Enterprise data collection and analysis method based on internet of things data collection terminal

Abstract

The application belongs to the technical field of Internet of Things, and specifically discloses an enterprise data collection and analysis method based on an Internet of Things data collection terminal, which comprises the following steps: extracting a material flow transfer path to generate a material flow transfer directed edge, generating a resource sharing coupling edge based on limited resource sharing, and constructing a directed relationship between projects; determining the weight coefficient on each edge according to the material flow transfer frequency, the overlap degree of the occupation time window and the occupation time length, identifying an associated project group through a closed loop; collecting energy consumption and fund data, comparing the data with a normal fluctuation interval after trend prediction, and identifying a suspected abnormal project; finding a directed edge with the suspected abnormal project as a downstream from an adjacency matrix, extracting an upstream deviation and obtaining a total conduction influence value through weighted accumulation; eliminating the total conduction influence value, and if the value is still higher than the upper limit of the interval after correction, an early warning is given. The application realizes accurate identification of abnormal signals and elimination of conduction interference, and avoids false alarms and resource mismatches.

Description

Technical Field

[0001] This invention belongs to the field of Internet of Things (IoT) technology and relates to a method for enterprise data collection and analysis based on IoT data acquisition terminals. Background Technology

[0002] With the rapid development and widespread application of IoT technology, IoT-based data acquisition terminals have gradually become an important infrastructure for enterprises to achieve information management and intelligent decision-making. In the scenario of multi-project parallel management in enterprises, each project generates a large amount of energy consumption data and financial data during operation. How to efficiently and accurately collect and analyze this multi-source heterogeneous data, and promptly identify projects with abnormal risks, has become an important issue for enterprises in refined management and risk control.

[0003] Existing technologies have already proposed relevant solutions. For example, Chinese invention patent CN115146977A discloses a method and system for enterprise energy efficiency data management based on the Internet of Things (IoT). This method collects data monitored by smart terminal devices and data acquisition devices, statistically analyzes, compares, and stores the collected data to obtain the actual operating data of the equipment, and then sends this data to a monitoring and management platform for user monitoring and management. This method enables enterprises to comprehensively monitor equipment energy efficiency and collect, share, manage, and optimize the scheduling of energy information, thus helping enterprises to save energy, reduce emissions, and lower energy costs to a certain extent.

[0004] However, the aforementioned existing technologies still have significant shortcomings:

[0005] First, this method only collects and monitors energy efficiency data for a single enterprise or a single device, without involving the analysis of the relationship between projects in the scenario of parallel management of multiple projects in an enterprise. As a result, it cannot identify whether there is a transmission path between projects when an anomaly occurs, which leads to the omission or misjudgment of project anomalies.

[0006] Second, this method can only perform independent statistical analysis and monitoring of equipment energy consumption data. It cannot quantify and eliminate the transmission impact between multiple projects. At the same time, it cannot distinguish whether the anomaly is caused by the project itself or by the deviation of upstream related projects transmitted through material flow or resource sharing relationship. This can easily lead to false alarms from management and misallocation of decision-making resources.

[0007] Therefore, there is an urgent need for a data collection and analysis method that can construct directed relationships between multiple projects within an enterprise, calculate and eliminate the transmission effects between projects, in order to accurately identify projects that truly have their own anomalies. Summary of the Invention

[0008] In view of this, in order to solve the problems mentioned in the background technology, a method for enterprise data collection and analysis based on IoT data collection terminals is proposed.

[0009] The objective of this invention can be achieved through the following technical solution: This invention provides an enterprise data collection and analysis method based on an Internet of Things (IoT) data collection terminal, including: constructing directed relationships between projects based on the material flow relationship and shared resource relationship between projects in the enterprise to be monitored.

[0010] Energy consumption data for each project is acquired through a data collection terminal, and financial data for each project is obtained through an interface with the enterprise management system.

[0011] The energy consumption data and funding data are time-aligned, and the multi-source data after time alignment is trend-predicted to obtain the energy consumption change trend value and funding change trend value of each project in the future time window. These values ​​are then compared with the corresponding historical benchmarks to identify suspected abnormal projects.

[0012] Based on the directed relationship between projects, the deviations in energy consumption and funding of projects that have upstream connections with the suspected abnormal project are retrieved. The transmission impact of the deviations on the suspected abnormal project is determined, and the transmission impact value is removed from the trend value of the suspected abnormal project for correlation correction.

[0013] If, after the correlation correction, the energy consumption change trend value or capital change trend value of the suspected abnormal project is still greater than or equal to the upper limit of the range, then it will be judged as an abnormal project and an early warning will be issued.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention generates directed edges of material flow by extracting material flow paths based on bill of materials and production work orders, and generates resource sharing coupling edges based on multiple projects calling the same limited resources, thereby constructing directed relationships between projects covering both material dependence and resource competition. At the same time, it identifies closed loops and merges projects with circular dependencies into related project groups, providing a complete basis for the transmission and tracking of abnormal signals, thereby avoiding abnormal misjudgment and omission due to lack of inter-project correlation analysis.

