Integrated deployment scheme of non-intrusive load monitoring system based on edge computing technology

Abstract

The invention relates to the technical field of load monitoring, in particular to an integrated deployment scheme of a non-intrusive load monitoring system based on an edge computing technology, which comprises the following steps of: based on load parameters of edge equipment, acquiring current, voltage and power information, extracting load characteristic values, matching current and voltage with time points, and detecting power change amplitude, so as to obtain an integrated deployment scheme of the non-intrusive load monitoring system. And judging whether a threshold value is exceeded or not, marking an abnormal state and generating load abnormal marking data. According to the invention, load distribution judgment is carried out by adopting the long-short-term memory network, intelligent transfer of high-frequency loads is realized by dynamically evaluating the load states of adjacent equipment, the risk of equipment overload can be effectively reduced, generation of local overload of the system is avoided, and the load transfer efficiency is improved based on the load transfer list and the real-time load state. According to the method, the low-load equipment is dynamically selected to transfer the high-frequency load, the load state between the equipment is balanced through global load redistribution, the pressure of the high-load equipment is reduced, and the operation efficiency and safety are improved.

Description

Technical Field

[0001] The present invention relates to the field of load monitoring technology, and in particular to an integrated deployment solution of a non-intrusive load monitoring system based on edge computing technology. Background Art

[0002] The field of load monitoring technology aims to evaluate the load status of equipment or systems, optimize the allocation of power resources and improve the safety of electricity use. By analyzing the external current or voltage signals of the equipment, it can avoid interference with the operation of the equipment and improve the convenience and applicability of the monitoring system.

[0003] The main purpose of the integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology is to complete the data collection, analysis and processing related to load monitoring at the device end close to the data source, reduce the delay of data transmission to the remote cloud, and realize the rapid deployment and response of the monitoring system through the distributed architecture of edge computing, thereby ensuring the flexibility and reliability of the system.

[0004] Traditional methods rely on fixed thresholds and rules for load regulation, lacking the ability to effectively predict and dynamically adjust future load changes, resulting in a higher risk of equipment overload. During high-frequency load transfers, a centralized architecture is used for decision-making, resulting in large delays and difficulty in responding to load fluctuations in real time, causing local equipment overload and affecting overall stability. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and propose an integrated deployment solution of a non-intrusive load monitoring system based on edge computing technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: an integrated deployment solution of a non-intrusive load monitoring system based on edge computing technology, comprising the following steps:

[0007] Step 1: Based on the load parameters of the edge device, obtain the current, voltage and power information, extract the load characteristic value, match the current, voltage and time point, detect the power change amplitude, determine whether the power fluctuation exceeds the threshold, mark it as an abnormal state, and generate load abnormal state marking data;

[0008] Step 2: Based on the load abnormal state marking data, obtain the current and voltage parameters of each device, calculate the change amplitude of the current and voltage, compare the device carrying capacity with the current load, determine whether the load upper limit is reached, mark the overloaded device, and generate the load carrying capacity analysis result;

[0009] Step 3: Based on the load carrying capacity analysis results, a long short-term memory network is used to make load distribution judgments, obtain the load status of adjacent equipment, select the equipment closest to full load, transfer high-frequency fluctuating loads, transfer high-frequency loads to lower-load equipment, and generate a load transfer list;

[0010] Step 4: Based on the load transfer list, the load status of all devices is obtained, the device with the lowest load is selected to receive the transferred load, and dynamic adjustment is performed. By judging the difference between the current load and the data before the transfer, the transfer effect is confirmed, and the load scheduling and transfer results are generated;

[0011] Step 5: Based on the load scheduling and transfer results, obtain the real-time load status of all equipment, perform load fluctuation analysis, select equipment with smaller load fluctuation to add additional load, reduce the load of overloaded equipment, determine the stability of the load after adjustment, and generate equipment load adjustment plan;

[0012] Step 6: Based on the equipment load adjustment plan, obtain the global load distribution situation, perform load balancing operations between equipment, monitor load changes, determine whether load reduction is effective, perform global load redistribution, ensure load balance between equipment, and generate a global load distribution result;

[0013] Step 7: Based on the global load distribution result, obtain the load status of all devices, check the difference between the actual load and the expected distributed load, periodically adjust the device load, perform load consistency check, ensure the balanced state of the device load, and generate load consistency verification data.

[0014] As a further solution of the present invention, the load carrying capacity analysis results include the current load of the equipment, the maximum carrying capacity of the equipment and the overload status mark, the load scheduling and transfer results include the load comparison before and after the transfer, the transfer success mark and the load balancing status, the equipment load adjustment plan includes the additional load distribution amount, the load reduction amount and the load balance status after adjustment, the global load distribution results include the load reduction equipment list, the load distribution ratio and the load balancing mark, and the load consistency verification data includes the load difference value, the load adjustment frequency and the load consistency status mark.

