Energy consumption device detection data management system and method based on big data

Abstract

This invention discloses a data management system and method for energy consumption equipment detection based on big data, belonging to the field of data management technology. This invention monitors data in real time to determine whether operational condition coupling interference occurs; when operational condition coupling interference occurs, it identifies an abnormal correlation dataset; it collects the occupancy data of each task on the target shared resource in the abnormal correlation dataset during the interference period; it uses artificial intelligence algorithms to analyze the occupancy data, calculates the independent occupancy ratio of each task on the corresponding shared resource, and obtains an occupancy ratio set; it determines the task priority of each task; it sets a model objective and constructs an optimization model; it generates a set of excluding tasks based on the optimization model; it generates task adjustment instructions based on the set of excluding tasks and executes the exclusion operation on the excluding tasks; and it formulates a resource scheduling scheme based on task priority, remaining data collection requirements, and the current occupancy status of shared resources.

Description

Technical Field

[0001] This invention relates to the field of data management technology, specifically to a data management system and method for energy consumption equipment detection based on big data. Background Technology

[0002] With the development of industrial intelligence and the refinement of scientific research, the application of energy consumption equipment testing in product development, energy efficiency assessment, and quality control is becoming increasingly widespread. Laboratories, as the core setting for energy consumption equipment testing, are typically equipped with shared resources such as power supply, temperature control, and load simulation to handle multiple types and batches of testing tasks. Meanwhile, the application of big data technology in the acquisition, storage, and preliminary analysis of testing data has become an important support for improving testing efficiency; however, its adaptability in core aspects such as handling operating condition interference and optimizing resource scheduling still needs improvement.

[0003] Existing technologies often rely solely on single data thresholds to identify anomalies, failing to establish a causal chain between energy consumption equipment detection data and shared resource parameters. This easily leads to confusion between operating condition-coupled interference and equipment malfunctions or sensor errors, resulting in misclassification of anomaly types. In multi-task resource-sharing scenarios, it's difficult to isolate the cumulative effect of resource consumption between tasks, relying solely on experience to allocate resources, which provides insufficient decision-making basis for subsequent interference handling. When handling interference, a one-size-fits-all approach of suspending tasks or removing data is often used, which can easily lead to the loss of core data from high-priority tasks or excessive resource consumption by low-priority tasks, affecting overall detection efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide a data management system and method for energy consumption equipment detection based on big data, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: Firstly, this application provides a method for managing energy consumption equipment testing data based on big data, including the following steps: Real-time monitoring of energy-consuming equipment test data and laboratory shared resource operating parameters to determine whether operating condition coupling interference has occurred; when operating condition coupling interference occurs, identify abnormal associated datasets. Collect data on the usage of shared resources for each task in the abnormal correlation dataset during the interference period; use artificial intelligence algorithms to analyze the usage data, calculate the independent usage ratio of each task on the corresponding shared resources, and obtain a set of usage ratios; Retrieve preset task priority rules to determine the task priority of each task; set model objectives and construct an optimization model based on the occupancy ratio set and task priorities; generate a set of tasks that can be excluded based on the optimization model. Based on the set of excluding tasks, generate task adjustment instructions and execute exclusion operations on the excluding tasks; formulate resource scheduling schemes based on task priority, remaining data collection requirements, and the current occupancy status of shared resources.

[0006] In conjunction with the first aspect, in the first embodiment of the first aspect of this application, the real-time monitoring of energy consumption equipment detection data and laboratory shared resource operating parameters to determine whether operating condition coupling interference has occurred includes: A real-time data acquisition and synchronization link is established, and edge computing nodes are used to collect energy consumption equipment detection data and laboratory shared resource operation parameters from nearby locations. Data is uploaded via the MQTT protocol, and each data entry is bound with a unique identifier. Preprocessing operations are performed on the collected data. Establish benchmarks for energy-consuming equipment testing data, including the theoretical test value range of the equipment under standard operating conditions and the fluctuation range of normal test data for similar historical tasks; establish benchmarks for shared resource operating parameters, specifically by setting normal operating thresholds for shared resources based on laboratory testing specifications; compare the energy-consuming equipment testing data with the benchmarks for energy-consuming equipment testing data, and compare the laboratory shared resource operating parameters with the benchmarks for shared resource operating parameters to determine whether operating condition coupling interference has occurred.

[0007] In conjunction with the first aspect, in the second embodiment of the first aspect of this application, the step of using artificial intelligence algorithms to analyze the occupancy data and calculate the independent occupancy ratio of each task on the corresponding shared resource to obtain an occupancy ratio set includes: Using the nonnegative factorization matrix algorithm, let the mixed occupancy matrix X be decomposed into the product of two nonnegative matrices, specifically X = W × H, where W is the resource feature matrix, with rows corresponding to shared resource types and columns corresponding to task IDs. The element W(i,k) in W represents the basic occupancy coefficient of the k-th task for the i-th type of shared resource, reflecting the task's resource occupancy capability characteristics. H is the task temporal coefficient matrix, with rows corresponding to task IDs and columns corresponding to time slices. The element H(k,j) in H represents the occupancy intensity coefficient of the k-th task in the j-th time slice, reflecting the temporal dynamic changes of the task's resource occupancy. Set preset parameters, including iteration number threshold, convergence error threshold, and task number constraint; use Euclidean distance as the loss function; The model training and resource occupancy superposition effects are separated; the proportion of a single resource in a single time slice is calculated, the proportion of the entire time period is integrated, and a set of occupancy proportions is constructed.

