, and the verification amplitude is the absolute value of the difference. The module outputs the verification features. The write judgment module is used to perform write judgment. When the measurement direction is the same as the verification direction and the measurement range meets the following conditions: the verification range minus the verification range multiplied by the proportional coefficient is not greater than the measurement range and the measurement range is not greater than the verification range plus the verification range multiplied by the proportional coefficient, the measurement reserve quota is output as the regional water budget base quota for this period; otherwise, a verification mark is output. The grouping correction module is used to group the metering processing set according to the metering device batch or operation and maintenance strategy to obtain the metering group set when outputting the verification mark. For each metering group, it calculates its group quota and its group range. The median of each group range is taken as the common range, and the difference between the common range and the verification range is used as the offset. The offset is used to correct the metering quota to obtain the corrected quota. The range of the corrected quota relative to the metering base of the previous period is limited to not exceeding the verification range. The corrected quota is output as the regional water budget base of the current period.

Images