[0015] As a further solution of the present invention, the specific steps of generating the load abnormal state marking data are:

[0016] Based on the load parameters of edge devices, current, voltage and power data are collected. The collected data is sorted in chronological order by real-time reading of each device, and the collected data is sorted by time tags to generate load parameter sorting data.

[0017] Based on the load parameters, the data is sorted, the change amplitude of the current and voltage is read for each device, and the time points are compared to calculate the difference of the current and voltage in each time period, match the power change, and generate the power change amplitude data;

[0018] Based on the power variation amplitude data, the power fluctuation is calculated for each time period, and by comparing the power amplitude of each time period with a preset threshold, the excess part is determined and marked to generate load abnormality status marking data.

[0019] As a further solution of the present invention, the specific steps of generating the load bearing capacity analysis result are:

[0020] Based on the load abnormality status mark data, read the current and voltage parameters of each device, obtain the real-time load status one by one, summarize the load data of the device, and generate the real-time load parameter data of the device according to the device number;

[0021] Based on the real-time load parameter data of the equipment, the current and voltage change amplitudes of each equipment are calculated, the change conditions of each equipment in different time periods are counted, the change difference is calculated, and the equipment load change amplitude data is generated;

[0022] Based on the equipment load change amplitude data, the load change of each equipment is compared with the upper limit of the equipment's carrying capacity, and it is determined one by one whether it exceeds the carrying capacity, and the overloaded equipment is marked to generate a load carrying capacity analysis result.

[0023] As a further solution of the present invention, the specific steps of generating the load transfer list are:

[0024] Based on the load carrying capacity analysis results, a long short-term memory network is used to obtain load data for each device one by one, and the difference between the current load of each device and its upper limit is calculated, and the devices are sorted according to the difference. The devices with loads close to full load are selected to generate a list of adjacent device load states.

[0025] Based on the adjacent equipment load status list, high-frequency fluctuation load data is obtained for each equipment, the current high-frequency load is calculated, and the load status of adjacent equipment is compared, and the equipment with lower load is selected as the load transfer target, and the high-frequency load is matched with the low-load equipment to generate a high-frequency load transfer plan;

[0026] Based on the high-frequency load transfer plan, the high-frequency load transfer operation is performed step by step. By monitoring the load changes of the equipment during the load transfer process, the load balance state of the equipment after the load transfer is recalculated, the results are recorded in a list, and a load transfer list is generated.

[0027] As a further solution of the present invention, the long short-term memory network is according to the formula:

[0028] D i =C i -L i +α·(T i -t i )+β·(E i -e i )

[0029] Where: D i is the difference between the current load of device i and its upper limit, C i is the upper limit of the load of device i, L i is the current load of equipment i, T i is the maximum working time of device i, t i is the current working time of device i, E i is the maximum energy efficiency value of equipment i, e i is the current energy efficiency value of device i, and α and β are weight coefficients.

[0030] As a further solution of the present invention, the specific steps of generating the load scheduling and transfer results are:

[0031] Based on the load transfer list, the load status of all devices is obtained one by one, and the current load of each device is compared with its carrying capacity, and the devices are sorted according to the load size, and the real-time load information of the devices is recorded to generate a device load status record;

[0032] Based on the equipment load status record, read the load value of each equipment one by one, select the equipment with the lowest current load as the load receiver, gradually perform load receiving and adjustment operations, and generate a dynamic load adjustment plan;

[0033] Based on the dynamic load adjustment plan, dynamic load transfer is performed, the load difference of the equipment before and after the transfer is calculated, the load distribution of each equipment is adjusted, and the transfer effect is confirmed. The transfer result is recorded to generate load scheduling and transfer results.

[0034] As a further solution of the present invention, the specific steps of generating the equipment load adjustment solution are:

[0035] Based on the load scheduling and transfer results, the real-time load status of each device is obtained, the current load value of each device is checked one by one, and the load fluctuation amplitude of each device is calculated to select the device with smaller fluctuation, determine the additional load it can bear, and generate the device load fluctuation analysis data;

[0036] Based on the equipment load fluctuation analysis data, check the load status of the equipment, calculate the overload value of the overloaded equipment one by one, gradually reduce the load of each overloaded equipment according to the calculation results, record the specific value of the reduced load during the reduction process, and generate an overload load reduction list;

[0037] Based on the overload reduction list, check the load change of each device after the reduction, calculate the stability after the load reduction one by one, analyze the amplitude of load fluctuation, evaluate the load balance after adjustment, and generate the equipment load adjustment plan.