[0015] (2) This invention obtains the time series of each consumption of each project through the acquisition terminal, performs trend extrapolation on each, and obtains the energy consumption change trend value by combining according to the preset fusion rules. At the same time, it obtains the time series of each type of capital through the enterprise management system interface and obtains the capital change trend value in the same way, thus realizing the collaborative acquisition and trend fusion evaluation of multi-dimensional data of energy consumption and capital.

[0016] (3) This invention obtains the change trend value sequence of each project within the same historical time window, takes the maximum and minimum values ​​to construct the normal fluctuation range, compares the current trend value with the corresponding range to identify suspected abnormal projects, thereby quickly locking in risk projects that need to be focused on, avoiding the inefficiency caused by checking one by one in a large group of projects.

[0017] (4) This invention finds all the associated directed edges with the suspected abnormal project as the downstream node from the directed relationship between projects, extracts the deviation of each upstream associated project and multiplies it by the corresponding influence weight coefficient, and then sums them up to obtain the total transmission influence value of the upstream project on the suspected abnormal project, thus realizing the quantitative assessment of the abnormal transmission influence between projects.

[0018] (5) The present invention removes the total transmission impact value from the current trend value to obtain the corrected trend value, and compares the corrected trend value with the normal fluctuation range again. It only determines the abnormal item when the corrected trend value is still greater than or equal to the upper limit of the range, thereby eliminating abnormal interference from upstream transmission and reducing false alarms and misallocation of decision-making resources caused by indirect transmission. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram showing the connections between the steps of the method of the present invention.

[0021] Figure 2 This is a schematic diagram illustrating the steps for constructing directed relationships between items in this invention.

[0022] Figure 3 This is a schematic diagram illustrating the steps involved in determining the connections between the various related project groups of this invention. Detailed Implementation

[0023] 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.

[0024] Please see Figure 1 As shown, the present invention provides a method for enterprise data collection and analysis based on an Internet of Things (IoT) data acquisition terminal, specifically including the following steps S1 to S5:

[0025] S1. Establish directed relationships between projects

[0026] Please see Figure 2 As shown, directed relationships between projects are constructed based on the material flow and shared resource relationships among projects in the enterprise to be monitored. This includes the following steps S1-1 to S1-5.

[0027] For example, the directed relationships between the construction projects include:

[0028] S1-1, Mark project nodes and extract material flow paths

[0029] Each project within the monitored enterprise is marked as a project node. The corresponding bill of materials and production orders for each project are collected, and the actual material flow paths between projects are extracted. Specifically, for any two projects A and B, if the products or semi-finished products produced by project A are used as input materials for project B, then project A is the material outputter, and project B is the material receiver. Directed edges for material flow are generated, with the material outputter as the upstream node and the material receiver as the downstream node. These directed edges are unique and irreversible, representing the unidirectional dependency relationship of material flow between projects.

[0030] The bill of materials records the types, specifications, and quantities of raw materials, semi-finished products, parts, components, and finished products required for each project; the production work order records the planned output, start date, completion date, and process arrangement for each project, serving as the basic data source for determining the time nodes and flow direction of material requirements.

[0031] S1-2, Generate resource-sharing coupling edges

[0032] When multiple projects call the same finite resource, a resource-sharing coupling edge is generated between the projects calling the finite resource. The resource-sharing coupling edge is an undirected edge used to represent that two or more projects have a sharing relationship on the finite resource. As long as two projects call the same type of finite resource sequentially or simultaneously in the scheduling plan, a resource-sharing coupling edge is generated, regardless of the specific time order in which the two projects call the resource.

[0033] The limited resources mentioned above refer to production and management resources with capacity limits, exclusive occupancy, or scheduling constraints during enterprise operation, including but not limited to: equipment resources, energy resources, warehousing resources, logistics channel resources, and human resources. Equipment resources are tracked through manufacturing execution systems or equipment management systems to obtain equipment scheduling plans and real-time operating status; energy resource occupancy information is obtained through energy management systems to obtain energy usage quotas and actual consumption for each project; warehousing resource occupancy information is obtained through warehousing management systems to obtain warehousing space allocation and occupancy status for each project; logistics channel resource occupancy information is obtained through logistics management systems to obtain transportation schedules and channel usage plans for each project; and human resource occupancy information is obtained through enterprise resource planning systems or human resource management systems to obtain skilled personnel allocation and work hour arrangements for each project.

[0034] It should be noted that the energy resource occupancy information mentioned in this step is used to characterize the scheduling competition relationship between projects in terms of energy quotas, and its data source is the planned quota data of the energy management system; the energy consumption data mentioned in step S2 is the actual consumption data measured by the data acquisition terminal. The two data sources and uses are different and do not conflict with each other.