[0008] In conjunction with the first aspect, in the third embodiment of the first aspect of this application, the step of separating the superimposed effects of model training and resource consumption includes: Input the constructed X into the nonnegative factorization matrix model, start iterative training according to the preset parameters, and alternately update the element values ​​of matrices W and H in each iteration until the loss function value is less than the convergence error threshold or the maximum number of iterations is reached, then stop training and obtain the optimal W and H. The independent resource usage of each task within each time slice is derived by inverse matrix multiplication. Specifically, the independent resource usage of each task k in time slice j is calculated. This process decomposes the mixed resource usage data, removes the superimposed impact of different tasks on the same resource in the same time slice, and obtains the independent contribution value of each task to the resource.

[0009] Specifically, the model iterative training is initiated. The predicted mixture matrix is ​​calculated by multiplying the current W and H matrices and compared with the original input matrix X. The deviation value is calculated using a loss function, which directly reflects the degree to which the current decomposition result matches the actual mixed resource occupancy data. Based on the deviation value, the elements of the W and H matrices are adjusted in reverse, employing an alternating update strategy to avoid convergence confusion caused by synchronous adjustments. First, the H matrix is ​​fixed, and the elements of the W matrix are optimized so that the product of W and H more closely reflects the resource occupancy characteristics in X. Then, the optimized W matrix is ​​fixed, and the elements of the H matrix are adjusted to match the occupancy intensity changes of each task in different time slices, maintaining the non-negativity of matrix elements after each update. This deviation calculation and matrix update process is repeated until the deviation value calculated by the loss function is less than the preset convergence error threshold, or the number of iterations reaches the maximum threshold. Training is then stopped, and the optimal W and H matrices at this point are output.

[0010] Furthermore, by multiplying the optimal W matrix and H matrix, the independent occupancy of a single task in a specific time slice and resource is derived in reverse. Specifically, the coefficient of a task and a resource in the W matrix is ​​multiplied by the coefficient of the task and a time slice in the H matrix. The result is the independent occupancy contribution of that task to that resource within that time slice, achieving task-by-task, time-by-time, and resource-by-resource decomposition of mixed data. For scenarios with the same time slice and the same resource, the sum of the independent occupancy contributions of all related tasks is consistent with the total occupancy at the corresponding position in the original mixed matrix X, and the independent contribution values ​​of each task do not interfere with each other. Through this decomposition, the originally superimposed total occupancy is broken down into the individual contribution of each task, completely eliminating the superimposed occupancy influence between different tasks and solving the problem of occupancy relationship confusion caused by multi-task concurrency. The deviation between the sum of the independent occupancy contributions of each task in each time slice and for each resource and the original total occupancy is checked one by one to ensure that the deviation is within an acceptable range, verify the accuracy of the decomposition results, and avoid distortion of independent contribution values ​​due to matrix update deviations.

[0011] In conjunction with the first aspect, in the fourth embodiment of the first aspect of this application, the step of setting a model objective and constructing an optimization model based on the set of occupancy ratios and task priorities includes: The specific objective of the model is to minimize the weighted sum of integrity loss across all tasks. The weights of the weighted sum correspond to the task priority protection weights, with higher priority tasks having a higher proportion of integrity loss in the weighted sum. The dimensions for determining integrity loss include the retention rate of core data items and the resource consumption of non-core data items. The more core data items retained, the lower the loss; the higher the resource consumption of non-core data items, the higher the loss. Set constraints; embed the model objective and constraints into the optimization model.

[0012] In conjunction with the first aspect, in the fifth embodiment of the first aspect of this application, the setting of constraints includes: Core data items must not be excluded; after the data item exclusion operation is completed, the total usage of all remaining data items for each type of shared resource must be controlled within the rated availability range of that resource; for tasks whose interference potential reaches a preset threshold, the exclusion ratio of their non-core data items must not be lower than the set standard; the decision attribute of each data item is limited to retention or exclusion, with no intermediate transition state.

[0013] In conjunction with the first aspect, in the sixth embodiment of the first aspect of this application, the step of generating a set of excluding task items based on the optimization model includes: The branch and bound method is used to solve the optimization model. Inputting the occupancy ratio set, task priority, data item attributes, and constraints, the optimal combination for retaining or excluding each data item is selected through iterative optimization. Result verification is performed, specifically checking whether the results fully comply with all constraints and are free of violations; assessing whether the overall integrity loss of the entire task is controlled within a preset acceptable range; if the results do not meet the standards, the leniency of the constraints is adjusted, and the solution process is restarted until the results meet the requirements; finally, a set of tasks that can be excluded is generated.

[0014] Specifically, for the initial node to be solved, the discrete constraints of the binary decision variables are relaxed, allowing the variables to take continuous values ​​between 0 and 1. The corresponding linear programming problem is then solved to obtain the initial upper and lower bounds. The upper bound is calculated by randomly generating a set of feasible solutions that satisfy the constraints, representing the upper limit of the current optimal possible value; the lower bound is determined by the optimal solution of the linear programming problem after the constraints are relaxed, representing the lower limit of the optimal possible value, thus initially locking in the range of optimal solution values.

[0015] From the initial node or the retained candidate nodes, prioritize the undetermined decision variable that has the greatest impact on the optimization objective as the branch point, and split the node into two child nodes: one child node forces the variable to be 1, indicating that the corresponding data item is retained, and the other forces it to be 0, indicating that the corresponding data item is excluded, so as to realize the stepwise decomposition of the problem.

[0016] For each child node, first check whether all constraints are met: if the constraints are violated, mark it as an infeasible node and discard it directly; if it is feasible, repeat the initial bounding logic, solve the linear programming problem after the constraints are relaxed, update the lower bound of the node, and optimize the global upper bound based on the current feasible solution to narrow the range of the optimal solution.