[0038] As a further solution of the present invention, the specific steps of generating the global load distribution result are:

[0039] Based on the equipment load adjustment scheme, the current load distribution status of each equipment is obtained, and by comparing the load differences between the equipment, the balance state between the maximum and minimum load values ​​is calculated to generate global load distribution inspection data;

[0040] Based on the global load distribution inspection data, the load reduction status of the equipment is checked step by step, the real-time load data of each equipment is calculated, and by comparing each equipment one by one, it is determined whether the load reduction has achieved the expected effect, and the load status of the equipment is recorded to generate the load reduction effectiveness result;

[0041] Based on the load reduction effectiveness result, global load redistribution among devices is gradually performed, and task allocation is adjusted according to the order of device load. By redistributing tasks of high-load devices, load balancing is ensured, and a global load distribution result is generated.

[0042] As a further solution of the present invention, the specific steps of generating the load consistency verification data are:

[0043] Based on the global load distribution result, current load data is obtained for each device, and load deviation values ​​are calculated by comparing the actual load of each device with the expected load, and device load state difference data is generated;

[0044] Based on the equipment load status difference data, the load fluctuation status of the equipment is checked step by step, the load of the equipment is redistributed periodically, and the load redistribution balance is ensured by adjusting the task amount of each equipment, and a periodic load adjustment plan is generated;

[0045] Based on the periodic load adjustment plan, the load consistency of each device is checked, and the load balancing state of the device is confirmed by comparing the actual load distribution of the device with the expected load distribution, and load consistency verification data is generated.

[0046] Compared with the prior art, the advantages and positive effects of the present invention are:

[0047] In the present invention, a long short-term memory network is used to make load distribution judgments. By dynamically evaluating the load status of adjacent equipment, intelligent transfer of high-frequency loads is achieved, which can effectively reduce the risk of equipment overload, avoid the occurrence of local overload of the system, and improve the accuracy of load scheduling and the overall stability of the system. Based on the load transfer list and real-time load status, low-load equipment is dynamically selected for high-frequency load transfer. Through global load redistribution, the load status between equipment is balanced, the pressure on high-load equipment is reduced, and the operation efficiency and safety are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a schematic diagram of the main steps of the present invention;

[0049] Figure 2 It is a schematic diagram of the process of load decomposition of the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0051] Embodiment 1

[0052] See also Figure 1 The present invention provides a technical solution: an integrated deployment solution of a non-intrusive load monitoring system based on edge computing technology, comprising the following steps:

[0053] Step 1: Based on the load parameters of the edge device, obtain the current, voltage and power information, extract the load characteristic value, match the current, voltage and time point, detect the power change amplitude, determine whether the power fluctuation exceeds the threshold, mark it as an abnormal state, and generate load abnormal state marking data;

[0054] Step 2: Based on the load abnormal status mark data, obtain the current and voltage parameters of each device, calculate the change amplitude of current and voltage, compare the device carrying capacity with the current load, determine whether the load limit is reached, mark the overloaded device, and generate the load carrying capacity analysis result;

[0055] Step 3: Based on the load carrying capacity analysis results, a long short-term memory network is used to make load distribution judgments, obtain the load status of adjacent equipment, select the equipment closest to full load, transfer high-frequency fluctuating loads, transfer high-frequency loads to lower-load equipment, and generate a load transfer list;

[0056] Step 4: Based on the load transfer list, obtain the load status of all devices, select the device with the lowest load to receive the transferred load, make dynamic adjustments, determine the difference between the current load and the data before the transfer, confirm the transfer effect, and generate the load scheduling and transfer results;

[0057] Step 5: Based on the load scheduling and transfer results, obtain the real-time load status of all equipment, perform load fluctuation analysis, select equipment with smaller load fluctuation to add additional load, reduce the load of overloaded equipment, determine the stability of the load after adjustment, and generate equipment load adjustment plan;

[0058] Step 6: Based on the equipment load adjustment plan, obtain the global load distribution situation, perform load balancing operations between equipment, monitor load changes, determine whether load reduction is effective, perform global load redistribution, ensure load balance between equipment, and generate global load distribution results;

[0059] Step 7: Based on the global load distribution results, obtain the load status of all devices, check the difference between the actual load and the expected distributed load, periodically adjust the device load, perform load consistency checks, ensure the balanced state of the equipment load, and generate load consistency verification data.

[0060] The load carrying capacity analysis results include the current load of the equipment, the maximum carrying capacity of the equipment and the overload status mark. The load scheduling and transfer results include the load comparison before and after the transfer, the transfer success mark and the load balancing status. The equipment load adjustment plan includes the additional load distribution amount, the load reduction amount and the load balance status after adjustment. The global load distribution results include the load reduction equipment list, the load distribution ratio and the load balancing mark. The load consistency verification data includes the load difference value, the load adjustment frequency and the load consistency status mark.

[0061] The specific steps for generating load abnormal status marking data are as follows:

[0062] Based on the load parameters of edge devices, current, voltage and power data are collected. The collected data is sorted in chronological order by real-time reading of each device, and the collected data is sorted by time tags to generate load parameter sorting data.