[0035] S1-3, Determine the influence weighting coefficients

[0036] Based on the material flow frequency of the directed edges in material flow and the time windows and durations of each project's occupation of the same limited resource corresponding to the resource-sharing coupling edges, the influence weight coefficients between project nodes are determined. The calculation of these influence weight coefficients is designed for ease of understanding, aiming to quantify the intensity of material dependence and resource competition among projects into directed edge weights that can participate in the calculation of the transmission impact value, providing a numerical basis for the subsequent quantitative assessment of abnormal transmission impacts. Specifically, this includes the following steps S1-3-1 to S1-3-5.

[0037] Furthermore, the determination of the influence weighting coefficients between each project node includes:

[0038] S1-3-1. Obtain the material flow frequency of the directed edge of the material flow within a preset statistical period, normalize it, and map it to the [0,1] interval to obtain the flow influence coefficient. The preset statistical period refers to the historical data statistical range used to calculate the influence weight coefficient, with its time direction pointing to the past, preferably the most recent 30 days, the most recent 90 days, or the most recent complete quarter. The material flow frequency refers to the number of times materials are sent from the upstream project to the downstream project within the preset statistical period.

[0039] Among these methods, the material flow frequency between all projects is normalized using Min-Max, mapping it to the [0,1] interval: In the formula, For material flow frequency, This represents the minimum frequency of material transfers between all projects. This represents the maximum frequency of material transfers between all projects. (Normalized) This refers to the flow influence coefficient of the directed edge of material flow. The closer the flow influence coefficient is to 1, the stronger the upstream project's dependence on the downstream project's material supply; the closer the value is to 0, the weaker the dependence. When the frequency of material flow is the same across all projects, the normalized flow impact coefficient is... The value is 1.

[0040] S1-3-2. Obtain the occupation time window and occupation duration of each item corresponding to the resource sharing coupling edge within a preset statistical period. Calculate the occupation time window overlap based on the occupation time window, and normalize the occupation duration. Then, perform a weighted summation of the normalized occupation duration and the occupation time window overlap to obtain the resource impact coefficient. The occupation time window includes the occupation start time and occupation end time.

[0041] Preferably, the occupation time is processed using the Min-Max normalization method, which maps it to the [0,1] interval. The overlap of the occupation time window is calculated by dividing the intersection length by the union length. Its value range is naturally [0,1] and it is a dimensionless quantity, so there is no need to perform normalization processing. It can directly participate in the weighted summation calculation.

[0042] The formula for calculating the resource impact coefficient is as follows: In the formula For resource impact coefficient, To account for the overlap of the time window, This is the normalized duration of use. and These are the weights for the overlap of the occupied time windows and the duration of occupation, used to quantify the different contributions of the degree of time overlap and the duration of occupation to the impact on resources. Preferably , .

[0043] By using weighted fusion to calculate the resource impact coefficient, on the one hand, the weight allocation can reflect the actual weights of the overlap of the occupation time window and the occupation duration on different dimensions of the impact of resource competition between projects, reflecting the difference in the contribution of the two factors to the degree of resource occupation; on the other hand, the two types of information, overlap and occupation duration, can be directly integrated to comprehensively consider the combined effect of the two on the resource occupation between projects.

[0044] The weights can be set based on the company's actual resource allocation strategy and project scheduling characteristics, or they can be obtained through fitting historical data. Preferably, production delay events caused by resource occupation among projects within a preset statistical period are statistically analyzed. The actual values ​​of the overlap of resource occupation time windows and the duration of resource occupation in each delay event are recorded. With the overlap and duration as independent variables and the degree of delay as the dependent variable, the contribution of the two to the impact of resource occupation is determined through multiple regression analysis. After normalization, the weights are converted into the weights of overlap and occupation duration, and the sum of the weights is 1. This allows for precise quantification of the resource impact coefficient.

[0045] Furthermore, the calculation of the time window overlap includes:

[0046] S1-3-2-1. Calculate the intersection and union lengths of the occupancy time windows of any two projects connected by the resource-sharing coupling edge within a preset statistical period, and use the ratio of these two lengths as the overlap of the occupancy time windows of the two projects. Wherein, when the intersection length is 0, the overlap of the occupancy time windows is 0. When a finite resource is shared by three or more projects, for all project nodes connected to the resource, calculate the overlap of the occupancy time windows between any two projects pairwise, and use the results of each pairwise calculation as input for calculating the resource influence coefficient between corresponding project pairs.

[0047] S1-3-3. For any two project nodes, when there is only a directed edge for material flow, the flow influence coefficient is used as the influence weight coefficient. When there are multiple directed edges for material flow between two project nodes, the maximum value of the flow influence coefficients corresponding to each directed edge for material flow is taken as the flow influence coefficient between the two nodes.