[0017] Compare the lower bound of all feasible child nodes with the global upper bound: if the lower bound of a child node is not less than the global upper bound, it means that the branch cannot obtain a better solution, so it is directly pruned and removed without further decomposition; if the lower bound is less than the global upper bound, the branch is retained as a candidate node for the next round to continuously optimize the solution efficiency.

[0018] From the remaining candidate nodes, select the node with the smallest lower bound and repeat the branching, bounding, and pruning process until all candidate nodes are pruned or decomposed into terminal nodes. At this point, select the combination with the smallest objective function value from all feasible terminal nodes; this is the optimal decision combination.

[0019] In conjunction with the first aspect, in the seventh embodiment of the first aspect of this application, the step of generating a task adjustment instruction based on the set of excluding task items and executing an exclusion operation on the excluding task items includes: Each task adjustment instruction corresponds to a set of data items to be excluded for a single task, including task ID, data item number and name to be excluded, data item storage path, exclusion operation timing requirements, resource release instructions, and exception handling plan. Adjustment instructions are generated for each task in order of task order. Instructions are reviewed and issued. After receiving the instruction, the execution end automatically reports the reception status and verifies the consistency between the local task data and the instruction requirements. When there are missing data items or mismatched paths, an exception warning is triggered and reported to the control end, and the operation is suspended for investigation. After verification, an execution confirmation receipt is generated, and the exclusion process is started. According to the instruction sequence requirements, the exclusion operation is executed in batches; after each set of data items is excluded, the task data index is updated, the index information of the excluded item is deleted, and a removal mark is added; the shared resource occupancy ledger is updated synchronously, the resource occupancy amount corresponding to the excluded item is deducted, and the resource release status is marked to ensure that the ledger is consistent with the actual resource occupancy; when the exclusion operation of multiple tasks involves the same shared resource, the exclusion operation of the low-priority task is executed only after the high-priority task's exclusion operation is completed and the resource is released.

[0020] In conjunction with the first aspect, in the eighth embodiment of the first aspect of this application, the step of formulating a resource scheduling scheme based on task priority, remaining data collection needs, and the current occupancy status of shared resources includes: High-priority tasks should be allocated high-quality resources first, while low-priority tasks should be avoided. Tasks with the same priority should be sorted according to their urgency. Resource allocation should match the task's collection needs to avoid resource overload or insufficient capacity, and a window should be reserved for core tasks to occupy resources continuously. Core resources should only be allocated to one task at a time. Prioritize the resource needs of high-priority tasks, allocate stable resources and lock the occupied time periods; allocate remaining resources to low-priority tasks and determine the upper limit of the occupied time; when resource competition occurs, prioritize the core collection needs, adjust the collection sequence of low-priority tasks, or split their non-core collection items to idle time periods; form a correspondence table of tasks, resources and occupied time periods, mark resource usage constraints and abnormal avoidance plans, and formulate resource scheduling schemes.

[0021] Secondly, this application provides a big data-based energy consumption equipment detection data management system, including: The abnormal correlation dataset generation module includes: a coupling interference determination unit that monitors energy consumption equipment test data and laboratory shared resource operation parameters in real time to determine whether operating condition coupling interference has occurred; and an abnormal correlation dataset generation unit that determines the abnormal correlation dataset when operating condition coupling interference occurs. Occupancy Ratio Set Generation Module: Includes: Occupancy Data Acquisition Unit: Acquires occupancy data of each task on the target shared resource in the abnormal correlation dataset during the interference occurrence period; Occupancy Ratio Calculation Unit: Analyzes the occupancy data using artificial intelligence algorithms, calculates the independent occupancy ratio of each task on the corresponding shared resource, and obtains the occupancy ratio set; The module for generating a set of tasks to be excluded includes: a task priority determination unit that retrieves preset task priority rules and determines the task priority of each task; an optimization model construction unit that sets a model objective and constructs an optimization model based on the occupancy ratio set and task priority; and a set of tasks to be excluded generation unit that generates a set of tasks to be excluded based on the optimization model. The resource scheduling scheme formulation module includes: an exclusion operation execution unit that generates task adjustment instructions based on the set of excludeable task items and executes exclusion operations on the excludeable task items; and a resource scheduling scheme formulation unit that formulates resource scheduling schemes based on task priorities, remaining data collection requirements, and the current occupancy status of shared resources.

[0022] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention establishes a correlation mapping and causal verification mechanism by real-time monitoring of energy consumption equipment detection data and shared resource operation parameters, which can accurately distinguish between operating condition coupling interference and other anomaly types, providing precise direction for subsequent processing and ensuring the effectiveness of anomaly handling.

[0023] 2. This invention uses artificial intelligence algorithms to remove the cumulative effect of resource occupation between tasks, accurately calculates the independent occupation ratio of each task for various shared resources, and forms a structured set of occupation ratios, providing data support for the construction of optimization models and solving the drawbacks of the experience-based allocation in existing technologies.

[0024] 3. This invention constructs an optimization model with the goal of minimizing the integrity loss of high-priority task detection, and generates a set of tasks that can be excluded. Only non-core and highly interfering data items are removed, which reduces resource conflicts and ensures that the core detection objectives of high-priority tasks are achieved, thereby improving the flexibility and targeting of task processing. Attached Figure Description

[0025] Figure 1 This is a schematic diagram illustrating the steps of the energy consumption equipment detection data management method based on big data according to the present invention; Figure 2 This is a system structure diagram of the energy consumption equipment detection data management system based on big data according to the present invention. Detailed Implementation

[0026] 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.

