A method and system for generating a dynamic inspection task based on energy-saving operation data

Abstract

The application discloses a kind of based on energy-saving operation data's inspection task dynamic generation method and system, one aspect, the energy-saving deviation evaluation model is utilized in the application, the deviation degree of the actual energy consumption of equipment and theoretical energy consumption is calculated, so as to be able to capture early implicit anomaly, such as energy consumption increase, efficiency decline, can actively warn before equipment occurs obvious fault;On the other hand, by introducing the correlation model between devices, the conduction influence of the energy-saving deviation of a single device to other devices in the system can be quantified, so as to preferentially schedule resources to inspect the key equipment with the greatest impact on system-level energy efficiency;Finally, an inspection benefit evaluation model is established to calculate the benefit index brought by each device after inspection, and further combined with the energy-saving deviation value, the device health risk index, the correlation influence value is comprehensively scored, so as to generate an inspection strategy, in this way, the overall energy utilization efficiency can be continuously improved while ensuring system reliability, and the optimal allocation of inspection resources is realized.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology based on energy management, and specifically relates to a method and system for dynamically generating inspection tasks based on energy-saving operation data. Background Technology

[0002] With the development of energy management technology, more and more buildings, parks and industrial settings are deploying energy-saving equipment and energy management systems, such as central air conditioning systems, cold and heat source systems, lighting control systems, and power equipment systems. These systems can monitor the operating status of equipment by collecting equipment operating parameters and energy consumption data, thereby improving energy utilization efficiency.

[0003] Currently, the main inspection methods for energy-saving equipment include the following: The first method is manual periodic inspection. This method arranges for inspectors to check the equipment according to a preset cycle, such as daily inspection, weekly inspection, or monthly inspection. Although this method can guarantee the basic maintenance needs of the equipment, the inspection plan is usually fixed and has low correlation with the actual operating status of the equipment. This can easily lead to repeated inspections when the equipment is operating normally, and problems such as failure to detect the decline in the energy-saving performance of the equipment in a timely manner.

[0004] The second approach is a passive inspection method based on fault alarms. This method monitors equipment operating parameters and generates inspection or maintenance tasks when the equipment exceeds a preset alarm threshold. However, energy-saving equipment often exhibits hidden abnormalities such as increased energy consumption and decreased operating efficiency before obvious faults occur. Traditional alarm mechanisms mainly focus on equipment fault conditions and cannot proactively identify early energy-saving losses.

[0005] The third approach is intelligent operation and maintenance based on equipment status evaluation. This approach typically establishes an evaluation model based on equipment operating parameters, fault records, and other information, and arranges maintenance tasks based on the evaluation results. However, most existing methods focus on equipment reliability or fault risk, lacking a comprehensive analysis of equipment energy-saving benefits and the correlation between equipment. They cannot determine whether the current inspection task can generate the greatest energy-saving value, thus leading to unreasonable allocation of inspection resources.

[0006] Therefore, based on the aforementioned shortcomings, how to proactively identify energy-saving anomalies by combining real-time equipment operation data and energy consumption trends, and dynamically generate more reasonable inspection tasks by comprehensively considering equipment health status, related impacts, and potential energy-saving benefits after inspection, has become an urgent problem to be solved in the field of intelligent operation and maintenance of energy-saving equipment. Summary of the Invention

[0007] The purpose of this invention is to provide a method and system for dynamically generating inspection tasks based on energy-saving operation data, in order to solve the problems of fixed inspection plans leading to repetitive inspections and the inability to detect equipment energy-saving performance degradation in a timely manner, the inability to detect energy-saving efficiency degradation in advance by relying on fault alarms, and the unreasonable allocation of inspection resources in the prior art.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a method for dynamically generating inspection tasks based on energy-saving operation data is provided, including: Acquire real-time operating data and historical operating records of each device in the system to be inspected; Based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model, the energy-saving deviation value of each device is calculated. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state. Based on the historical operating records of each device and using a pre-built device health risk assessment model, the device health risk index of each device is calculated. A device association model of the system to be inspected is constructed, and the association impact value of each device is calculated based on the device association model. The association impact value of any device is used to characterize the degree of influence of the energy-saving deviation of the source device on the device. The larger the association impact value, the greater the degree of influence. The source device is all devices in the system to be inspected that have an influence path with the device. Establish an inspection benefit evaluation model, and calculate the inspection benefit index obtained by each device after inspection based on the inspection benefit evaluation model. The comprehensive inspection score of each piece of equipment is calculated by using the energy-saving deviation value, equipment health risk index, correlation impact value and inspection benefit index of each piece of equipment. Based on the comprehensive inspection scores of each device, an inspection strategy is generated for the system to be inspected.

[0009] Based on the aforementioned disclosures, this invention constructs a multi-dimensional intelligent evaluation system to achieve dynamic generation and proactive optimization of energy-saving equipment inspection tasks. On one hand, unlike traditional fixed-cycle inspections or fault alarm triggering mechanisms, this invention utilizes an energy-saving deviation evaluation model to calculate in real-time the degree of deviation between actual and theoretical energy consumption. This allows for the keen detection of early, latent anomalies such as increased energy consumption and decreased efficiency, providing proactive warnings before significant equipment failures occur, thus effectively preventing continuous energy waste. On the other hand, by introducing an inter-equipment correlation model, it can quantify the transmission impact of energy-saving deviation of a single device on other devices within the system, prioritizing resource allocation. The invention involves inspecting key equipment that has the greatest impact on system-level energy efficiency. Finally, an inspection benefit evaluation model is established to calculate the benefit index of each piece of equipment after inspection. This is further combined with energy-saving deviation values, equipment health risk indices, and related impact values ​​for a comprehensive score. Based on this, an inspection strategy is generated. This approach ensures that inspection decisions not only focus on the status of individual equipment but also consider the coupling effect of equipment and energy-saving returns. It can continuously improve overall energy utilization efficiency while ensuring system reliability, achieving optimal allocation of inspection resources and significantly enhancing the intelligence level and operational economy of the energy management system. Therefore, this invention is highly suitable for large-scale application and promotion.

[0010] In one possible design, based on the real-time operating data of each device and a pre-built energy-saving deviation evaluation model, the energy-saving deviation value of each device is calculated, including: For any device, obtain the operating parameters of the device during normal operation as historical benchmark operating data; Based on historical benchmark operating data, a benchmark energy consumption dynamic prediction model for any of the devices is constructed. Based on the real-time operating data of any of the devices and the benchmark energy consumption dynamic prediction model, the theoretical energy consumption of any of the devices in the current operating state is calculated. Collect the actual energy consumption of any of the devices under the current operating state; Based on actual energy consumption, theoretical energy consumption, and the energy-saving deviation evaluation model, the energy-saving deviation value of any of the devices is calculated. Wherein, the energy-saving deviation value of any of the devices is: ; In the formula, The model value of the energy-saving deviation evaluation model represents the energy-saving deviation value of any given device. This represents the actual energy consumption of any of the aforementioned devices. This represents the theoretical energy consumption of any of the aforementioned devices. This indicates the runtime of any given device.

[0011] In one possible design, the historical operation record of any device includes: the continuous operating time of the device, as well as the number of historical failures, maintenance intervals, alarm frequencies, and fluctuation rates of specified operating parameters within a historical preset duration; Based on the historical operating records of each device and using a pre-built device health risk assessment model, the device health risk index for each device is calculated, including: The continuous operating time, the number of historical faults, the maintenance interval, the alarm frequency, and the fluctuation rate of the specified operating parameter are normalized to obtain the continuous operating time score, the number of historical faults score, the maintenance interval score, the alarm frequency score, and the fluctuation rate of the specified operating parameter score. Based on continuous operating time score, historical failure count score, maintenance interval score, alarm frequency score, and specified operating parameter fluctuation score, and using an equipment health risk assessment model, the equipment health risk index of any of the equipment is calculated. The equipment health risk assessment model is as follows: R_h = w_1·X_r + w_2·X_f + w_3·X_a + w_4·X_v + w_5·X_m; In the formula, R_h is the equipment health risk index of any device, X_r represents the continuous running time score, X_f represents the historical failure count score, X_a represents the alarm frequency score, X_v represents the specified operating parameter fluctuation score, X_m represents the maintenance interval score, and w_1 to w_5 are the weights of each score of any device during the current inspection.