[0063] Organize data based on load parameters, read the change amplitude of current and voltage for each device, compare them by time points, calculate the difference of current and voltage in each time period, match the power change, and generate power change amplitude data;

[0064] Based on the power change amplitude data, the power fluctuation is calculated for each time period. By comparing the power amplitude of each time period with the preset threshold, the excess part is determined and marked to generate load abnormal status marking data;

[0065] Based on the load parameters of edge devices, the Modbus protocol is used to collect current, voltage and power data. The function code 03H is sent, the data length is set to 2 bytes, the reading start address is the register address of the specified device, and the data of each device is read in real time. According to the register value in the read data packet, the current, voltage and power values ​​are parsed into floating point numbers, the collected data are sorted in chronological order, and the parsed data are labeled according to the timestamp. The timestamp sorting algorithm is used, and the UNIX timestamp is used as the sorting key to sort the time tags and generate load parameter sorting data;

[0066] The data is sorted based on the load parameters, and the current and voltage data read from each device are processed using a differential algorithm. The differences of the current and voltage values ​​in each time period are calculated item by item, and the data are compared by time point. The power change amplitude is matched using a binary search algorithm, and the result is matched with the data item at the corresponding time point to generate the power change amplitude data.

[0067] Based on the power change amplitude data, an amplitude comparison algorithm is used to calculate the power fluctuation in each time period. The power change value is compared with the preset threshold in each time period to determine whether it exceeds the threshold. The excess part is marked through the comparison result to generate load abnormal status marking data.

[0068] The specific steps to generate the load carrying capacity analysis results are:

[0069] Based on the abnormal load status mark data, read the current and voltage parameters of each device, obtain the real-time load status one by one, summarize the load data of the device, and generate the real-time load parameter data of the device according to the device number;

[0070] Based on the real-time load parameter data of the equipment, calculate the current and voltage change range of each equipment, collect statistics on the changes of each equipment in different time periods, calculate the change difference, and generate the equipment load change range data;

[0071] Based on the equipment load change amplitude data, the load change of each equipment is compared with the upper limit of the equipment's carrying capacity, and whether the carrying capacity is exceeded is determined one by one, and the overloaded equipment is marked to generate the load carrying capacity analysis results;

[0072] Based on the abnormal load status mark data, the Modbus TCP protocol is used to read the current and voltage parameters of each device. By sending the function code 03H, the register address is specified to read the current and voltage values. The read data is classified according to the device number, and the number of each device is mapped and stored with the read data through the hash mapping algorithm. The real-time load status is obtained one by one, the load data is summarized by the device number, and the real-time load parameter data of the device is generated;

[0073] Based on the real-time load parameter data of the equipment, the sliding window algorithm is used to calculate the current and voltage change amplitude of each device. The window size of the sliding window is set to 5 time units. Statistics are collected according to the data changes in each time period, and the change difference of current and voltage in adjacent time periods is gradually calculated to calculate the change amplitude. The result is associated with the equipment number to generate the equipment load change amplitude data;

[0074] Based on the equipment load change amplitude data, an item-by-item comparison algorithm is used to compare the load change of each device with its upper limit of carrying capacity. The equipment's carrying capacity parameter list is searched item by item to determine whether the load change of each device exceeds its upper limit of carrying capacity. The overloaded equipment is marked to generate the load carrying capacity analysis results.

[0075] The specific steps to generate a load transfer list are:

[0076] Based on the load carrying capacity analysis results, a long short-term memory network is used to obtain load data for each device. The difference between the current load of each device and its upper limit is calculated, and the devices are sorted according to the difference. The devices with loads close to full load are selected to generate a list of adjacent device load states.

[0077] Based on the load status list of adjacent equipment, obtain high-frequency fluctuation load data for each equipment, calculate the current high-frequency load, and compare the load status of adjacent equipment. Select the equipment with lower load as the load transfer target, match the high-frequency load with the low-load equipment, and generate a high-frequency load transfer plan;

[0078] Based on the high-frequency load transfer plan, the high-frequency load transfer operation is gradually carried out. By monitoring the load changes of the equipment during the load transfer process, the load balance state of the equipment after the load transfer is recalculated, the results are recorded in the list, and the load transfer list is generated;

[0079] Based on the results of load carrying capacity analysis, a long short-term memory network is used to obtain load data for each device. The input layer is set to the historical load data of the device, the time step is set to 10, the number of hidden layer neurons is 128, and the output layer generates the current load forecast value of each device. By calculating the difference between the current load of each device and its upper limit, the difference calculation uses a subtraction operation, records the difference value of each device, and uses a quick sorting algorithm to sort the difference values. The difference values ​​are sorted according to the size of the difference, and the devices with loads close to full load are selected to generate a list of adjacent device load states.