[0048] S1-3-4. When only resource-sharing coupling edges exist, the resource influence coefficient is used as the influence weight coefficient. Specifically, for resource-sharing coupling edges, the corresponding influence weight coefficients are assigned values ​​to the two directions that the two project nodes point to each other when stored in the adjacency matrix.

[0049] S1-3-5. When both directed edges for material flow and coupled edges for resource sharing exist simultaneously, the influence coefficients for flow and resources are weighted and summed to obtain the influence weight coefficient.

[0050] It should be added that the formula for calculating the influence weighting coefficient is as follows: In the formula The influence weight coefficient on the directed edge, The circulation impact coefficient, For resource impact coefficient, and These are the weights of the circulation impact coefficient and the resource impact coefficient, used to quantify the different contributions of material circulation to the degree of influence between projects due to resource occupation competition. Their specific values ​​can be pre-set by management personnel according to the actual production management characteristics of the enterprise and stored in the enterprise management system, participating in the calculation of the impact weight coefficient as fixed configuration parameters. Preferably, , .

[0051] By using weighted fusion to calculate the influence weight coefficients on each side, on the one hand, the weight allocation can reflect the actual weights of different dimensions of the influence of material flow relationship and resource sharing relationship on the inter-project relationship, reflecting the difference in the contribution of the two types of relationship to the inter-project dependence intensity; on the other hand, it can directly integrate the information of the two types of relationships, material flow and resource sharing, and comprehensively consider the combined effect of the two on the transmission influence between projects.

[0052] S1-4. Identify the relevant project teams

[0053] Based on the directed edges of material flow and the coupling edges of resource sharing, each associated project group is determined.

[0054] Please see Figure 3 As shown, further, determining each associated project group includes:

[0055] S1-4-1, Initialize the traversal

[0056] Using all directed edges of material flow for each project as the traversal object, a recursive stack is established. This recursive stack is a temporary data structure dynamically created during the depth-first traversal, used to record all project nodes on the current traversal path. Its lifespan only exists during the traversal process of identifying closed loops. Starting from any unvisited project node, recursively traversing is performed along the direction of the directed edges of material flow.

[0057] S1-4-2, Determining a Closed Loop

[0058] During the recursive traversal, each item node on the current traversal path is pushed onto the recursive stack in sequence. When traversing to the downstream node pointed to by any directed edge of the current node, if the downstream node already exists in the current recursive stack, it is determined that all item nodes between the traversed downstream node and the current node form a closed loop.

[0059] In one specific embodiment, if the traversal path is item A → item B → item C → item D, when traversing to the downstream node of item D, it is found that item B already exists in the recursion stack, then item B, item C, and item D form a closed loop.

[0060] S1-4-3. Closed loops are merged into related project groups.

[0061] When a closed loop is determined to exist, all project nodes constituting that closed loop are merged into an associated project group. A strongly connected component merging strategy is adopted: the largest set of nodes where any two nodes have a mutually reachable path are merged into the same associated project group. All project nodes within this associated project group are marked as visited. If multiple associated project groups share at least one project node, these multiple associated project groups are merged into a single associated project group.

[0062] S1-4-4, Backtracking the path that did not form a loop

[0063] If no closed loop is formed on the current traversal path, the current item node is marked as visited, and the process backtracks to the previous node, continuing to traverse other downstream branches of the current node. During backtracking, the visited mark of the current item node is retained to record that the node has been searched, but the node is popped from the recursion stack. After backtracking to the previous node, the process continues to traverse other untraversed downstream branches of that previous node. The recursion stack only records nodes on the current traversal path for closed loop detection, and the visited mark is used to record whether all global nodes have been searched.

[0064] For example, if project A participates in two closed loops simultaneously (loop 1: A→B→C→A; loop 2: A→D→E→A), then all project nodes (A, B, C, D, E) in the two closed loops are merged into the same associated project group.

[0065] S1-4-5, Mark the associated project group as visited

[0066] Mark each associated project group as visited, and its internal nodes will not be traversed repeatedly in subsequent traversals.

[0067] S1-4-6, Repeat until all nodes have been visited.

[0068] Repeat the above traversal process until all project nodes are marked as visited.

[0069] S1-5, Store in the adjacency matrix and construct the directed relationships between items.

[0070] Each associated project group and each project not included in any associated project group is treated as a node. The upstream and downstream relationships between nodes are determined by the material transfer direction. Specifically, for projects not included in any associated project group, their upstream and downstream nodes are determined by the direction of their directed material flow edges. For associated project groups, they are treated as a single node, with their upstream nodes being all upstream nodes of all projects within the group (excluding internal edges within the group), and their downstream nodes being all downstream nodes of all projects within the group (excluding internal edges within the group). The influence weight coefficients on each edge are stored in an adjacency matrix, and directed relationships between projects are constructed based on this adjacency matrix. In the adjacency matrix, rows represent upstream nodes, columns represent downstream nodes, and matrix elements contain the influence weight coefficients on each edge. If there is no direct directed relationship between two nodes, the corresponding matrix element value is set to 0. The adjacency matrix is ​​stored in the enterprise management system database in a two-dimensional array or sparse matrix format, allowing subsequent steps to quickly retrieve directed relationships and corresponding weight coefficients between any nodes using node indexes.