[0012] In one possible design, the weights of each score are calculated using a gradient update strategy based on prediction error, whereby the gradient update strategy based on prediction error is: Obtain the true health label of any device at the time of the last inspection, as well as the historical health risk index and historical weight of each score at the time of the last inspection, wherein the true health label is 1 or 0; The health risk prediction error is calculated based on the true health label and the device health risk index of any of the devices. Based on the health risk prediction error, the historical weights of each score are updated to obtain the weights of each score for any device during the current inspection.

[0013] In one possible design, based on the health risk prediction error, the historical weights of each score are updated to obtain the weights of each score for any device during the current inspection, including: Based on the health risk prediction error, the initial weights of each score of any device during the current inspection are calculated, and the initial weights are normalized to obtain the weights of each score of any device during the current inspection. The initial weights of each score for any device during the current inspection are calculated according to the following formula. w_k,temp^(b) = max{0, w_k^(b-1) + η·e^(b-1)·X_k^(b-1)}; In the formula, w_k,temp^(b) represents the initial weight of the k-th score of any device in the current inspection, w_k^(b-1) represents the historical weight of the k-th score, η represents the preset learning rate, e^(b-1) represents the health risk prediction error, and X_k^(b-1) represents the k-th score of any device in the previous inspection, where k=1,2,...,5.

[0014] In one possible design, a device relationship model of the system to be inspected is constructed, and based on the device relationship model, the relationship impact value of each device is calculated, including: Establish a directed topology graph of the equipment in the system to be inspected, wherein each node in the directed topology graph is used to represent each device, and if there is a directed edge between two nodes, it indicates that there is an influence path between the devices corresponding to the two nodes. Based on the directed topology graph of the devices, a physical topology adjacency matrix is ​​determined. In the physical topology adjacency matrix, the element a_ij represents the physical topology constraint value between device i and device j. a_ij is 1 or 0. When a_ij is 1, it means that there is an influence path between device i and device j. Device i is used to influence device j through energy flow, pipeline connection or control signal. When a_ij is 0, it means that there is no influence path between device i and device j. The device combinations corresponding to the elements with a physical topology constraint value of 1 are selected from the physical topology adjacency matrix, and a candidate device combination set is formed using the selected device combinations. For any candidate device combination in the candidate device combination set, obtain the historical energy saving deviation value sequence of two candidate devices in the candidate device combination within a preset historical statistical period. For any candidate device combination, the affected candidate device in the candidate device combination is taken as the target device, and the other candidate device is taken as the source device of influence. Based on the historical energy-saving deviation value sequence corresponding to each of the two candidate devices, the correlation weight between the two candidate devices in any candidate device combination is calculated, and after all candidate device combinations have been traversed, the correlation weight between the two candidate devices in all candidate device combinations is obtained. Based on the association weights between two candidate devices in all candidate device combinations, an association matrix is ​​constructed, wherein the row subscripts in the association matrix represent the source device, the column subscripts represent the target device, and the element r_qp in the association matrix represents the association weight between the source device p and the target device q. Based on the correlation matrix and the energy-saving deviation values ​​of each influencing source device, the correlation impact value of each device is calculated.

[0015] In one possible design, based on the historical energy-saving deviation value sequences corresponding to each of the two candidate devices, the correlation weight between the two candidate devices in any combination of candidate devices is calculated, including: Based on the historical energy-saving deviation value sequences of the two candidate devices, the time-delay Pearson correlation coefficients of the two candidate devices at different time delays within the preset time delay range are calculated, and the maximum value among the absolute values ​​of each time-delay Pearson correlation coefficient is taken as the data correlation strength between the two candidate devices. The initial association weight between the two candidate devices is obtained by multiplying the physical topology constraint values ​​corresponding to the two candidate devices by the data association strength. Based on the directed topology graph of the devices, downstream candidate devices of the source device in any candidate device combination are determined; The initial correlation weights between each downstream candidate device and the source device are obtained. Based on the initial correlation weights between each downstream candidate device and the source device, the initial correlation weights between the two candidate devices in any candidate device combination are normalized to obtain the correlation weights between the two candidate devices in any candidate device combination.

[0016] In one possible design, the time-delay Pearson correlation coefficients of the two candidate devices are calculated for different time delays within a preset time delay range, including: The time-delay Pearson correlation coefficient of the two candidate devices at any time delay is calculated using the following formula; ; In the formula, Indicates the time delay of the two candidate devices at any given time. The time-delayed Pearson correlation coefficient, This represents the historical energy-saving deviation value of candidate device p at time t in the historical energy-saving deviation value sequence of the two candidate devices. This represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device p. This represents the first value in the historical energy-saving deviation sequence of candidate device q among two candidate devices. Historical energy saving deviation at time _____ The value q represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device q, and T represents the total duration of the preset historical statistical period. ,and This indicates the preset time delay range; Accordingly, based on the correlation matrix and the energy-saving deviation values ​​of each influencing source device, the correlation impact value of each device is calculated, including: For any given device, obtain all influencing source devices corresponding to that device. Based on the correlation matrix, the correlation weight between any device and its corresponding source devices is determined. The correlation weight between any device and its corresponding source devices is multiplied by the energy-saving deviation value of each source device, and the sum of the products is obtained to obtain the correlation impact value of any device.

[0017] In one possible design, based on the inspection benefit evaluation model, the inspection benefit index corresponding to each device after inspection is calculated, including: For any device, calculate the theoretical energy consumption of the device under the current operating state, and collect the actual energy consumption of the device under the current operating state, so as to calculate the energy loss rate of the device per unit time based on the theoretical energy consumption and actual energy consumption of the device. From the historical maintenance database, filter out historical inspection and maintenance records that match the current operating status of any equipment; For any historical inspection and maintenance record, calculate the first average energy loss rate of any device during a preset statistical period before the historical inspection and maintenance record, and the second average energy loss rate during a preset period after the historical inspection and maintenance record. Based on the first average energy loss rate and the second average energy loss rate, the energy efficiency recovery rate corresponding to any historical inspection and maintenance record is calculated, and after all historical inspection and maintenance records have been traversed, the energy efficiency recovery rate corresponding to each historical inspection and maintenance record is obtained. Take the arithmetic mean or median of all energy efficiency recovery rates as the expected energy recovery rate for any of the devices. Obtain the inspection cost and equipment importance level of any of the aforementioned devices; Based on the energy loss rate per unit time, the expected energy-saving recovery rate, the inspection cost, the equipment importance level, and the energy-saving deviation value of any equipment, and using the inspection benefit evaluation model, the inspection benefit index corresponding to any equipment is calculated. The inspection benefit evaluation model is as follows: ; In the formula, This represents the model value of the inspection benefit evaluation model, and also represents the inspection benefit index corresponding to any of the aforementioned devices. This indicates the energy-saving deviation value of any device. This represents the energy loss rate per unit time. Indicates the importance level of the equipment. This indicates the projected energy recovery rate. This represents the inspection cost.

[0018] Secondly, a system for dynamically generating inspection tasks based on energy-saving operation data is provided, including: The acquisition unit is used to acquire real-time operating data and historical operating records of each device in the system to be inspected. The energy-saving deviation calculation unit is used to calculate the energy-saving deviation value of each device based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state. The equipment health risk assessment unit is used to calculate the equipment health risk index of each piece of equipment based on the historical operating records of each piece of equipment and using a pre-built equipment health risk assessment model. The correlation impact assessment unit is used to construct the equipment correlation relationship model of the system to be inspected, and calculate the correlation impact value of each equipment based on the equipment correlation relationship model. The correlation impact value of any equipment is used to characterize the degree of influence of the energy saving deviation of the source equipment on the equipment. The larger the correlation impact value, the greater the degree of influence. The source equipment is all equipment in the system to be inspected that has an influence path with the equipment. The inspection benefit calculation unit is used to establish an inspection benefit evaluation model and, based on the inspection benefit evaluation model, calculate the inspection benefit index obtained by each device after inspection. The inspection strategy generation unit is used to calculate the comprehensive inspection score of each piece of equipment by utilizing the energy-saving deviation value, equipment health risk index, correlation impact value and inspection benefit index of each piece of equipment. The inspection strategy generation unit is also used to generate inspection strategies for the system to be inspected based on the comprehensive inspection scores of each device.