[0080] Based on the load status list of adjacent equipment, the wavelet transform algorithm is used to obtain high-frequency fluctuation load data for each equipment. The wavelet basis function uses db4. By decomposing the load signal of each equipment, the high-frequency load component is extracted, and the high-frequency load amount is calculated. The high-frequency load amount is compared with the load status of adjacent equipment. The nearest neighbor search algorithm is used to select the equipment with lower load as the load transfer target. The high-frequency load is matched with the low-load equipment through the load matching algorithm to generate a high-frequency load transfer plan.

[0081] Based on the high-frequency load transfer plan, the high-frequency load transfer operation is carried out step by step. The load distribution algorithm is used to gradually distribute the high-frequency load to the low-load equipment. During the transfer process, the load changes of each device are monitored in real time through the load monitoring module, and the load balance status of the equipment after the transfer is recalculated. The balance judgment formula is used to compare the difference before and after the load, and the load changes of each device are recorded. The load transfer list is generated and recorded.

[0082] Long short-term memory network, according to the formula:

[0083] D i =C i -L i +α·(T i -t i )+β·(E i -e i )

[0084] Where: D i is the difference between the current load of device i and its upper limit, C i is the upper limit of the load of device i, L i is the current load of equipment i, T i is the maximum working time of device i, t i is the current working time of device i, E i is the maximum energy efficiency value of equipment i, e i is the current energy efficiency value of equipment i, α and β are weight coefficients;

[0085] Execution process: First, obtain the current load L of device i i , and extract the load limit C of device i from the system parameters i , further introduce the maximum working time T of device i i and current working time t i , through T i -t i Calculate the difference in working time of the equipment, and then adjust the impact of the difference through the weight coefficient α to obtain the maximum energy efficiency E of the equipment i and current energy efficiency i , through E i -ei Calculate the energy efficiency difference, adjust the impact of the difference by the weight coefficient β, and calculate the comprehensive load difference D of each device by combining the load difference, working time difference and energy efficiency difference of the equipment. i , and then sort the equipment according to the size of the difference, and select the equipment with a load close to full load. The weight coefficients α and β can be obtained by regression analysis of the load data, working hours and energy efficiency fluctuations of the historical equipment. The statistical data of the long-term operation of the equipment is used to determine the reasonable weight coefficient value by fitting methods such as the least squares method to ensure that the impact of each difference factor on the load calculation is reasonable.

[0086] The specific steps to generate load scheduling and transfer results are:

[0087] Based on the load transfer list, the load status of all devices is obtained one by one. By comparing the current load of each device with its carrying capacity, the devices are sorted according to the load size, and the real-time load information of the devices is recorded to generate a record of the device load status;

[0088] Based on the equipment load status record, read the load value of each device one by one, select the device with the lowest current load as the load receiver, gradually perform load receiving and adjustment operations, and generate a dynamic load adjustment plan;

[0089] Based on the dynamic load adjustment plan, the dynamic transfer of load is performed. By calculating the load difference of the equipment before and after the transfer, the load distribution of each equipment is adjusted, and the transfer effect is confirmed. The transfer results are recorded to generate load scheduling and transfer results;

[0090] Based on the load transfer list, a multi-threaded concurrent algorithm is used to obtain the load status of all devices one by one, and the current load data of each device is read in real time. By comparing the current load of the device with its carrying capacity, the load value of each device is recorded in the hash table structure, and the load values ​​of all devices are sorted using a quick sorting algorithm. The sorting results are stored in order according to the load size, and the device load status record is generated;

[0091] Based on the equipment load status record, read the load value of each device one by one, use the greedy algorithm to select the device with the lowest current load as the load receiver, and then call the data in the load status record to select the device with the lowest load in the sorting as the transfer target, and gradually perform load receiving and adjustment operations. The load adjustment adopts the proportional allocation algorithm, calculates the received load according to the ratio of the current load to the carrying capacity, and generates a dynamic load adjustment plan;

[0092] Based on the dynamic load adjustment plan, the dynamic load transfer operation is performed. The dynamic load transfer algorithm is used to transfer the high load to the low load device one by one according to the transfer sequence in the adjustment plan. By reading the load data of each device before and after the transfer, the differential algorithm is used to calculate the load difference, and the load status is redistributed according to the adjusted load data. The load changes of each device are confirmed, the load values ​​during the transfer process are recorded, and the load scheduling and transfer results are generated and recorded.

[0093] See also Figure 2 It should be noted that steps 1, 2, 3 and 4 complete the non-intrusive load decomposition, and the process is as follows:

[0094] By analyzing the power consumption signal of the equipment, the power consumption cycle and power consumption trend are first extracted as the fingerprint features of the electrical equipment. The input signal passes through the feature extraction module to extract the periodic power consumption pattern and trend power consumption change, and then merge them to fully describe the load behavior of the equipment. The power consumption cycle feature reflects the switching and operation rules of the equipment, and the power consumption trend describes the power consumption change of the equipment within a certain period of time.