[0071] S2. Obtain energy consumption data and financial data.

[0072] Energy consumption data for each project is acquired through a data collection terminal, and financial data for each project is obtained through an interface with the enterprise management system.

[0073] S3. Identification of Suspected Abnormal Items

[0074] The energy consumption data and funding data are time-aligned, and the multi-source data after time alignment is trend-predicted to obtain the energy consumption change trend value and funding change trend value of each project in the future time window. These values ​​are then compared with the corresponding historical benchmarks to identify suspected abnormal projects.

[0075] For example, obtaining the energy consumption change trend value and the capital change trend value includes:

[0076] The energy consumption time series of each item within a preset sampling period is obtained from the energy consumption data. These consumption time series include, but are not limited to, electricity consumption time series, compressed air consumption time series, and cooling water consumption time series. For items without compressed air or cooling water monitoring capabilities, only the electricity consumption time series may be collected. Trend extrapolation is then performed on each item to obtain the predicted value series for each consumption within future time windows. Preferably, linear regression is used for trend extrapolation; if the historical data exhibits periodic or non-linear fluctuations, well-known time series forecasting methods (such as moving average or exponential smoothing) can also be used for trend forecasting. The specific forecasting method used can be selected based on the fitting accuracy of the historical data.

[0077] The preset sampling period refers to the fixed time length at which the IoT data acquisition terminal periodically reads consumption data from energy consumption monitoring points (such as smart meters, flow meters, etc.) of various projects at preset time intervals. This preset sampling period can be configured by administrators based on the data change rate. For example, for rapidly changing electricity consumption, it can be set to 1 minute / time; for relatively stable compressed air or cooling water consumption, it can be set to 5 minutes / time or 10 minutes / time. Those skilled in the art can flexibly set the above sampling period according to actual monitoring needs, and the specific values ​​do not constitute a limitation of the present invention.

[0078] In one specific embodiment, let the historical energy consumption data sequence be: ,in Number the time points, , For the first The energy consumption value at each time point, then the slope of the fitted straight line The calculation formula is: The slope Extending to future time windows, we obtain predicted values ​​for various future time points: ,in , This refers to the predicted number of points within a future time window. The length of the future time window can be set according to the actual management needs of the enterprise, preferably 7 days, 14 days, or 30 days.

[0079] Based on the predicted value sequence, the slope of each consumption amount changing over time is calculated as the trend value of each consumption amount.

[0080] The trend values ​​of each consumption amount are combined according to the preset fusion rules to obtain the energy consumption trend value of each project.

[0081] The preset fusion rule is as follows: when there is only one type of consumption with a positive trend value, the positive trend value is used as the energy consumption trend value.

[0082] When multiple consumption types have positive trend values, the trend value with the largest value is selected as the energy consumption trend value.

[0083] When all consumption type change trend values ​​are non-positive, the average of all change trend values ​​is taken as the energy consumption change trend value.

[0084] The capital change trend value of each project is calculated according to the method for obtaining the energy consumption change trend value.

[0085] Specifically, obtaining the trend values ​​of fund changes for each project within a future time window includes: obtaining the time series of each fund type for each project within a preset statistical period from the fund data, wherein the time series of each fund type includes, but is not limited to, the time series of actual expenditure amount, the time series of budget execution rate, and the time series of fund recovery progress.

[0086] By extrapolating the time series of each type of capital, we can obtain the predicted value series of each type of capital within the future time window.

[0087] Based on the predicted value sequence of each type of fund within the future time window, the slope of the change of each type of fund over time is calculated as the trend value of each type of fund.

[0088] The trend values ​​of the various funding types are combined according to a preset fusion rule to obtain the funding trend value for each project:

[0089] When only one type of capital has a positive trend value, that trend value is used as the capital trend value; when multiple types of capital have positive trend values, the trend value with the largest value is selected as the capital trend value; when all types of capital have non-positive trend values, the average of all trend values ​​is taken as the capital trend value.

[0090] It should be noted that a positive trend value for the actual expenditure amount indicates an increase in expenditure, a positive trend value for the budget execution rate indicates an increase in the execution rate, a negative trend value for the progress of fund recovery indicates a decrease in the recovery speed, and a positive trend value for the funds indicates an increase in financial pressure.