[0019] Thirdly, a device for dynamically generating inspection tasks based on energy-saving operation data is provided. Taking the device as an electronic device as an example, it includes a memory, a processor, and a transceiver that are connected in sequence. The memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer programs and execute the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect or any possible design in the first aspect.

[0020] Fourthly, a storage medium is provided, on which instructions are stored, which, when executed on a computer, perform the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect or any possible design of the first aspect.

[0021] Fifthly, a computer program product containing instructions is provided, which, when executed on a computer, causes the computer to execute the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect or any possible design of the first aspect.

[0022] Beneficial effects: (1) Breaking through the limitations of passive inspection, realizing the active identification and early warning of energy-saving hazards; This invention no longer relies on fixed cycles or fault alarms to trigger inspections, but instead constructs an "energy-saving deviation evaluation model" to calculate the degree of deviation between the actual energy consumption and theoretical energy consumption of the equipment in real time; In this way, this invention can keenly capture the hidden abnormal states of the equipment such as decreased energy efficiency and increased energy consumption before the occurrence of obvious faults, thereby turning the inspection task from "post-event remediation" to "pre-event intervention", and thus effectively avoiding long-term energy waste caused by the degradation of energy-saving performance.

[0023] (2) Introducing equipment correlation impact analysis to achieve system-level global optimal operation and maintenance decision-making; by constructing an equipment correlation relationship model and calculating the correlation impact value, this invention can quantify the degree of transmission impact of energy-saving deviation of a certain equipment on other equipment. This mechanism makes the inspection task no longer view a single piece of equipment in isolation, but starts from the coupling relationship of the entire energy system (such as central air conditioning, cold and heat sources, lighting, power equipment, etc.), and prioritizes the inspection of key equipment with the greatest impact on system energy efficiency, thereby maximizing the system energy-saving benefits of inspection investment.

[0024] (3) Integrating multi-dimensional evaluation factors significantly improves the scientific nature and accuracy of inspection task generation; This invention innovatively performs comprehensive weighted calculation of four types of indicators: energy-saving deviation value, equipment health risk index, correlation impact value, and inspection benefit index, forming a unified comprehensive inspection score. Compared with existing technologies that only focus on equipment reliability or failure risk, this invention also takes into account the energy-saving status of the equipment itself, safety risks, system-level linkage impact, and economic returns, making inspection decisions more comprehensive and objective.

[0025] (4) Improve the overall economic efficiency and energy-saving sustainability of the energy management system; Since the inspection benefit index is directly incorporated into the evaluation model, this invention can guide maintenance resources to be prioritized for the equipment with the highest energy-saving return after inspection. This intelligent operation and maintenance method, which is guided by energy-saving value, not only helps to maintain the equipment in the high-efficiency operating range in the long term, but also continuously optimizes the overall system energy efficiency level, which can bring quantifiable economic benefits such as electricity and heat saving to enterprises, and help achieve carbon neutrality and sustainable development goals.

[0026] (5) Dynamically optimize the inspection strategy to avoid redundant inspections and reduce operation and maintenance costs. The present invention dynamically calculates the comprehensive score based on the real-time operation data of the equipment and generates the inspection strategy on demand accordingly, which completely changes the traditional mechanical mode of "fixed cycle full coverage". In this way, the inspection objects, priorities, contents and frequencies can be adaptively adjusted with the changes in equipment status and energy consumption, thereby effectively avoiding invalid inspections and excessive inspections and improving the intelligent operation and maintenance level of the energy system. Attached Figure Description

[0027] Figure 1 A flowchart illustrating the steps of the method for dynamically generating inspection tasks based on energy-saving operation data provided in this embodiment of the invention; Figure 2 This is a structural diagram of the inspection task dynamic generation system based on energy-saving operation data provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in conjunction with the accompanying drawings and descriptions of the embodiments or the prior art. Obviously, the following description of the structure of the accompanying drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation of the present invention.

[0029] It should be understood that although the terms first, second, etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit, without departing from the scope of the exemplary embodiments of the invention.

[0030] It should be understood that the term "and / or" that may appear in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" that may appear in this document describes another relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this document generally indicates that the related objects before and after it are in an "or" relationship.

[0031] Example: See Figure 1 As shown, the method for dynamically generating inspection tasks based on energy-saving operation data provided in the first aspect of this embodiment can be executed by a computer device with certain computing resources, including but not limited to a server or edge computer. It is understood that the aforementioned execution subject does not constitute a limitation on the embodiments of this application. Accordingly, the operation steps of this method can be, but are not limited to, the steps S1 to S7 below.

[0032] S1. Obtain real-time operating data and historical operating records of each device in the system to be inspected. In specific implementation, the system to be inspected may include, but is not limited to, central air conditioning systems, cold and heat source systems, lighting control systems, power equipment systems, etc., and will be specifically selected according to actual use. At the same time, the real-time operating data of any device may include, but is not limited to, information such as energy consumption per unit time, load rate, running time, ambient temperature, and set parameters. At the same time, the historical operating records of any device may include, but are not limited to, the continuous running time of the device, as well as the number of historical faults, maintenance intervals, alarm frequencies, and fluctuation rates of specified operating parameters within a historical preset time period. Among them, corresponding specified operating parameters can be pre-configured according to different devices. For example, for chiller units, specified operating parameters can be set as load rate, chilled water return temperature, chilled water supply temperature, etc. At present, the specified operating parameters of other different devices will not be described in detail.

[0033] Thus, after obtaining the real-time operating data and historical operating records of each device, the energy-saving deviation value of each device can be calculated by using the pre-built energy-saving deviation evaluation model, as shown in step S2 below.

[0034] S2. Based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model, calculate the energy-saving deviation value of each device. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state. In specific applications, this embodiment first establishes a benchmark energy consumption dynamic prediction model, and then calculates the theoretical energy consumption of the device based on the prediction model. Finally, the energy-saving deviation value of the device is calculated by combining the theoretical energy consumption, actual energy consumption and the aforementioned energy-saving deviation evaluation model. The process is shown in steps S21 to S25 below.

[0035] S21. For any device, obtain the operating parameters of the device during normal operation as historical benchmark operating data; in this embodiment, the aforementioned operating parameters during normal operation are the historical operating parameters of the device when it is in a normal, stable and fault-free state, and the data type is the same as the data type in the real-time operating data of the device; thus, after obtaining the historical benchmark operating data, a benchmark energy consumption dynamic prediction model can be constructed based on it, as shown in step S22 below.

[0036] S22. Based on historical benchmark operating data, construct a benchmark energy consumption dynamic prediction model for any of the devices. In specific implementation, for example, but not limited to, using a multiple linear regression algorithm to establish the benchmark energy consumption dynamic prediction model for any device, that is, the theoretical energy consumption E_b of the device is not a fixed lookup value, but is calculated in real time by the model based on the current operating conditions.

[0037] Specifically, in one embodiment, a chiller unit is used as the modeling object. Historical operating parameters for at least 30 consecutive days are selected as initial samples, and a sample sequence is formed according to a preset sampling period. Then, after removing abnormal data from equipment start-up / shutdown phases, fault alarm phases, sensor malfunctions, and deviations from the sample mean by more than three standard deviations, the following are selected as input features: load rate L1, outdoor ambient temperature T_e, chilled water supply temperature T_s, chilled water return temperature T_r, cooling water inlet temperature T_c, and cumulative equipment operating time T_h. The actual energy consumption E within the corresponding sampling period is used as the target value, and each input feature is standardized. Then, the training set and validation set are divided in an 8:2 ratio. Finally, the least squares method is used to solve for the regression coefficients, thereby constructing a benchmark energy consumption dynamic prediction model, the expression of which is: E_b = β_0 + β_1·L 1+ β_2·T_e + β_3·T_s + β_4·T_r + β_5·T_c + β_6·T_h; In the formula, E_b represents the theoretical energy consumption, β_0 is the intercept term, and β_1 to β_6 are the regression coefficients obtained based on the training set.