[0027] Example: Figures 1-2 As shown, the present invention provides a technical solution: like Figure 1 As shown, this application provides a method for managing energy consumption equipment detection data based on big data, including the following steps: Step S100: Monitor the energy consumption equipment test data and laboratory shared resource operation parameters in real time to determine whether operating condition coupling interference has occurred; when operating condition coupling interference occurs, determine the abnormal associated dataset; Specifically, a real-time data acquisition and synchronization link is established, and edge computing nodes are used to collect energy consumption equipment detection data and laboratory shared resource operation parameters from nearby locations. Data is uploaded via the MQTT protocol, and each data item is bound with a unique identifier. The collected data is preprocessed. Establish benchmarks for energy-consuming equipment testing data, including the theoretical test value range of the equipment under standard operating conditions and the fluctuation range of normal test data for similar historical tasks; establish benchmarks for shared resource operating parameters, specifically by setting normal operating thresholds for shared resources based on laboratory testing specifications; compare the energy-consuming equipment testing data with the benchmarks for energy-consuming equipment testing data, and compare the laboratory shared resource operating parameters with the benchmarks for shared resource operating parameters to determine whether operating condition coupling interference has occurred.

[0028] In one specific embodiment, this embodiment takes the parallel testing of two pieces of equipment in the laboratory as the scenario, namely the energy consumption testing task of 1# three-phase asynchronous motor (rated power 7.5kW) and the energy consumption testing task of 2# centrifugal water pump (rated power 2.2kW). The two share the laboratory power supply system and temperature control system, simulating the entire process of determining the coupling interference under the working conditions.

[0029] Two edge computing nodes are deployed next to the motor test bench and the water pump test bench to collect two types of data nearby: energy consumption equipment detection data, including the energy consumption values ​​corresponding to the motor stator current, input power, and water pump outlet pressure, collected once per second; and shared resource parameters, including power supply circuit voltage and ambient temperature, collected once every 0.5 seconds. The data is encrypted and uploaded to the core processing unit via the MQTT protocol, and each data is bound to a unique identifier in the format of "task ID-device number-collection timestamp". For example, the current data collected at 10:05:00.000 for motor #1 is identified as "T001-M001-20260125100500000".

[0030] Preprocessing was performed on the collected data: noise filtering eliminated isolated anomalies, such as a momentary erroneous collection value of 15A (far exceeding the normal range) in motor current, which was directly filtered out; data standardization unified units, converting motor power from "watts" to "kilowatts," and retaining one decimal place for temperature; synchronous calibration of timing deviations ensured accurate correspondence between device data and shared resource data at the same timestamp. After preprocessing, the motor current stabilized in the 8.2-8.5A range, the water pump energy consumption stabilized in the 1.8-1.9kW range, and the initial power supply voltage stabilized at 380V±0.5V.

[0031] The specific benchmarks for energy consumption equipment testing data are as follows: For motor #1, under standard operating conditions (75% rated load, ambient temperature 25℃), the theoretical range of input power is 5.2-5.6kW, and the normal fluctuation range for similar historical tasks is ±0.3kW; For pump #2, under standard operating conditions, the theoretical range of energy consumption is 1.7-1.9kW, and the historical fluctuation range is ±0.1kW.

[0032] The specific benchmark parameters for shared resource operation are as follows: According to laboratory testing standards, the normal voltage threshold of the power supply system is 380V±3% (368.6-391.4V), and the ambient temperature threshold of the temperature control system is 25℃±2℃ (23-27℃).

[0033] At 10:08, an anomaly was observed in the data comparison: the input power of motor #1 rose to 5.9kW, exceeding the historical fluctuation limit by 0.3kW, and remained there for one minute without decreasing; simultaneously, the supply voltage dropped to 367.2V, 1.4V below the lower limit of the supply threshold, and the ambient temperature rose to 27.8℃, exceeding the upper limit of the temperature control threshold by 0.8℃. Further investigation revealed that both the voltage fluctuation and the excessive temperature were caused by the simultaneous activation of the load simulation module by two devices. The parallel use of shared resources caused a coupling effect, and the timing of the device data anomalies and the anomalies in the shared resource parameters were completely synchronized, indicating that a condition coupling interference had occurred.

[0034] The abnormal correlation dataset was then identified, which included 180 structured data entries for the full energy consumption data, power supply voltage, and ambient temperature parameters of tasks T001 and T002 during the period from 10:08 to 10:09.

[0035] Step S200: Collect the occupancy data of each task on the target shared resource in the abnormal correlation dataset during the interference period; use artificial intelligence algorithms to analyze the occupancy data, calculate the independent occupancy ratio of each task on the corresponding shared resource, and obtain the occupancy ratio set; Specifically, using the nonnegative factorization matrix algorithm, let the mixed occupancy matrix X be decomposed into the product of two nonnegative matrices, specifically X = W × H, where W is the resource feature matrix, with rows corresponding to shared resource types and columns corresponding to task IDs. The element W(i,k) in W represents the basic occupancy coefficient of the k-th task for the i-th type of shared resource, reflecting the task's resource occupancy capability characteristics. H is the task temporal coefficient matrix, with rows corresponding to task IDs and columns corresponding to time slices. The element H(k,j) in H represents the occupancy intensity coefficient of the k-th task in the j-th time slice, reflecting the temporal dynamic changes of the task's resource occupancy. Set preset parameters, including iteration number threshold, convergence error threshold, and task number constraint; use Euclidean distance as the loss function; The model training and resource occupancy superposition effects are separated; the proportion of a single resource in a single time slice is calculated, the proportion of the entire time period is integrated, and a set of occupancy proportions is constructed.