[0091] For example, the identification of suspected abnormal items includes:

[0092] Obtain the energy consumption and capital change trend sequences for each project across multiple historical time windows. The historical time window refers to a historical period with the same or similar time characteristics as the current point in time. For example, if the current period is the 30th week of 2026, data from the 30th week of 2025, the 30th week of 2024, and the 30th week of 2023 are used as historical time samples. If the data is less than three years, data from at least the previous year's same period can be used as a substitute. For projects with seasonality or periodicity, the historical time window can also be adjusted according to the project's actual operating cycle, such as selecting corresponding historical time periods by quarter, month, or specific production cycle.

[0093] The maximum and minimum values ​​in the sequence are determined respectively, and the normal fluctuation range of energy consumption and normal fluctuation range of funds for each project are constructed based on the interval between the minimum and maximum values.

[0094] The energy consumption and capital fluctuation trends of each project are compared with their corresponding normal fluctuation ranges for energy consumption and capital, respectively. If a project's energy consumption or capital fluctuation trend is outside its normal fluctuation range, the project is marked as a suspected abnormal project. Specifically, if the current energy consumption fluctuation trend is greater than or equal to the upper limit of the normal fluctuation range or lower than the lower limit, the project is suspected of being abnormal in terms of energy consumption; similarly, if the current capital fluctuation trend is greater than or equal to the upper limit of the normal fluctuation range or lower than the lower limit, the project is suspected of being abnormal in terms of capital.

[0095] In a preferred embodiment of the invention, an anomaly flag is triggered only when the value is greater than or equal to the upper limit of the interval (exceeding the positive value direction), while an anomaly flag is not triggered when the value is lower than the lower limit of the interval (improvement in the negative value direction). If the enterprise needs to monitor both bidirectional deviations simultaneously (i.e., excessive improvement may also indicate anomalies such as production interruptions), the judgment condition can be set to trigger an anomaly flag when the value is lower than the lower limit of the interval.

[0096] If no suspected abnormal items are marked after the above identification process, all current items are determined to be normal items, the process is terminated, and no further association correction and abnormal judgment steps are required.

[0097] S4, Correlation Correction

[0098] Based on the directed relationship between projects, the deviations in energy consumption and funding of projects that have upstream connections with the suspected abnormal project are retrieved. The transmission impact of the deviations on the suspected abnormal project is determined, and the transmission impact value is removed from the trend value of the suspected abnormal project for correlation correction.

[0099] For example, the calculation of the deviation includes:

[0100] From the directed relationships between the projects, all upstream related projects of the suspected anomalous project are obtained. If the suspected anomalous project belongs to a certain related project group, only the upstream nodes outside that related project group are considered as upstream related projects. It should be noted that when the suspected anomalous project belongs to a certain related project group, since the projects within the related project group form a closed loop due to material flow, the anomalous signal may circulate within the group, making it difficult to accurately distinguish the independent anomalous contribution of a single project. Therefore, only nodes outside the group are considered as upstream related projects, and other nodes within the group are not included in the deviation calculation to avoid duplicate or erroneous measurement caused by cyclic propagation.

[0101] The energy consumption trend value of each upstream related project is compared with the normal energy consumption fluctuation range. When the energy consumption trend value is greater than or equal to the upper limit of the normal energy consumption fluctuation range, the energy consumption deviation is obtained by subtracting the upper limit of the range from the energy consumption trend value. Otherwise, 0 is taken as the energy consumption deviation.

[0102] The reason for adopting the above method is that the purpose of calculating the transmission impact value in this invention is to quantify the deteriorating abnormal impact transmitted from the upstream project to the current project. Values ​​with a trend below the lower limit of the normal fluctuation range or within the range do not represent an abnormality in the direction of deterioration and do not need to be included as deviations in the calculation of the transmission impact value.

[0103] Similarly, the capital deviation is calculated using the same method as the energy consumption deviation.

[0104] Specifically, the calculation of the capital deviation includes:

[0105] The trend value of capital change of each upstream related project is compared with the normal fluctuation range of capital. When the trend value of capital change is greater than or equal to the upper limit of the normal fluctuation range of capital, the upper limit of the range is subtracted from the trend value of capital change to obtain the capital deviation. Otherwise, 0 is used as the capital deviation.

[0106] For example, determining the transmission impact value of a suspected abnormal item includes:

[0107] From the directed relationships between the projects, find all directed edges with suspected abnormal projects as downstream nodes and their corresponding upstream nodes. Record the found directed edges as each associated directed edge and the found upstream nodes as upstream associated projects.

[0108] Multiply the energy consumption deviation of the upstream associated project corresponding to each associated directed edge by the influence weight coefficient on that associated directed edge to obtain the energy consumption transmission influence value of each associated directed edge.

[0109] The energy consumption transmission impact values ​​of all associated directed edges are summed to obtain the total energy consumption transmission impact value received by the suspected abnormal project. When summing the energy consumption transmission impact values ​​of each associated directed edge, if multiple associated directed edges are connected to the same upstream associated project, they are calculated separately and then summed.