[0038] Thus, after the model training is completed, it is validated using a validation set. That is, when the mean absolute percentage error on the validation set is not higher than the preset error threshold, the model is determined to be an effective benchmark energy consumption model. If it exceeds the aforementioned threshold, the training samples are re-selected and the model is trained. As new normal operation data accumulates, the training samples and regression coefficients can be updated on a rolling basis according to a preset cycle to adapt to changes in equipment performance and seasonal operating conditions.

[0039] After constructing the benchmark energy consumption dynamic prediction model, the theoretical energy consumption of any device can be calculated based on it, as shown in step S23 below.

[0040] S23. Based on the real-time operating data of any device and the benchmark energy consumption dynamic prediction model, calculate the theoretical energy consumption of any device under the current operating state. In specific applications, input the aforementioned characteristic parameters (i.e., the real-time operating data of any device) of the current sampling period into the model to output the theoretical energy consumption E_b corresponding to the current load and environmental conditions. Then, based on the actual energy consumption of any device and in conjunction with the aforementioned energy-saving deviation evaluation model, calculate the energy-saving deviation value of any device, as shown in steps S24 and S25 below. S24. Collect the actual energy consumption of any of the devices in the current operating state.

[0041] S25. Calculate the energy-saving deviation value of any of the devices based on actual energy consumption, theoretical energy consumption, and the energy-saving deviation evaluation model.

[0042] In practical applications, the energy-saving deviation evaluation model can be expressed as: ; In the formula, The model value of the energy-saving deviation evaluation model represents the energy-saving deviation value of any given device. This represents the actual energy consumption of any of the aforementioned devices. This represents the theoretical energy consumption of any of the aforementioned devices. This indicates the runtime (i.e. duration) of any device.

[0043] Thus, based on the aforementioned formula, it can be seen that when the equipment energy consumption is continuously higher than the baseline energy consumption, the energy-saving deviation value continues to increase as the duration of the abnormality increases. Among them, when the energy-saving deviation value exceeds the preset threshold, even if the equipment has not yet triggered an alarm, it can be determined that the equipment has an abnormal energy-saving trend and enters the inspection candidate team. Therefore, this model focuses on the downward trend of equipment energy-saving efficiency, rather than just the equipment fault status.

[0044] Thus, based on the aforementioned steps S21 to S25, after calculating the energy-saving deviation value of each device, the device health risk index can be calculated, as shown in step S3 below.

[0045] S3. Based on the historical operating records of each device and using a pre-built device health risk assessment model, calculate the device health risk index for each device. In specific applications, the continuous operating time, the number of historical failures, the maintenance interval, the alarm frequency, and the fluctuation rate of the specified operating parameters can be normalized first to obtain a continuous operating time score, a historical failure count score, a maintenance interval score, an alarm frequency score, and a specified operating parameter fluctuation score. Then, based on the continuous operating time score, the historical failure count score, the maintenance interval score, the alarm frequency score, and the specified operating parameter fluctuation score, and using the device health risk assessment model, calculate the device health risk index for any device.

[0046] In one specific implementation, the reason for performing normalization is to eliminate the dimensional differences between the above evaluation factors. For example, but not limited to, the Min-Max method can be used for normalization, that is, the minimum and maximum values ​​of each evaluation factor of the same type of equipment within a preset historical statistical period are used as the normalization boundary.

[0047] Specifically, for the four positive evaluation factors—continuous operating time, number of historical failures, alarm frequency, and volatility of specified operating parameters—where higher values ​​indicate higher risk, the dimensionless feature scores are calculated using the following formula: X_i = (x_i - x_i,min) / (x_i,max - x_i,min); X_i represents the dimensionless feature score of the i-th evaluation factor (i=1,2,3,4, representing continuous running time, number of historical failures, alarm frequency and fluctuation rate of specified operating parameters, respectively), x_i represents the i-th evaluation factor, and x_i,min and x_i,max represent the minimum and maximum values ​​of the i-th evaluation factor, respectively.

[0048] For the converse indicator of lower maintenance intervals, which represents higher risk, the dimensionless characteristic score is calculated using the following formula: X_m = (T_m,max - T_m) / (T_m,max - T_m,min); In the formula, X_m represents the dimensionless feature score (i.e., maintenance interval score) corresponding to the maintenance interval, T_m represents the maintenance interval, and T_m,max and T_m,min represent the maximum and minimum values ​​corresponding to the maintenance interval.

[0049] Thus, the normalized feature scores are all limited to the [0,1] interval; if the current value exceeds the historical boundary, the score is truncated to 0 or 1; when the maximum value and minimum value of a certain indicator are equal, the feature score is set to 0.

[0050] Based on this, after normalizing the aforementioned continuous operating time, the number of historical failures, the maintenance interval, the alarm frequency, and the volatility of the specified operating parameters, the equipment health risk assessment model can be used to calculate the equipment health risk assessment model for any given device.

[0051] The equipment health risk assessment model can be expressed as follows: R_h = w_1·X_r + w_2·X_f + w_3·X_a + w_4·X_v + w_5·X_m; In the formula, R_h is the equipment health risk index of any device, X_r represents the continuous running time score, X_f represents the historical failure count score, X_a represents the alarm frequency score, X_v represents the specified operating parameter fluctuation score, X_m represents the maintenance interval score, and w_1 to w_5 are the weights of each score of any device during the current inspection.

[0052] Thus, based on the aforementioned formula, the equipment health risk index R_h ranges from [0,1]. The larger the value, the higher the potential risk of equipment abnormality, and the higher its priority in inspection decisions.

[0053] Furthermore, the weights of the aforementioned scores can be calculated, but are not limited to, using a gradient update strategy based on prediction error, wherein the gradient update strategy based on prediction error can be, but is not limited to: Step 1: Obtain the health authenticity label of any device at the time of the last inspection, as well as the historical health risk index and historical weight of each score at the time of the last inspection, wherein the health authenticity label is 1 or 0.

[0054] In practice, after the previous inspection, on-site personnel use the inspection terminal to input the fault type, fault location, degree of abnormality, and handling result of any equipment. The system converts the inspection results into a feedback label y (i.e., a health accuracy label). When a fault or energy efficiency abnormality is confirmed, y is set to 1; when the equipment is confirmed to be normal, y is set to 0. At the same time, for inspection results with different degrees of abnormality, y can also be set to a continuous value in the range [0,1] according to a preset mapping table.

[0055] Thus, after obtaining the true health label, the prediction error can be calculated, as shown in step 2 below.

[0056] Step 2: Calculate the health risk prediction error based on the true health label and the device health risk index of any of the devices; in specific implementation, e^(b-1) = y^(b-1) - R_h^(b-1); where e^(b-1) represents the health risk prediction error, y^(b) represents the true health label, and R_h^(b-1) represents the historical health risk index.

[0057] Thus, after obtaining the health risk prediction error, the historical weights of each score can be updated based on this, as shown in step 3 below.

[0058] Step 3: Based on the health risk prediction error, update the historical weights of each score to obtain the weights of each score for any device during the current inspection.

[0059] In practice, but not limited to, the initial weights of each score of any device during the current inspection can be calculated first based on the health risk prediction error; then, the initial weights are normalized to obtain the weights of each score of any device during the current inspection.

[0060] Optionally, for example but not limited to, the initial weights of each score for any device during the current inspection can be calculated according to the following formula.

[0061] w_k,temp^(b) = max{0, w_k^(b-1) + η·e^(b-1)·X_k^(b-1)}; In the formula, w_k,temp^(b) represents the initial weight of the k-th score of any device in the current inspection, w_k^(b-1) represents the historical weight of the k-th score, η represents the preset learning rate, e^(b-1) represents the health risk prediction error, and X_k^(b-1) represents the k-th score of any device in the previous inspection, where k=1,2,...,5.

[0062] Thus, after calculating the initial weights of each score during the current inspection, they can be normalized by summing the initial weights of all scores to obtain the total weights. Then, by dividing the initial weights of each score by the total weights, the weights of each score for any device during the current inspection can be obtained.

[0063] Based on this, the max operation is used to ensure that the weights are non-negative; the normalization process is used to ensure that the sum of the updated weights is 1; when the normalization denominator is 0, the weights before this update remain unchanged.