[0095] The extracted feature information is processed by a feedforward neural network. The neural network outputs multiple decomposition signals according to the power consumption behavior of different devices, and is optimized by a convolutional neural network. The device load signal is processed by weight superposition to generate a decomposition result to ensure the accuracy of load decomposition. After multiple training and optimization, the decomposition model is stored in the edge layer for the load decomposition module to call and deploy on demand.

[0096] The central layer is responsible for data aggregation and abnormal warning functions, monitoring the working status of the equipment, sending warning information in a timely manner when abnormal or illegal equipment is detected, and performing energy management on the user's overall electricity consumption and providing energy-saving strategies.

[0097] The specific steps to generate a device load adjustment plan are:

[0098] Based on the load scheduling and transfer results, the real-time load status of each device is obtained, and the current load value of each device is checked. By calculating the load fluctuation amplitude of each device, the device with smaller fluctuation is selected to determine the additional load it can bear, and the device load fluctuation analysis data is generated;

[0099] Based on the equipment load fluctuation analysis data, check the load status of the equipment, calculate the overload value of the overloaded equipment one by one, and gradually reduce the load of each overloaded equipment according to the calculation results. During the reduction process, record the specific value of the reduced load and generate an overload reduction list;

[0100] Based on the overload reduction list, check the load change of each device after the reduction, calculate the stability after the load reduction one by one, analyze the amplitude of load fluctuation, evaluate the load balance after adjustment, and generate the equipment load adjustment plan;

[0101] Based on the results of load scheduling and transfer, the real-time monitoring algorithm is used to obtain the real-time load status of each device. The current load value of each device is checked by calling the device load monitoring module. The load fluctuation calculation algorithm is used to analyze the load value in each time period and calculate the load fluctuation amplitude. By comparing the fluctuation amplitudes of each device, the device with smaller fluctuation is selected to determine the additional load it can bear, and generate the equipment load fluctuation analysis data;

[0102] Based on the equipment load fluctuation analysis data, check the load status of the equipment one by one, use the overload calculation algorithm to calculate the overload value of the overloaded equipment, call the load reduction function for each overloaded equipment, and gradually reduce the load according to the overload value. The reduction operation is executed through the load control interface of the equipment, and the load distribution algorithm is used to determine the specific value of the load reduction. The reduced load is recorded in the log table to generate an overload load reduction list;

[0103] Based on the overload reduction list, the load change tracking algorithm is used to check the load changes of each device after the reduction, and the load stability after the reduction is calculated one by one. The load balancing algorithm is used to analyze the amplitude of the load fluctuation of each device. The load balance after adjustment is evaluated by calling the load balancing function multiple times, and the load data of all devices is recorded to generate the equipment load adjustment plan.

[0104] The specific steps to generate the global load distribution results are:

[0105] Based on the equipment load adjustment plan, the current load distribution status of each equipment is obtained, and the balance state between the maximum and minimum load values ​​is calculated by comparing the load differences between the equipment to generate global load distribution inspection data;

[0106] Based on the global load distribution inspection data, the load reduction status of the equipment is checked step by step, the real-time load data of each equipment is calculated, and by comparing each equipment, it is determined whether the load reduction has achieved the expected effect, and the load status of the equipment is recorded to generate the load reduction effectiveness result;

[0107] Based on the load reduction effectiveness results, global load redistribution among devices is gradually carried out. Task allocation is adjusted according to the order of high and low loads of devices. By redistributing tasks of high-load devices, load balancing is ensured and global load allocation results are generated.

[0108] Based on the equipment load adjustment plan, the load distribution monitoring algorithm is used to obtain the current load distribution status of each device one by one, and the equipment load monitoring module is called to read the real-time load data of each device one by one, compare the load differences between the devices, and use the maximum-minimum value calculation algorithm to calculate the difference between the maximum and minimum values ​​of the load between the devices, determine the load balance status between the devices, and generate global load distribution inspection data;

[0109] Based on the global load distribution inspection data, the load reduction status of each device is checked step by step using the real-time data comparison algorithm. The load reduction log record is called to read the current load of each device. By comparing the current load data with the previous reduction data of each device, the reduction effect judgment algorithm is used to determine whether the load reduction has achieved the expected effect, and the real-time load status of each device is recorded to generate the load reduction effectiveness result.

[0110] Based on the load reduction effectiveness results, a global load redistribution algorithm is used to gradually redistribute the load between devices. The device load status module is called to adjust the task allocation in the order of device load. The task redistribution algorithm is used to redistribute the tasks of high-load devices and evenly distribute the load to low-load devices. The results of the redistributed tasks are recorded to generate the global load allocation results.