[0110] Similarly, the total capital transmission impact value is calculated using the same method as the total energy consumption transmission impact value.

[0111] It should be added that the calculation of the total capital transmission impact value includes: multiplying the capital deviation of the upstream related project corresponding to each related directed edge by the influence weight coefficient on the related directed edge to obtain the capital transmission impact value of each related directed edge.

[0112] The total impact value of fund transmission received by the suspected abnormal project is obtained by summing up the impact values ​​of fund transmission of all related directed edges.

[0113] S5. Abnormal Item Judgment and Warning

[0114] If, after correlation correction, the energy consumption change trend value or capital change trend value of the suspected abnormal project is still greater than or equal to the upper limit of the interval, it indicates that the abnormal situation of the project still exists after eliminating the upstream transmission influence, and it is then judged as an abnormal project and an early warning is issued.

[0115] In this embodiment, "deterioration" refers to an energy consumption correction trend value being greater than or equal to the upper limit of the normal energy consumption fluctuation range for the project, or a capital correction trend value being greater than or equal to the upper limit of the normal capital fluctuation range for the project. In other words, a project is considered to have its own anomaly only if, after eliminating the transmission effects of upstream related projects, the correction trend value of the suspected abnormal project itself still deviates from its historical normal fluctuation range.

[0116] For example, the abnormal item determination includes:

[0117] Obtain the energy consumption trend value and capital trend value of the suspected abnormal project. Subtract the total energy consumption transmission impact value from the energy consumption trend value to obtain the energy consumption correction trend value. At the same time, subtract the total capital transmission impact value from the capital trend value to obtain the capital correction trend value.

[0118] Compare the energy consumption correction trend value and the capital correction trend value with their normal fluctuation ranges for energy consumption and capital, respectively.

[0119] If the energy consumption correction trend value is greater than or equal to the upper limit of the normal fluctuation range of energy consumption or the capital correction trend value is greater than or equal to the upper limit of the normal fluctuation range of capital, then the suspected abnormal project is determined to be an abnormal project; otherwise, the suspected abnormal project is determined to be a normal project.

[0120] This invention, through steps S1 to S5, constructs a complete enterprise data collection and analysis method based on an IoT data acquisition terminal. By systematically constructing directed relationships between projects, quantifying the impact of anomaly transmission, and performing correlation correction, it provides reliable quantitative basis and automated means for anomaly identification and risk warning in multi-project parallel management scenarios.

[0121] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0122] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0123] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0124] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0125] Finally, the above description is only 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 enterprise data collection and analysis based on an Internet of Things (IoT) data acquisition terminal, characterized in that: The method includes: Based on the material flow and shared resource relationships among projects in the enterprise to be monitored, a directed relationship is constructed between projects; Energy consumption data for each project is acquired through data collection terminals, and financial data for each project is acquired through the enterprise management system interface. The energy consumption data and funding data are time-aligned, and the multi-source data after time alignment is trend-predicted to obtain the energy consumption change trend value and funding change trend value of each project in the future time window. These values ​​are then compared with the corresponding historical benchmarks to identify suspected abnormal projects. Based on the directed relationship between projects, the deviations in energy consumption and funding of projects that have upstream connections with the suspected abnormal project are retrieved. The transmission impact of the deviations on the suspected abnormal project is determined, and the transmission impact value is removed from the trend value of the suspected abnormal project for correlation correction. If, after the correlation correction, the energy consumption change trend value or capital change trend value of the suspected abnormal project still exceeds the corresponding normal fluctuation range, it will be judged as an abnormal project and an early warning will be issued.

2. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 1, characterized in that: The directed relationships between the construction projects include: Mark each project as a project node, collect the bill of materials and production work orders corresponding to each project, extract the actual material flow path between projects, and generate directed edges for material flow with the material output party as the upstream node and the material receiving party as the downstream node. When multiple projects call the same limited resource, a resource-sharing coupling edge is generated between the projects that call the limited resource; Based on the material flow frequency of the directed edges of material flow, and the time window and duration of each project's occupation on the same limited resource corresponding to the resource sharing coupling edges, the influence weight coefficient between each project node is determined. Based on the directed edges of material flow and the coupling edges of resource sharing, each associated project group is determined; Each associated project group and each project not included in any associated project group are treated as nodes. The upstream and downstream affiliation between each node is determined by the material transfer direction. The influence weight coefficients on each edge are stored in the adjacency matrix, and the directed relationship between projects is constructed based on the adjacency matrix.

3. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 2, characterized in that: The determination of the influence weighting coefficients between each project node includes: The material flow frequency of the directed edge of the material flow is obtained within a preset statistical period, and normalized to obtain the flow influence coefficient. Obtain the time window and duration of each item corresponding to the resource sharing coupling edge within a preset statistical period, calculate the overlap of the time window based on the time window, normalize the duration, and weight and sum the normalized duration and the overlap of the time window to obtain the resource impact coefficient. For any two project nodes, when there is only a directed edge for material flow, the flow impact coefficient is used as the impact weight coefficient. When only resource-sharing coupling edges exist, the resource influence coefficient is used as the influence weight coefficient; When both directed edges for material flow and coupled edges for resource sharing exist simultaneously, the influence coefficients of material flow and resource sharing are weighted and summed to obtain the influence weight coefficient.

4. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 3, characterized in that: The calculation of the time window overlap includes: The intersection and union lengths of the occupancy time windows of any two items connected by the resource-sharing coupling edge within a preset statistical period are calculated, and the ratio of the two is used as the overlap of the occupancy time windows of the two items.

5. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 2, characterized in that: The determination of each associated project group includes: Using all the directed edges of material flow for each project as the traversal objects, a recursive stack is established. Starting from any unvisited project node, the recursive traversal is performed along the direction of the directed edges of material flow. During the recursive traversal, each item node on the current traversal path is pushed onto the recursive stack in turn. When the downstream node pointed to by any directed edge of the current node is reached, if the downstream node already exists in the current recursive stack, it is determined that all item nodes between the traversed downstream node and the current node form a closed loop. When a closed loop is determined to be formed, all project nodes that constitute the closed loop are merged into an associated project group, and all project nodes contained in the associated project group are marked as visited. If multiple associated project groups share at least one project node, the multiple associated project groups are merged into the same associated project group. If no closed loop is formed on the current traversal path, mark the current project node as visited and backtrack to the previous node to continue traversing other downstream branches of the current node; Mark each associated project group as visited, and its internal nodes will not be traversed repeatedly in subsequent traversals; Repeat the above traversal process until all project nodes are marked as visited.

6. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 1, characterized in that: The acquisition of the energy consumption change trend value and the capital change trend value includes: The energy consumption data is used to obtain the time series of each item's consumption within a preset sampling period, and the trends are extrapolated to obtain the predicted value series of each consumption within a future time window. Based on the predicted value sequence, the slope of each consumption amount changing with time is calculated as the trend value of each consumption amount. The trend values ​​of each consumption amount are combined according to the preset fusion rules to obtain the energy consumption trend value of each project. The capital change trend value of each project is calculated according to the method for obtaining the energy consumption change trend value.

7. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 1, characterized in that: The items to be identified as potentially abnormal include: Obtain the energy consumption and capital change trend sequences for each project within multiple historical time windows; Determine the maximum and minimum values ​​in the sequence respectively, and construct the normal fluctuation range of energy consumption and normal fluctuation range of capital for each project based on the interval between the minimum and maximum values; When the energy consumption trend value of a project is not within its normal energy consumption fluctuation range, or the capital fluctuation value is not within its normal capital fluctuation range, the project will be marked as a suspected abnormal project.

8. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 1, characterized in that: The calculation of the deviation includes: Obtain all upstream related projects of the suspected abnormal project from the directed relationships between the projects; The energy consumption change trend value of each upstream related project is compared with the normal energy consumption fluctuation range. When the energy consumption change trend value is greater than or equal to the upper limit of the normal energy consumption fluctuation range, the energy consumption change trend value is subtracted from the upper limit of the range to obtain the energy consumption deviation. Otherwise, 0 is taken as the energy consumption deviation. Similarly, the capital deviation is calculated using the same method as the energy consumption deviation.

9. The enterprise data collection and analysis method based on an IoT data acquisition terminal according to claim 1, characterized in that: The determination of the transmission impact value of suspected abnormal items includes: From the directed relationships between the projects, find all directed edges with suspected abnormal projects as downstream nodes and their corresponding upstream nodes. Record the found directed edges as each associated directed edge and the found upstream nodes as upstream associated projects. Multiply the energy consumption deviation of the upstream associated project corresponding to each associated directed edge by the influence weight coefficient on that associated directed edge to obtain the energy consumption transmission influence value of each associated directed edge. The total energy consumption transmission impact value received by the suspected abnormal project is obtained by summing up the energy consumption transmission impact values ​​of all associated directed edges. Similarly, the total capital transmission impact value is calculated using the same method as the total energy consumption transmission impact value.

10. The enterprise data collection and analysis method based on an IoT data collection terminal according to claim 9, characterized in that: The determination of abnormal items includes: Obtain the energy consumption trend value and capital trend value of the suspected abnormal project. Subtract the total energy consumption transmission impact value from the energy consumption trend value to obtain the energy consumption correction trend value. At the same time, subtract the total capital transmission impact value from the capital trend value to obtain the capital correction trend value. If the energy consumption correction trend value is greater than or equal to the upper limit of the normal fluctuation range of energy consumption or the capital correction trend value is greater than or equal to the upper limit of the normal fluctuation range of capital, then the suspected abnormal project is determined to be an abnormal project; otherwise, the suspected abnormal project is determined to be a normal project.

Images