[0064] Thus, when a fault is confirmed on-site and R_h^(b-1) is low, e^(b-1)>0, and the weight corresponding to the feature that simultaneously shows a high score for that fault increases in the next inspection. Conversely, when the equipment is confirmed to be normal on-site and R_h^(b-1) is high, e^(b-1)<0, and the corresponding feature weight decreases in the next inspection. This allows the subsequent equipment health risk index to gradually approach the on-site inspection results. Simultaneously, to prevent drastic weight changes caused by a single abnormal feedback, the learning rate η is set within a preset range, and the weights can be updated in batches after accumulating a preset number of valid inspection samples.

[0065] Furthermore, this implementation also sets a health risk judgment threshold so that after calculating the current equipment health risk index, it is compared with the health risk judgment threshold to identify whether the equipment is faulty and to trigger a fault alarm.

[0066] Optionally, the predicted label can be obtained by first comparing the historical health risk index R_h^(b-1) with the health risk judgment threshold R_h,0^(b-1) at the time of the last inspection. ^(b-1), that is: when R_h^(b-1)≥R_h,0^(b-1), let ^(b-1)=1, otherwise, let ^(b-1)=0; Meanwhile, the health risk assessment threshold for the next inspection will be adjusted according to the following formula: R_h,0^(b) = clip[R_h,0^(b-1) - η_h·(y^(b-1)- ^(b-1)), R_h,min, R_h,max]; In the formula, R_h,0^(b) represents the updated health risk assessment threshold, R_h,0^(b-1) represents the health risk assessment threshold, and η_h represents the threshold learning rate. ^(b-1) represents the predicted label, y^(b-1) represents the true health label, R_h,min and R_h,max represent the minimum and maximum decision thresholds, respectively, and clip[] represents the truncation function.

[0067] In this embodiment, the clip function is used to limit the updated threshold within a preset range [R_h,min,R_h,max]. When y^(b-1)=1 and When y^(b-1)=0, the threshold is lowered when a false negative occurs; when y^(b-1)=0 and ^(b-1)=1, meaning that the threshold is increased when a misjudgment occurs, and the threshold remains unchanged when the prediction is correct.

[0068] In this way, through the above feedback and correction, the system forms a closed-loop management mechanism of "operation monitoring - intelligent decision-making - on-site inspection - error calculation - weight and threshold update - model retraining".

[0069] Therefore, based on the aforementioned step S3 and its sub-steps, after calculating the health risk index of each device, the associated impact value at each time point can be calculated, as shown in step S4 below.

[0070] S4. Construct a device association model for the system to be inspected, and calculate the association impact value of each device based on the device association model. The association impact value of any device is used to characterize the degree of influence of the energy-saving deviation of the source device on that device. The larger the association impact value, the greater the degree of influence. The source device is all devices in the system to be inspected that have an influence path with that device. In this embodiment, the influence path is an energy flow, pipeline connection, or control signal path. That is, devices that can have an energy flow, pipeline connection, or control signal transmission relationship with that device are considered as the source devices of that device.

[0071] In practical implementation, to address the issue of interconnected relationships between various devices in the system to be inspected, this embodiment establishes a device relationship model for correlation analysis; for example, chiller unit → cooling pump → cooling tower → air conditioning terminal; if one of the devices deviates from energy efficiency, not only is the device itself evaluated, but also the degree of impact on its related devices is analyzed; for example, a decrease in cooling tower efficiency may lead to an increase in chiller power consumption and an increase in the overall energy consumption of the chiller plant.

[0072] In one specific implementation, the inter-device association weight is not determined solely by physical topology or data correlation, but is calculated by fusing physical topology constraints and operational data association strength. The calculation process can be, but is not limited to, the steps S41 to S47 below.

[0073] S41. Establish a directed topology graph of the equipment in the system to be inspected, wherein each node in the directed topology graph represents each device, and if there is a directed edge between two nodes, it represents that there is an influence path between the devices corresponding to the two nodes; in specific implementation, for example, but not limited to, the directed topology graph of the equipment can be established based on the connection relationship of the devices, the flow direction of the energy medium and the control logic relationship in the system to be inspected, that is, if there is an energy flow, pipeline connection or control information (i.e., influence path) between device i and device j, it means that device i can affect device j through energy flow, pipeline connection or control signal, and there is a directed edge between device i and device j.

[0074] Thus, after constructing the aforementioned directed topology graph of the devices, the physical topology adjacency matrix can be determined based on it, as shown in step S42 below.

[0075] S42. Based on the directed topology graph of the devices, determine the physical topology adjacency matrix, wherein the element a_ij in the physical topology adjacency matrix represents the physical topology constraint value between device i and device j, a_ij is 1 or 0, and when a_ij is 1, it indicates that there is an influence path between device i and device j, and device i is used to influence device j through energy flow, pipeline connection or control signal; when a_ij is 0, it indicates that there is no influence path between device i and device j.

[0076] In practical applications, the physical topology relationship in this step is only used to determine whether there is a potential influence path between device i and device j and to limit the candidate influence range. It does not directly determine the magnitude of the final association weight. In this way, it can avoid generating an association relationship that lacks actual physical meaning based solely on data correlation. Among them, the direction of device influence is determined by the aforementioned directed topology graph, so a_ij and a_ji do not have to be equal.

[0077] After constructing the physical topology adjacency matrix, device screening can be performed, as shown in step S43 below.

[0078] S43. Select the device combinations corresponding to the elements with a physical topology constraint value of 1 from the physical topology adjacency matrix, and use the selected device combinations to form a candidate device combination set; in this embodiment, a physical topology constraint value of 1 indicates that there is an influence path between two devices, that is, one device can influence another device through energy flow, pipeline connection or control signal; therefore, only the device combinations corresponding to the elements with a physical topology constraint value of 1 are considered as valid association combinations.

[0079] After obtaining the candidate device combination set, the corresponding historical energy saving deviation value sequence can be obtained. Based on this, the correlation weight between the two candidate devices in each candidate device combination can be calculated. The process is shown in steps S44 and S45 below.

[0080] S44. For any candidate device combination in the candidate device combination set, obtain the historical energy-saving deviation value sequence of two candidate devices in the candidate device combination within a preset historical statistical period. For any candidate device combination, the affected candidate device in the candidate device combination is taken as the target device, and the other candidate device is taken as the source device. In specific implementation, the calculation process of historical energy-saving deviation value is the same as the aforementioned energy-saving offset value calculation process, and will not be repeated here.

[0081] After obtaining the historical energy-saving deviation value sequence of two candidate devices in any candidate device combination, the correlation weight between the two can be calculated based on this, as shown in step S45 below.

[0082] S45. Based on the historical energy-saving deviation value sequence corresponding to each of the two candidate devices, calculate the correlation weight between the two candidate devices in any candidate device combination, and obtain the correlation weight between the two candidate devices in all candidate device combinations after traversing all candidate device combinations; in specific implementation, for example, but not limited to, the following steps S45a to S45d can be used to calculate the correlation weight between the aforementioned two candidate devices.

[0083] S45a. Based on the historical energy-saving deviation value sequences of the two candidate devices, calculate the time-delay Pearson correlation coefficients of the two candidate devices at different time delays within the preset time delay range, and take the maximum value among the absolute values ​​of each time-delay Pearson correlation coefficient as the data correlation strength between the two candidate devices.

[0084] Optionally, for example but not limited to, the time-delay Pearson correlation coefficient of two candidate devices at any time delay can be calculated according to the following formula.

[0085] ; In the formula, Indicates the time delay of the two candidate devices at any given time. The time-delayed Pearson correlation coefficient, This represents the historical energy-saving deviation value of candidate device p at time t in the historical energy-saving deviation value sequence of the two candidate devices. This represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device p. This represents the first value in the historical energy-saving deviation sequence of candidate device q among two candidate devices. Historical energy saving deviation at time _____ The value q represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device q, and T represents the total duration of the preset historical statistical period. ,and This indicates the preset time delay range.

[0086] Thus, by updating the time lag time according to the preset step size, the time lag Pearson correlation coefficients of the two candidate devices at each time lag time can be obtained; then, by taking the maximum value among the absolute values ​​of each time lag Pearson correlation coefficient, the data association strength between the two candidate devices can be obtained; furthermore, when the data association strength is less than the set value, the data association strength is set to 0, thereby reducing the impact of random fluctuations.