[0111] The specific steps to generate load consistency verification data are:

[0112] Based on the global load distribution results, the current load data of each device is obtained, and the load deviation value is calculated by comparing the actual load of each device with the expected load, and the device load status difference data is generated;

[0113] Based on the equipment load status difference data, the load fluctuation status of the equipment is checked step by step, and the load of the equipment is redistributed periodically. By adjusting the task volume of each equipment, the load redistribution balance is ensured, and a periodic load adjustment plan is generated;

[0114] Based on the periodic load adjustment plan, check the load consistency of each device, confirm the load balancing status of the device by comparing the actual load distribution of the device with the expected load distribution, and generate load consistency verification data;

[0115] Based on the global load distribution results, a real-time data acquisition algorithm is used to obtain the current load data of each device one by one. The actual load value of the device is read by calling the device load monitoring interface. The expected load comparison algorithm is used to compare the actual load of each device with the expected load. The load deviation value is calculated using differential operation, and the deviation information of each device is recorded to generate device load status difference data.

[0116] Based on the equipment load status difference data, the load fluctuation detection algorithm is used to gradually check the load fluctuation of the equipment. The load fluctuation value in each cycle is obtained by calling the periodic load monitoring module. The load fluctuation status of each device is checked according to the set cycle. The task volume adjustment algorithm is used to dynamically adjust the task volume in each cycle. The task volume is reallocated according to the current load status of the equipment to ensure load distribution balance and generate a periodic load adjustment plan.

[0117] Based on the periodic load adjustment plan, a consistency detection algorithm is used to check the load consistency of each device. The load distribution module is called to compare the actual load distribution of the device with the expected load distribution. The cosine similarity calculation method is used to evaluate the load distribution consistency of each device, confirm the load distribution balance state, record the load consistency of the equipment, and generate load consistency verification data.

[0118] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An integrated deployment solution for a non-intrusive load monitoring system based on edge computing technology, characterized in that: The following steps are involved: Step 1: Based on the load parameters of the edge device, obtain the current, voltage and power information, extract the load characteristic value, match the current, voltage and time point, detect the power change amplitude, determine whether the power fluctuation exceeds the threshold, mark it as an abnormal state, and generate load abnormal state marking data; Step 2: Based on the load abnormal state marking data, obtain the current and voltage parameters of each device, calculate the change amplitude of the current and voltage, compare the device carrying capacity with the current load, determine whether the load upper limit is reached, mark the overloaded device, and generate the load carrying capacity analysis result; Step 3: Based on the load carrying capacity analysis results, a long short-term memory network is used to make load distribution judgments, obtain the load status of adjacent equipment, select the equipment closest to full load, transfer high-frequency fluctuating loads, transfer high-frequency loads to lower-load equipment, and generate a load transfer list; Step 4: Based on the load transfer list, the load status of all devices is obtained, the device with the lowest load is selected to receive the transferred load, and dynamic adjustment is performed. By judging the difference between the current load and the data before the transfer, the transfer effect is confirmed, and the load scheduling and transfer results are generated; Step 5: Based on the load scheduling and transfer results, obtain the real-time load status of all equipment, perform load fluctuation analysis, select equipment with smaller load fluctuation to add additional load, reduce the load of overloaded equipment, determine the stability of the load after adjustment, and generate equipment load adjustment plan; Step 6: Based on the equipment load adjustment plan, obtain the global load distribution situation, perform load balancing operations between equipment, monitor load changes, determine whether load reduction is effective, perform global load redistribution, ensure load balance between equipment, and generate a global load distribution result; Step 7: Based on the global load distribution result, obtain the load status of all devices, check the difference between the actual load and the expected distributed load, periodically adjust the device load, perform load consistency check, ensure the balanced state of the device load, and generate load consistency verification data.

2. According to claim 1, the integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology is characterized in that: The load carrying capacity analysis results include the current load of the equipment, the maximum carrying capacity of the equipment and the overload status mark; the load scheduling and transfer results include the load comparison before and after the transfer, the transfer success mark and the load balancing status; the equipment load adjustment plan includes the additional load distribution amount, the load reduction amount and the load balance status after adjustment; the global load distribution results include the load reduction equipment list, the load distribution ratio and the load balancing mark; the load consistency verification data includes the load difference value, the load adjustment frequency and the load consistency status mark.

3. According to the integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology of claim 1, it is characterized in that: The specific steps of generating the load abnormal state marking data are: Based on the load parameters of edge devices, current, voltage and power data are collected. The collected data is sorted in chronological order by real-time reading of each device, and the collected data is sorted by time tags to generate load parameter sorting data. Based on the load parameters, the data is sorted, the change amplitude of the current and voltage is read for each device, and the time points are compared to calculate the difference of the current and voltage in each time period, match the power change, and generate the power change amplitude data; Based on the power variation amplitude data, the power fluctuation is calculated for each time period, and by comparing the power amplitude of each time period with a preset threshold, the excess part is determined and marked to generate load abnormality status marking data.

4. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the load bearing capacity analysis result are: Based on the load abnormality status mark data, read the current and voltage parameters of each device, obtain the real-time load status one by one, summarize the load data of the device, and generate the real-time load parameter data of the device according to the device number; Based on the real-time load parameter data of the equipment, the current and voltage change amplitudes of each equipment are calculated, the change conditions of each equipment in different time periods are counted, the change difference is calculated, and the equipment load change amplitude data is generated; Based on the equipment load change amplitude data, the load change of each equipment is compared with the upper limit of the equipment's carrying capacity, and it is determined one by one whether it exceeds the carrying capacity, and the overloaded equipment is marked to generate a load carrying capacity analysis result.

5. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the load transfer list are: Based on the load carrying capacity analysis results, a long short-term memory network is used to obtain load data for each device one by one, and the difference between the current load of each device and its upper limit is calculated, and the devices are sorted according to the difference. The devices with loads close to full load are selected to generate a list of adjacent device load states. Based on the adjacent equipment load status list, high-frequency fluctuation load data is obtained for each equipment, the current high-frequency load is calculated, and the load status of adjacent equipment is compared, and the equipment with lower load is selected as the load transfer target, and the high-frequency load is matched with the low-load equipment to generate a high-frequency load transfer plan; Based on the high-frequency load transfer plan, the high-frequency load transfer operation is performed step by step. By monitoring the load changes of the equipment during the load transfer process, the load balance state of the equipment after the load transfer is recalculated, the results are recorded in a list, and a load transfer list is generated.

6. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The long short-term memory network is based on the formula: D i =C i -L i +α·(T i -t i )+β·(E i -e i ) Where: D i is the difference between the current load of device i and its upper limit, C i is the upper limit of the load of device i, L i is the current load of equipment i, T i is the maximum working time of device i, t i is the current working time of device i, E i is the maximum energy efficiency value of equipment i, e i is the current energy efficiency value of device i, and α and β are weight coefficients.

7. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the load scheduling and transfer results are: Based on the load transfer list, the load status of all devices is obtained one by one, and the current load of each device is compared with its carrying capacity, and the devices are sorted according to the load size, and the real-time load information of the devices is recorded to generate a device load status record; Based on the equipment load status record, read the load value of each equipment one by one, select the equipment with the lowest current load as the load receiver, gradually perform load receiving and adjustment operations, and generate a dynamic load adjustment plan; Based on the dynamic load adjustment plan, dynamic load transfer is performed, the load difference of the equipment before and after the transfer is calculated, the load distribution of each equipment is adjusted, and the transfer effect is confirmed. The transfer result is recorded to generate load scheduling and transfer results.

8. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the equipment load adjustment plan are: Based on the load scheduling and transfer results, the real-time load status of each device is obtained, the current load value of each device is checked one by one, and the load fluctuation amplitude of each device is calculated to select the device with smaller fluctuation, determine the additional load it can bear, and generate the device load fluctuation analysis data; Based on the equipment load fluctuation analysis data, check the load status of the equipment, calculate the overload value of the overloaded equipment one by one, gradually reduce the load of each overloaded equipment according to the calculation results, record the specific value of the reduced load during the reduction process, and generate an overload load reduction list; Based on the overload reduction list, check the load change of each device after the reduction, calculate the stability after the load reduction one by one, analyze the amplitude of load fluctuation, evaluate the load balance after adjustment, and generate the equipment load adjustment plan.

9. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the global load distribution result are: Based on the equipment load adjustment scheme, the current load distribution status of each equipment is obtained, and by comparing the load differences between the equipment, the balance state between the maximum and minimum load values ​​is calculated to generate global load distribution inspection data; Based on the global load distribution inspection data, the load reduction status of the equipment is checked step by step, the real-time load data of each equipment is calculated, and by comparing each equipment one by one, it is determined whether the load reduction has achieved the expected effect, and the load status of the equipment is recorded to generate the load reduction effectiveness result; Based on the load reduction effectiveness result, global load redistribution among devices is gradually performed, and task allocation is adjusted according to the order of device load. By redistributing tasks of high-load devices, load balancing is ensured, and a global load distribution result is generated.

10. The integrated deployment solution of the non-intrusive load monitoring system based on edge computing technology according to claim 1 is characterized in that: The specific steps of generating the load consistency verification data are: Based on the global load distribution result, current load data is obtained for each device, and load deviation values ​​are calculated by comparing the actual load of each device with the expected load, and device load state difference data is generated; Based on the equipment load status difference data, the load fluctuation status of the equipment is checked step by step, the load of the equipment is redistributed periodically, and the load redistribution balance is ensured by adjusting the task amount of each equipment, and a periodic load adjustment plan is generated; Based on the periodic load adjustment plan, the load consistency of each device is checked, and the load balancing state of the device is confirmed by comparing the actual load distribution of the device with the expected load distribution, and load consistency verification data is generated.

Images