[0036] Furthermore, the constructed X is input into the nonnegative factorization matrix model, and iterative training is started according to the preset parameters. During each iteration, the element values ​​of matrices W and H are updated alternately until the loss function value is less than the convergence error threshold or the maximum number of iterations is reached, at which point the training stops and the optimal W and H are obtained. The independent resource usage of each task within each time slice is derived by inverse matrix multiplication. Specifically, the independent resource usage of each task k in time slice j is calculated. This process decomposes the mixed resource usage data, removes the superimposed impact of different tasks on the same resource in the same time slice, and obtains the independent contribution value of each task to the resource.

[0037] In one specific embodiment, based on the abnormal correlation dataset of step S100, the focus is on the two types of target shared resources, namely the power supply system and the temperature control system. Resource occupancy data of T001 (motor task) and T002 (water pump task) during the period from 10:08 to 10:09 are collected. The data is divided into 6 time slices with 10 seconds as each time slice. The mixed occupancy of the two types of resources is collected in each time slice.

[0038] A nonnegative matrix factorization algorithm is used to construct the model. The mixed occupancy matrix X is set to 2 rows and 6 columns (rows correspond to power supply and temperature control resources, and columns correspond to 6 time slices). The elements in X are the mixed occupancy values ​​of the corresponding resources within the time slices (power supply unit: kW, temperature control unit: kW·h), specifically X = [[5.8, 6.1, 5.9, 6.0, 5.7, 5.9], [2.1, 2.3, 2.2, 2.4, 2.2, 2.3]]. Matrix X is decomposed into W × H, where W is 2 rows and 2 columns (rows correspond to resources, columns correspond to tasks), and H is 2 rows and 6 columns (rows correspond to tasks, columns correspond to time slices).

[0039] Set the preset parameters: 500 iterations, 1e-6 convergence error, and 2 tasks (corresponding to T001 and T002). Use Euclidean distance as the loss function to measure the decomposition deviation.

[0040] Start model training, alternately updating the values ​​of W and H elements. When the loss function value drops to 8.2e-7 (less than the convergence threshold) after the 320th iteration, training stops. The optimal W matrix is ​​obtained: W=[[0.82,0.35],[0.38,0.21]], where W(1,1)=0.82 represents the basic occupancy coefficient of power supply resources for T001, and W(2,1)=0.38 represents the basic occupancy coefficient of temperature control resources for T001; the optimal H matrix element range is 0.65-0.92, reflecting the fluctuation of the occupancy intensity of the two tasks in each time slice.

[0041] By reverse-engineering the independent occupancy rate using W×H, the proportion of a single resource per time slice is calculated. For example, in time slice 1, T001's occupancy rate for power supply resources is (0.82×0.71) / (0.82×0.71+0.35×0.68)≈69%, while T002's is 31%. Integrating the proportions of the six time slices, a weighted average based on time slice duration is obtained, resulting in the following set of occupancy rates: T001 accounts for 68% of power supply resources and 62% of temperature control resources; T002 accounts for 32% of power supply resources and 38% of temperature control resources.

[0042] Step S300: Retrieve the preset task priority rules and determine the task priority of each task; based on the occupancy ratio set and task priority, set the model objective and construct the optimization model; generate a set of tasks that can be excluded based on the optimization model; Specifically, the model objective is to minimize the weighted sum of integrity loss across all tasks. The weights of the weighted sum correspond to the task priority protection weights, with higher priority tasks having a higher proportion of integrity loss in the weighted sum. The dimensions for determining integrity loss include the retention rate of core data items and the resource consumption of non-core data items. The more core data items retained, the lower the loss; the higher the resource consumption of non-core data items, the higher the loss. Set constraints; embed the model objective and constraints into the optimization model.

[0043] Furthermore, core data items are prohibited from being excluded; after the data item exclusion operation is completed, the total usage of all remaining data items for each type of shared resource must be controlled within the rated availability range of that resource; for tasks whose interference potential reaches a preset threshold, the exclusion ratio of their non-core data items is required to be no less than the set standard; the decision attribute of each data item is limited to retention or exclusion, with no intermediate transition state.

[0044] Furthermore, the branch and bound method is used to solve the optimization model. The input includes the set of occupancy ratios, task priorities, data item attributes, and constraints. Through iterative optimization, the optimal combination for retaining or excluding each data item is selected. Result verification is then performed, specifically checking whether the results fully comply with all constraints and are free of violations. The overall integrity loss of the entire task is assessed to ensure it is within a preset acceptable range. If the results do not meet the standards, the leniency of the constraints is adjusted, and the solution process is restarted until the results meet the requirements. Finally, a set of tasks that can be excluded is generated.

[0045] In one specific embodiment, based on the set of occupancy ratios obtained in S200, and combined with the laboratory's preset task priority rules, model construction and solution are carried out to generate a set of tasks that can be excluded, and the T001 (motor task) and T002 (water pump task) scenarios are used throughout the process.

[0046] First, the priority rules are retrieved. T001 is a product type testing task with high accuracy requirements, so it is designated as Level 1 (high priority); T002 is a routine calibration task, designated as Level 2 (low priority). The protection weights are quantified, with Level 1 set to 2.0 and Level 2 set to 1.0, to ensure that the integrity loss of high-priority tasks accounts for a higher proportion in the weighted sum.

[0047] The model objectives and integrity loss judgment dimensions are set as follows: the objective is to minimize the weighted sum of integrity losses across all tasks, with a retention rate weight of 0.7 for core data items and a resource consumption weight of 0.3 for non-core data items. Data items are split into two tasks: T001 core items are rated load steady-state power data (8 items), and non-core items are start-stop transition energy consumption data (5 items, 0.3-0.5kW per power supply line and 0.1-0.2kW·h for temperature control); T002 core items are energy consumption data corresponding to outlet pressure (6 items), and non-core items are environmental adaptation energy consumption data (4 items, 0.1-0.2kW per power supply line and 0.05-0.1kW·h for temperature control).