[0087] Thus, after calculating the data correlation strength between the two candidate devices, a fusion calculation can be performed and the weights normalized, as shown in steps S45b to S45d below.

[0088] S45b. Multiply the physical topology constraint values ​​corresponding to the two candidate devices by the data association strength to obtain the initial association weight between the two candidate devices.

[0089] Thus, after calculating the initial association weights, normalization can be performed, as shown in steps S45c and S45d below.

[0090] S45c. Based on the directed topology graph of the devices, determine the downstream candidate devices of the source device in any candidate device combination.

[0091] S45d. Obtain the initial association weights between each downstream candidate device and the source device, and normalize the initial association weights between the two candidate devices in any candidate device combination based on the initial association weights between each downstream candidate device and the source device to obtain the association weights between the two candidate devices in any candidate device combination.

[0092] In this embodiment, the initial association weights between each downstream candidate device and the influencing source device are summed to obtain the total association weights. Then, the initial association weights are divided by the total initial association weights to obtain the association weights between two candidate devices in any candidate device combination.

[0093] Based on the aforementioned steps S45a to S45d, after calculating the correlation weight between two candidate devices in each candidate device combination, a correlation matrix can be constructed based on this, as shown in step S46 below.

[0094] S46. Based on the association weights between two candidate devices in all candidate device combinations, an association matrix is ​​constructed, wherein the row subscripts in the association matrix represent the source device, the column subscripts represent the target device, and the element r_qp in the association matrix represents the association weight between the source device p and the target device q.

[0095] In practice, physical topology constraints are used to determine whether there are potential influence paths between devices. The data association strength represented by the time-delay Pearson correlation coefficient determines the strength of the candidate influence relationship in actual operation. Finally, the association weight is determined by both physical influence constraints and historical operation data association, rather than by either factor alone.

[0096] Optionally, the correlation matrix can be represented as: ; In the formula, r1n represents the association weight of the first source influencing device to the target device n, and rnn represents the association weight of the nth source influencing device to the target device n.

[0097] Based on this, the matrix element in row p and column q represents the influence weight (association weight) of the source device p on the target device q. The larger the value of r_pq, the higher the degree of influence of the energy-saving deviation of the source device p on the target device q. At the same time, since the device association is directional, r_pqj represents the influence weight in the p→q direction, while r_qp represents the influence weight in the q→p direction. Therefore, the two have different meanings and their values ​​do not have to be equal.

[0098] Thus, based on the aforementioned steps, after constructing the correlation matrix, the correlation impact value of each device can be calculated by combining the energy-saving deviation values ​​of each influencing source device, as shown in step S47 below.

[0099] S47. Based on the correlation matrix and the energy-saving deviation values ​​of each influencing source device, calculate the correlation impact value of each device. In this embodiment, for any device (referring to the target device), it is possible, but not limited to, to obtain all influencing source devices corresponding to the target device. Then, based on the correlation matrix, determine the correlation weight between the target device and each corresponding influencing source device. Finally, multiply the correlation weight between the target device and each corresponding influencing source device by the energy-saving deviation values ​​of each influencing source device corresponding to the target device, and calculate the sum of the products to obtain the correlation impact value of the target device.

[0100] In practice, for any device q (i.e. target device q), the column in the correlation matrix of any device q is obtained. Then, the elements of the column are weighted and summed with the energy-saving deviation values ​​of the corresponding source devices to obtain the correlation influence value of any device q.

[0101] Thus, the higher the correlation impact value of the equipment, the more significant the comprehensive impact of the energy-saving deviation of the upstream or related influence source equipment is. Therefore, the inspection priority of the equipment and related influence source equipment should be increased accordingly.

[0102] Based on the aforementioned steps S41 to S47, after calculating the associated impact values ​​of each device, the inspection benefits can be calculated, as shown in step S5 below.

[0103] S5. Establish an inspection benefit evaluation model, and calculate the inspection benefit index obtained by each device after inspection based on the inspection benefit evaluation model. In specific applications, the constructed inspection benefit evaluation model no longer uses the failure probability as the sole basis, but comprehensively analyzes the energy-saving benefits that may be brought about after inspection. Its benefit evaluation factors include: energy-saving deviation value, energy loss rate per unit time, equipment importance level, inspection cost, and expected skill recovery rate. Therefore, in this embodiment, the inspection benefit index obtained by each device after inspection is calculated by calculating the aforementioned benefit evaluation factors, and the process is shown in steps S51 to S57 below.

[0104] S51. For any device, calculate the theoretical energy consumption of the device under the current operating state, and collect the actual energy consumption of the device under the current operating state, so as to calculate the energy loss rate of the device per unit time based on the theoretical energy consumption and actual energy consumption of the device; in this embodiment, the calculation process of theoretical energy consumption has been described in the aforementioned step S2, and the process will not be repeated.

[0105] Once the theoretical energy consumption is obtained, the rate of energy loss per unit time can be calculated using the following formula: L = max(E_a-E_b, 0) / Δt; In the formula, L represents the energy loss rate per unit time of any device, E_a is the actual energy consumption of any device, E_b is the theoretical energy consumption of any device, and Δt represents the current sampling period duration.

[0106] In this embodiment, L represents the energy loss of the device per unit time that exceeds the baseline energy consumption; when E_a≤E_b, let L=0; where L is not directly converted from S_d, but is calculated from the difference between the actual energy consumption and the baseline energy consumption; S_d is used to reflect the degree of relative deviation and its persistence, and L is used to reflect the absolute energy loss per unit time. The two respectively characterize the relative severity of the anomaly and the absolute loss rate.

[0107] Thus, based on the aforementioned formula, after calculating the energy loss rate per unit time for each device, the expected energy recovery rate can be calculated, as shown in steps S52 to S55 below.

[0108] S52. From the historical maintenance database, select historical inspection and maintenance records that match the current operating status of any equipment. In this embodiment, the historical maintenance database may include, but is not limited to, the equipment management database and the maintenance work order database. Among them, historical inspection and maintenance records that are the same as or similar to the current equipment type, fault type and load range (load difference within a preset range) are selected from the database. Then, for each historical inspection and maintenance record, calculate the first average energy loss rate of any equipment in the preset statistical period before each record and the second average energy loss rate in the preset stable operation period after each record (i.e., the average energy loss rate before and after maintenance). Then, calculate the expected energy-saving recovery rate based on the two calculated average energy loss rates, as shown in steps S53 to S55 below.

[0109] S53. For any historical inspection and maintenance record, calculate the first average energy loss rate of the device within a preset statistical period before the historical inspection and maintenance record, and the second average energy loss rate within a preset period after the historical inspection and maintenance record; in this embodiment, the energy loss rate per unit time for each sampling period within the expected statistical period is calculated; then, the average is taken to obtain the first average energy loss rate; of course, the calculation process of the second average energy loss rate is the same, and will not be repeated here.

[0110] After obtaining the average energy loss rate before and after maintenance, the energy efficiency recovery rate can be calculated based on this, as shown in step S54 below.

[0111] S54. Based on the first average energy loss rate and the second average energy loss rate, calculate the energy efficiency recovery rate corresponding to any historical inspection and maintenance record, and obtain the energy efficiency recovery rate corresponding to each historical inspection and maintenance record after all historical inspection and maintenance records have been traversed.

[0112] In practical implementation, examples, but not limited to, using the following formula, can be used to calculate the aforementioned energy efficiency recovery rate.

[0113] P_c = max{0, min[1, (L_pre-L_post) / L_pre]} In practical implementation, P_c represents the energy efficiency recovery rate corresponding to any historical inspection and maintenance record, L_pre is the first average energy loss rate, and L_post represents the second average energy loss rate.

[0114] In this embodiment, when the first average energy loss rate is less than 0, the record is discarded and the energy efficiency recovery rate corresponding to the next historical inspection and maintenance record is calculated.

[0115] Thus, based on the aforementioned formula, after calculating the energy efficiency recovery rate corresponding to each historical inspection and maintenance record, the expected energy-saving recovery rate of any of the devices can be calculated, as shown in step S55 below.

[0116] S55. Take the arithmetic mean or median of all energy efficiency recovery rates as the expected energy-saving recovery rate of any device; in this embodiment, the expected energy-saving recovery rate ranges from [0,1]; wherein, when there is no similar historical record that meets the conditions, the corresponding conservative preset value can be read from the pre-established statistical table of "Equipment Type - Abnormal Type - Maintenance Measures - Energy-Saving Recovery Rate"; the statistical table is updated periodically according to the energy efficiency data after the new maintenance.

[0117] After calculating the expected energy recovery rate of any device, the equipment importance level and inspection cost of any device can be obtained, as shown in step S56 below.

[0118] S56. Obtain the inspection cost and importance level of any of the devices; in this embodiment, the labor cost per hour, estimated inspection hours, spare parts and consumables costs, transportation costs, and estimated downtime losses can be automatically obtained based on the device number and the planned inspection items of any of the devices through the interface of an ERP system, enterprise asset management system, or maintenance management system. Then, the inspection cost is calculated according to the following formula: C = C_labor·T_check + C_material + C_travel + C_shutdown In the formula, C_labor is the cost per labor unit hour, T_check is the estimated inspection time, C_material is the cost of spare parts and consumables, C_travel is the transportation cost, and C_shutdown is the estimated downtime loss. The estimated inspection time is determined by the historical average time of similar inspection work orders. When interface data is missing, the corresponding cost is read from the preset inspection project cost quota table. At the same time, to avoid the denominator being 0 in subsequent calculations, when the calculated C=0, C is set to the preset minimum cost value C_min.

[0119] Furthermore, the importance level of equipment is automatically read from the equipment ledger in the enterprise asset management system and mapped to a preset dimensionless level coefficient.

[0120] Thus, after obtaining the inspection cost and the importance level of the equipment, the inspection benefit index of any equipment can be calculated by combining the aforementioned energy loss rate per unit time, the expected energy recovery rate, and the energy deviation value, as shown in step S57 below.

[0121] S57. Based on the energy loss rate per unit time, the expected energy-saving recovery rate, the inspection cost, the equipment importance level, and the energy-saving deviation value of any equipment, and using the inspection benefit evaluation model, calculate the inspection benefit index corresponding to any equipment.

[0122] In this embodiment, the inspection benefit evaluation model can be expressed as: ; In the formula, This represents the model value of the inspection benefit evaluation model, and also represents the inspection benefit index corresponding to any of the aforementioned devices. This indicates the energy-saving deviation value of any device. This represents the energy loss rate per unit time. Indicates the importance level of the equipment. This indicates the projected energy recovery rate. This represents the inspection cost.

[0123] In this embodiment, the higher the inspection benefit index, the greater the energy-saving benefits that can be obtained by prioritizing inspections.

[0124] Therefore, after calculating the inspection benefit index of each device through the aforementioned steps S51 to S57, the comprehensive inspection score of each device can be calculated by combining the energy-saving deviation value, equipment health risk index, and related impact value of each device. The process is shown in step S6 below.

[0125] S6. Calculate the comprehensive inspection score of each device using the energy-saving deviation value, device health risk index, correlation impact value, and inspection benefit index of each device. In this embodiment, the comprehensive inspection score of each device is obtained by weighted summation of the energy-saving deviation value, device health risk index, correlation impact value, and inspection benefit index of each device. The total weight of the aforementioned four indicators is 1.

[0126] Thus, after obtaining the comprehensive inspection score of each device, an inspection strategy can be generated based on this, as shown in step S7 below.

[0127] S7. Based on the comprehensive inspection score of each device, an inspection strategy for the system to be inspected is generated. In this embodiment, the devices are sorted in descending order of their comprehensive inspection scores. The sorting order of the devices is the order of their inspection priorities, and the sorting order of the devices is the inspection path. Then, the inspection content for each device is retrieved from a preset inspection content library. Finally, an inspection strategy is generated based on the inspection priority, inspection route, and inspection content. The inspection strategy is then sent to the inspection terminal to complete the inspection of each device in the system to be inspected.

[0128] Therefore, through the method for dynamically generating inspection tasks based on energy-saving operation data described in detail in steps S1 to S7 above, this invention achieves dynamic generation and proactive optimization of energy-saving equipment inspection tasks by constructing a multi-dimensional intelligent evaluation system. On the one hand, unlike traditional fixed-cycle inspections or fault alarm triggering mechanisms, this invention utilizes an energy-saving deviation evaluation model to calculate the degree of deviation between the actual energy consumption and theoretical energy consumption of equipment in real time. This allows for the keen detection of early hidden anomalies such as increased energy consumption and decreased efficiency, providing proactive warnings before significant equipment failures occur, thus effectively avoiding the problem of continuous energy waste. On the other hand, by introducing a correlation model between equipment, it is possible to quantify the impact of energy-saving deviation of a single piece of equipment on the overall system. The system prioritizes the inspection of critical equipment that has the greatest impact on system-level energy efficiency by considering the transmission effects of other equipment. Finally, an inspection benefit evaluation model is established to calculate the benefit index of each equipment inspection. This is further combined with energy-saving deviation values, equipment health risk indices, and related impact values ​​for comprehensive scoring. Based on this, an inspection strategy is generated. This approach ensures that inspection decisions not only focus on the status of individual equipment but also consider the coupling effects of equipment and energy-saving returns. It can continuously improve overall energy utilization efficiency while ensuring system reliability, achieving optimal allocation of inspection resources and significantly enhancing the intelligence level and operational economy of the energy management system. Therefore, this invention is highly suitable for large-scale application and promotion.

[0129] like Figure 2 As shown, the second aspect of this embodiment provides a software system for implementing the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect of the embodiment, comprising: The acquisition unit is used to acquire real-time operating data and historical operating records of each device in the system to be inspected.

[0130] The energy-saving deviation calculation unit is used to calculate the energy-saving deviation value of each device based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state.

[0131] The equipment health risk assessment unit is used to calculate the equipment health risk index of each piece of equipment based on its historical operating records and using a pre-built equipment health risk assessment model.

[0132] The correlation impact assessment unit is used to construct a device correlation relationship model of the system to be inspected, and calculate the correlation impact value of each device based on the device correlation relationship model. The correlation impact value of any device is used to characterize the degree of influence of the energy-saving deviation of the source device on the device. The larger the correlation impact value, the greater the degree of influence. The source device is all devices in the system to be inspected that have an influence path with the device.

[0133] The inspection benefit calculation unit is used to establish an inspection benefit evaluation model and, based on the inspection benefit evaluation model, calculate the inspection benefit index obtained by each device after inspection.

[0134] The inspection strategy generation unit is used to calculate the comprehensive inspection score of each piece of equipment by utilizing the energy-saving deviation value, equipment health risk index, correlation impact value and inspection benefit index of each piece of equipment.

[0135] The inspection strategy generation unit is also used to generate inspection strategies for the system to be inspected based on the comprehensive inspection scores of each device.

[0136] The working process, working details and technical effects of the system provided in this embodiment can be found in the first aspect of the embodiment, and will not be repeated here.

[0137] like Figure 3 As shown, the third aspect of this embodiment provides a device for dynamically generating inspection tasks based on energy-saving operation data. Taking the device as an electronic device as an example, it includes: a memory, a processor, and a transceiver connected in sequence. The memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect of the embodiment.

[0138] For specific examples, the memory may include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out (FIFO) memory, and / or first-in-last-out (FILO) memory, etc.; specifically, the processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented using at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor, also known as the CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state.

[0139] In some embodiments, the processor may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. For example, the processor may not be limited to microprocessors of the STM32F105 series, reduced instruction set computer (RISC) microprocessors, x86 architecture processors, or processors with integrated neural network processing units (NPUs). The transceiver may be, but is not limited to, a Wi-Fi transceiver, a Bluetooth transceiver, a General Packet Radio Service (GPRS) transceiver, a ZigBee transceiver (a low-power LAN protocol based on the IEEE 802.15.4 standard), a 3G transceiver, a 4G transceiver, and / or a 5G transceiver. Furthermore, the device may also include, but is not limited to, a power module, a display screen, and other necessary components.