[0048] Set constraints: Core data items cannot be excluded; the rated available capacity of the power supply system is 6.5kW and the temperature control system is 3.0kW·h; the interference potential threshold is set to 50%, T001 (power supply ratio 68%, temperature control 62%) meets the standard, and the exclusion ratio of its non-core items is required to be no less than 40%; all data items are only given a binary decision attribute of "retain / exclude".

[0049] The objective and constraints are embedded in the optimization model, and the branch and bound method is used for solution. The inputs are the occupancy ratio set, priority weights, and data item attributes. The iteration threshold is set to 1000 times, and the convergence error is 1e-6. The optimal combination is obtained at the 410th iteration: T001 excludes 2 non-core start-stop data items (accounting for 40% of non-core items, releasing 0.8kW of power supply and 0.3kW·h of temperature control), and T002 excludes 1 non-core environmental adaptation data item (releasing 0.15kW of power supply and 0.08kW·h of temperature control).

[0050] Results verification: No core items were excluded. The remaining data consumed a total of 5.4kW of power and 1.9kW·h of temperature control, both below the rated values. The overall integrity loss of the entire task was 8%, which was controlled within the preset range of 10%. Finally, an exclusion set was generated, including two start-stop transition data items from T001 and one environmental adaptation data item from T002. The exclusion criteria were marked as "non-core + high interference / resource consumption".

[0051] Step S400: Based on the set of tasks that can be excluded, generate task adjustment instructions and execute exclusion operations on the tasks that can be excluded; based on task priority, remaining data collection requirements and the current occupancy status of shared resources, formulate a resource scheduling plan.

[0052] Specifically, each task adjustment instruction corresponds to a set of data items to be excluded for a single task, including the task ID, the number and name of the data item to be excluded, the data item storage path, the timing requirements for the exclusion operation, resource release instructions, and anomaly handling contingency plans. Adjustment instructions are generated for each task in order of task order. The instructions are reviewed and issued. After receiving the instructions, the execution end automatically reports the reception status and verifies the consistency between the local task data and the instructions. If there are missing data items or mismatched paths, an anomaly warning is triggered and reported to the control end, and the operation is suspended pending investigation. After verification, an execution confirmation receipt is generated, and the exclusion process is started. According to the instruction sequence requirements, the exclusion operation is executed in batches; after each set of data items is excluded, the task data index is updated, the index information of the excluded item is deleted, and a removal mark is added; the shared resource occupancy ledger is updated synchronously, the resource occupancy amount corresponding to the excluded item is deducted, and the resource release status is marked to ensure that the ledger is consistent with the actual resource occupancy; when the exclusion operation of multiple tasks involves the same shared resource, the exclusion operation of the low-priority task is executed only after the high-priority task's exclusion operation is completed and the resource is released.

[0053] Furthermore, high-priority tasks are allocated high-quality resources first, while low-priority tasks are avoided, and tasks with the same priority are sorted according to their urgency. Resource allocation must match the task's collection needs to avoid resource overload or insufficient capacity, and a window for continuous use by core tasks should be reserved. Core resources are allocated to only one task at a time. Prioritize the resource needs of high-priority tasks, allocate stable resources and lock the occupied time periods; allocate remaining resources to low-priority tasks and determine the upper limit of the occupied time; when resource competition occurs, prioritize the core collection needs, adjust the collection sequence of low-priority tasks, or split their non-core collection items to idle time periods; form a correspondence table of tasks, resources and occupied time periods, mark resource usage constraints and abnormal avoidance plans, and formulate resource scheduling schemes.

[0054] In one specific embodiment, tasks are prioritized (T001 takes precedence), and dedicated adjustment instructions are generated one by one. The core information of the T001 instruction is: Task ID=T001, items to be excluded numbered S001 and S002 (named "Start / Stop Transition Energy Consumption Data"), storage path " / data / T001 / non-core / 202601251008 / ", timing requirement "Complete exclusion before 10:10:00", resource release description "Release power supply 0.8kW, temperature control 0.3kW·h", and contingency plan "Pause operation if path mismatch occurs, and provide feedback to the management terminal within 10 seconds". The T002 instruction is: Task ID=T002, item to be excluded numbered H003 (named "Environment Adaptation Energy Consumption Data"), storage path " / data / T002 / non-core / 202601251008 / ", timing requirement "Execute after T001 exclusion is completed", release power supply 0.15kW, temperature control 0.08kW·h. After being double-verified, the instructions are sent to the corresponding edge nodes via the MQTT protocol, and the sending record is retained.

[0055] At 10:09:30, the T001 execution terminal received the instruction, verified that the local data matched the instruction, and sent back a confirmation receipt. At 10:09:45, it completed the removal of two data items and updated the task index synchronously, marking "Removal time 10:09:45, Instruction number CMD001". The shared resource ledger was also updated synchronously, with power supply usage decreasing from 5.4kW to 4.6kW and temperature control from 1.9kW·h to 1.6kW·h. After the T001 resource release was completed, the T002 execution terminal started the operation, and at 10:10:00, the removal of H003 was completed. The ledger was finally updated to power supply 4.45kW and temperature control 1.52kW·h, both below the rated values, with no residual resource usage.

[0056] Based on task priority, remaining data acquisition requirements (T001 needs to acquire 8 core steady-state data points, and T002 needs to acquire 5 core data points), and resource status, the following plan is formulated: T001 is prioritized for allocation to the laboratory's #1 power supply circuit (voltage stability ±0.3V, high-quality resource), with the 10:12-10:32 time slot (20-minute continuous acquisition window). The temperature control system will prioritize ensuring the temperature in its test area remains stable at 25℃±0.5℃. T002 is allocated to the #2 power supply circuit (remaining resource), occupying the 10:33-10:43 time slot (maximum duration 10 minutes), sharing the remaining temperature control adjustment capacity. There are no resource contention conflicts. The constraint "#1 circuit is exclusively for T001; T002 must not occupy it" is marked, and an anomaly avoidance plan is attached (prioritizing T001's data acquisition during power fluctuations). This plan is now in effect.

[0057] like Figure 2 As shown, this application provides a big data-based energy consumption equipment detection data management system, including: The abnormal correlation dataset generation module includes: a coupling interference determination unit that monitors energy consumption equipment test data and laboratory shared resource operation parameters in real time to determine whether operating condition coupling interference has occurred; and an abnormal correlation dataset generation unit that determines the abnormal correlation dataset when operating condition coupling interference occurs. Occupancy Ratio Set Generation Module: Includes: Occupancy Data Acquisition Unit: Acquires occupancy data of each task on the target shared resource in the abnormal correlation dataset during the interference occurrence period; Occupancy Ratio Calculation Unit: Analyzes the occupancy data using artificial intelligence algorithms, calculates the independent occupancy ratio of each task on the corresponding shared resource, and obtains the occupancy ratio set; The module for generating a set of tasks to be excluded includes: a task priority determination unit that retrieves preset task priority rules and determines the task priority of each task; an optimization model construction unit that sets a model objective and constructs an optimization model based on the occupancy ratio set and task priority; and a set of tasks to be excluded generation unit that generates a set of tasks to be excluded based on the optimization model. The resource scheduling scheme formulation module includes: an exclusion operation execution unit that generates task adjustment instructions based on the set of excludeable task items and executes exclusion operations on the excludeable task items; and a resource scheduling scheme formulation unit that formulates resource scheduling schemes based on task priorities, remaining data collection requirements, and the current occupancy status of shared resources.

[0058] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for managing energy consumption equipment detection data based on big data, characterized in that, Includes the following steps: Real-time monitoring of energy-consuming equipment test data and laboratory shared resource operating parameters to determine whether operating condition coupling interference has occurred; When operating condition coupling disturbances occur, identify the abnormal associated dataset; Collect data on the usage of shared resources of the target by each task in the abnormal correlation dataset during the period of interference; Artificial intelligence algorithms are used to analyze the occupancy data, calculate the independent occupancy ratio of each task on the corresponding shared resources, and obtain a set of occupancy ratios; Retrieve the preset task priority rules and determine the task priority of each task; Based on the occupancy ratio set and task priority, set the model objective and construct an optimization model; generate a set of tasks that can be excluded based on the optimization model. Based on the set of excluding tasks, generate task adjustment instructions and execute exclusion operations on the excluding tasks; formulate resource scheduling schemes based on task priority, remaining data collection requirements, and the current occupancy status of shared resources.

2. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The real-time monitoring of energy consumption equipment detection data and laboratory shared resource operating parameters are used to determine whether operational condition coupling interference has occurred, including: A real-time data acquisition and synchronization link is established, and edge computing nodes are used to collect energy consumption equipment detection data and laboratory shared resource operation parameters from nearby locations. Data is uploaded via the MQTT protocol, and each data entry is bound with a unique identifier. Preprocessing operations are performed on the collected data. Establish benchmarks for energy-consuming equipment testing data, including the theoretical test value range of the equipment under standard operating conditions and the fluctuation range of normal test data for similar historical tasks; establish benchmarks for shared resource operating parameters, specifically by setting normal operating thresholds for shared resources based on laboratory testing specifications; compare the energy-consuming equipment testing data with the benchmarks for energy-consuming equipment testing data, and compare the laboratory shared resource operating parameters with the benchmarks for shared resource operating parameters to determine whether operating condition coupling interference has occurred.

3. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The method employs artificial intelligence algorithms to analyze the occupancy data, calculates the independent occupancy ratio of each task on the corresponding shared resources, and obtains a set of occupancy ratios, including: Using the nonnegative factorization matrix algorithm, let the mixed occupancy matrix X be decomposed into the product of two nonnegative matrices, specifically X = W × H, where W is the resource feature matrix, with rows corresponding to shared resource types and columns corresponding to task IDs. The element W(i,k) in W represents the basic occupancy coefficient of the k-th task for the i-th type of shared resource, reflecting the task's resource occupancy capability characteristics. H is the task temporal coefficient matrix, with rows corresponding to task IDs and columns corresponding to time slices. The element H(k,j) in H represents the occupancy intensity coefficient of the k-th task in the j-th time slice, reflecting the temporal dynamic changes of the task's resource occupancy. Set preset parameters, including iteration number threshold, convergence error threshold, and task number constraint; use Euclidean distance as the loss function; The model training and resource occupancy superposition effects are separated; the proportion of a single resource in a single time slice is calculated, the proportion of the entire time period is integrated, and a set of occupancy proportions is constructed.

4. The method for managing energy consumption equipment detection data based on big data according to claim 3, characterized in that, The process of separating the combined effects of model training and resource consumption includes: Input the constructed X into the nonnegative factorization matrix model, start iterative training according to the preset parameters, and alternately update the element values ​​of matrices W and H in each iteration until the loss function value is less than the convergence error threshold or the maximum number of iterations is reached, then stop training and obtain the optimal W and H. The independent resource usage of each task within each time slice is derived by inverse matrix multiplication. Specifically, the independent resource usage of each task k in time slice j is calculated. This process decomposes the mixed resource usage data, removes the superimposed impact of different tasks on the same resource in the same time slice, and obtains the independent contribution value of each task to the resource.

5. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The process of setting model objectives and constructing an optimization model based on the set of occupancy ratios and task priorities includes: The specific objective of the model is to minimize the weighted sum of integrity loss across all tasks. The weights of the weighted sum correspond to the task priority protection weights, with higher priority tasks having a higher proportion of integrity loss in the weighted sum. The dimensions for determining integrity loss include the retention rate of core data items and the resource consumption of non-core data items. The more core data items retained, the lower the loss; the higher the resource consumption of non-core data items, the higher the loss. Set constraints; embed the model objective and constraints into the optimization model.

6. The method for managing energy consumption equipment detection data based on big data according to claim 5, characterized in that, The set constraints include: Core data items must not be excluded; after the data item exclusion operation is completed, the total usage of all remaining data items for each type of shared resource must be controlled within the rated availability range of that resource; for tasks whose interference potential reaches a preset threshold, the exclusion ratio of their non-core data items must not be lower than the set standard; the decision attribute of each data item is limited to retention or exclusion, with no intermediate transition state.

7. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The generation of the set of tasks to be excluded based on the optimization model includes: The branch and bound method is used to solve the optimization model. Inputting the occupancy ratio set, task priority, data item attributes, and constraints, the optimal combination for retaining or excluding each data item is selected through iterative optimization. Result verification is performed, specifically checking whether the results fully comply with all constraints and are free of violations; assessing whether the overall integrity loss of the entire task is controlled within a preset acceptable range; if the results do not meet the standards, the leniency of the constraints is adjusted, and the solution process is restarted until the results meet the requirements; finally, a set of tasks that can be excluded is generated.

8. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The process of generating task adjustment instructions based on the set of excluding tasks and executing exclusion operations on the excluding tasks includes: Each task adjustment instruction corresponds to a set of data items to be excluded for a single task, including task ID, data item number and name to be excluded, data item storage path, exclusion operation timing requirements, resource release instructions, and exception handling plan. Adjustment instructions are generated for each task in order of task order. Instructions are reviewed and issued. After receiving the instruction, the execution end automatically reports the reception status and verifies the consistency between the local task data and the instruction requirements. When there are missing data items or mismatched paths, an exception warning is triggered and reported to the control end, and the operation is suspended for investigation. After verification, an execution confirmation receipt is generated, and the exclusion process is started. According to the instruction sequence requirements, the exclusion operation is executed in batches; after each set of data items is excluded, the task data index is updated, the index information of the excluded item is deleted, and a removal mark is added; the shared resource occupancy ledger is updated synchronously, the resource occupancy amount corresponding to the excluded item is deducted, and the resource release status is marked to ensure that the ledger is consistent with the actual resource occupancy; when the exclusion operation of multiple tasks involves the same shared resource, the exclusion operation of the low-priority task is executed only after the high-priority task's exclusion operation is completed and the resource is released.

9. The method for managing energy consumption equipment detection data based on big data according to claim 1, characterized in that, The resource scheduling scheme, based on task priority, remaining data collection needs, and the current occupancy status of shared resources, includes: High-priority tasks should be allocated high-quality resources first, while low-priority tasks should be avoided. Tasks with the same priority should be sorted according to their urgency. Resource allocation should match the task's collection needs to avoid resource overload or insufficient capacity, and a window should be reserved for core tasks to occupy resources continuously. Core resources should only be allocated to one task at a time. Prioritize the resource needs of high-priority tasks, allocate stable resources and lock the occupied time periods; allocate remaining resources to low-priority tasks and determine the upper limit of the occupied time; when resource competition occurs, prioritize the core collection needs, adjust the collection sequence of low-priority tasks, or split their non-core collection items to idle time periods; form a correspondence table of tasks, resources and occupied time periods, mark resource usage constraints and abnormal avoidance plans, and formulate resource scheduling schemes.

10. A big data-based energy consumption equipment testing data management system, using the big data-based energy consumption equipment testing data management method according to any one of claims 1-9, characterized in that, include: The abnormal correlation dataset generation module includes: a coupling interference determination unit that monitors energy consumption equipment test data and laboratory shared resource operation parameters in real time to determine whether operating condition coupling interference has occurred; and an abnormal correlation dataset generation unit that determines the abnormal correlation dataset when operating condition coupling interference occurs. Occupancy Ratio Set Generation Module: Includes: Occupancy Data Acquisition Unit: Acquires occupancy data of each task on the target shared resource in the abnormal correlation dataset during the interference occurrence period; Occupancy Ratio Calculation Unit: Analyzes the occupancy data using artificial intelligence algorithms, calculates the independent occupancy ratio of each task on the corresponding shared resource, and obtains the occupancy ratio set; The module for generating a set of tasks to be excluded includes: a task priority determination unit that retrieves preset task priority rules and determines the task priority of each task; an optimization model construction unit that sets a model objective and constructs an optimization model based on the occupancy ratio set and task priority; and a set of tasks to be excluded generation unit that generates a set of tasks to be excluded based on the optimization model. The resource scheduling scheme formulation module includes: an exclusion operation execution unit that generates task adjustment instructions based on the set of excludeable task items and executes exclusion operations on the excludeable task items; and a resource scheduling scheme formulation unit that formulates resource scheduling schemes based on task priorities, remaining data collection requirements, and the current occupancy status of shared resources.

Images