[0140] The working process, working details and technical effects of the electronic device provided in this embodiment can be found in the first aspect of the embodiment, and will not be repeated here.

[0141] The fourth aspect of this embodiment provides a storage medium that stores instructions containing the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect of the embodiment. That is, the storage medium stores instructions, and when the instructions are run on a computer, the method for dynamically generating inspection tasks based on energy-saving operation data as described in the first aspect of the embodiment is executed.

[0142] The storage medium refers to a carrier for storing data, which may include, but is not limited to, floppy disks, optical disks, hard disks, flash memory, USB flash drives, and / or memory sticks. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0143] The working process, working details and technical effects of the storage medium provided in this embodiment can be found in the first aspect of the embodiment, and will not be repeated here.

[0144] The fifth aspect of this embodiment provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform the dynamic generation method for inspection tasks based on energy-saving operation data as described in the first aspect of this embodiment. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0145] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dynamically generating inspection tasks based on energy-saving operation data, characterized in that, include: Acquire real-time operating data and historical operating records of each device in the system to be inspected; Based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model, the energy-saving deviation value of each device is calculated. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state. Based on the historical operating records of each device and using a pre-built device health risk assessment model, the device health risk index of each device is calculated. A device association model of the system to be inspected is constructed, and the association impact value of each device is calculated based on the device association model. The association impact value of any device is used to characterize the degree of influence of the energy-saving deviation of the source device on the device. The larger the association impact value, the greater the degree of influence. The source device is all devices in the system to be inspected that have an influence path with the device. Establish an inspection benefit evaluation model, and calculate the inspection benefit index obtained by each device after inspection based on the inspection benefit evaluation model. The comprehensive inspection score of each piece of equipment is calculated by using the energy-saving deviation value, equipment health risk index, correlation impact value and inspection benefit index of each piece of equipment. Based on the comprehensive inspection scores of each device, an inspection strategy is generated for the system to be inspected.

2. The method according to claim 1, characterized in that, Based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model, the energy-saving deviation value of each device is calculated, including: For any device, obtain the operating parameters of the device during normal operation as historical benchmark operating data; Based on historical benchmark operating data, a benchmark energy consumption dynamic prediction model for any of the devices is constructed. Based on the real-time operating data of any of the devices and the benchmark energy consumption dynamic prediction model, the theoretical energy consumption of any of the devices in the current operating state is calculated. Collect the actual energy consumption of any of the devices under the current operating state; Based on actual energy consumption, theoretical energy consumption, and the energy-saving deviation evaluation model, the energy-saving deviation value of any of the devices is calculated. Wherein, the energy-saving deviation value of any of the devices is: ； In the formula, The model value of the energy-saving deviation evaluation model represents the energy-saving deviation value of any given device. This represents the actual energy consumption of any of the aforementioned devices. This represents the theoretical energy consumption of any of the aforementioned devices. This indicates the runtime of any given device.

3. The method according to claim 1, characterized in that, The historical operation record of any device includes: the continuous operating time of the device, as well as the number of historical failures, maintenance intervals, alarm frequencies, and fluctuation rates of specified operating parameters within a preset historical time period; Based on the historical operating records of each device and using a pre-built device health risk assessment model, the device health risk index for each device is calculated, including: The continuous operating time, the number of historical faults, the maintenance interval, the alarm frequency, and the fluctuation rate of the specified operating parameter are normalized to obtain the continuous operating time score, the number of historical faults score, the maintenance interval score, the alarm frequency score, and the fluctuation rate of the specified operating parameter score. Based on continuous operating time score, historical failure count score, maintenance interval score, alarm frequency score, and specified operating parameter fluctuation score, and using an equipment health risk assessment model, the equipment health risk index of any of the equipment is calculated. The equipment health risk assessment model is as follows: R_h = w_1·X_r + w_2·X_f + w_3·X_a + w_4·X_v + w_5·X_m; In the formula, R_h is the equipment health risk index of any device, X_r represents the continuous running time score, X_f represents the historical failure count score, X_a represents the alarm frequency score, X_v represents the specified operating parameter fluctuation score, X_m represents the maintenance interval score, and w_1 to w_5 are the weights of each score of any device during the current inspection.

4. The method according to claim 3, characterized in that, The weights of each score are calculated using a gradient update strategy based on prediction error, whereby the gradient update strategy based on prediction error is as follows: Obtain the true health label of any device at the time of the last inspection, as well as the historical health risk index and historical weight of each score at the time of the last inspection, wherein the true health label is 1 or 0; The health risk prediction error is calculated based on the true health label and the device health risk index of any of the devices. Based on the health risk prediction error, the historical weights of each score are updated to obtain the weights of each score for any device during the current inspection.

5. The method according to claim 4, characterized in that, Based on the health risk prediction error, the historical weights of each score are updated to obtain the weights of each score for any device during the current inspection, including: Based on the health risk prediction error, the initial weights of each score of any device during the current inspection are calculated, and the initial weights are normalized to obtain the weights of each score of any device during the current inspection. The initial weights of each score for any device during the current inspection are calculated according to the following formula. w_k,temp^(b) = max{0, w_k^(b-1) + η·e^(b-1)·X_k^(b-1)}; In the formula, w_k,temp^(b) represents the initial weight of the k-th score of any device in the current inspection, w_k^(b-1) represents the historical weight of the k-th score, η represents the preset learning rate, e^(b-1) represents the health risk prediction error, and X_k^(b-1) represents the k-th score of any device in the previous inspection, where k=1,2,...,5.

6. The method according to claim 1, characterized in that, A device relationship model of the system to be inspected is constructed, and the relationship impact value of each device is calculated based on the device relationship model, including: Establish a directed topology graph of the equipment in the system to be inspected, wherein each node in the directed topology graph is used to represent each device, and if there is a directed edge between two nodes, it indicates that there is an influence path between the devices corresponding to the two nodes. Based on the directed topology graph of the devices, a physical topology adjacency matrix is ​​determined. In the physical topology adjacency matrix, the element a_ij represents the physical topology constraint value between device i and device j. a_ij is 1 or 0. When a_ij is 1, it means that there is an influence path between device i and device j. Device i is used to influence device j through energy flow, pipeline connection or control signal. When a_ij is 0, it means that there is no influence path between device i and device j. The device combinations corresponding to the elements with a physical topology constraint value of 1 are selected from the physical topology adjacency matrix, and a candidate device combination set is formed using the selected device combinations. For any candidate device combination in the candidate device combination set, obtain the historical energy saving deviation value sequence of two candidate devices in the candidate device combination within a preset historical statistical period. For any candidate device combination, the affected candidate device in the candidate device combination is taken as the target device, and the other candidate device is taken as the source device of influence. Based on the historical energy-saving deviation value sequence corresponding to each of the two candidate devices, the correlation weight between the two candidate devices in any candidate device combination is calculated, and after all candidate device combinations have been traversed, the correlation weight between the two candidate devices in all candidate device combinations is obtained. Based on the association weights between two candidate devices in all candidate device combinations, an association matrix is ​​constructed, wherein the row subscripts in the association matrix represent the source device, the column subscripts represent the target device, and the element r_qp in the association matrix represents the association weight between the source device p and the target device q. Based on the correlation matrix and the energy-saving deviation values ​​of each influencing source device, the correlation impact value of each device is calculated.

7. The method according to claim 6, characterized in that, Based on the historical energy-saving deviation value sequences corresponding to each of the two candidate devices, the correlation weight between the two candidate devices in any combination of candidate devices is calculated, including: Based on the historical energy-saving deviation value sequences of the two candidate devices, the time-delay Pearson correlation coefficients of the two candidate devices at different time delays within the preset time delay range are calculated, and the maximum value among the absolute values ​​of each time-delay Pearson correlation coefficient is taken as the data correlation strength between the two candidate devices. The initial association weight between the two candidate devices is obtained by multiplying the physical topology constraint values ​​corresponding to the two candidate devices by the data association strength. Based on the directed topology graph of the devices, downstream candidate devices of the source device in any candidate device combination are determined; The initial correlation weights between each downstream candidate device and the source device are obtained. Based on the initial correlation weights between each downstream candidate device and the source device, the initial correlation weights between the two candidate devices in any candidate device combination are normalized to obtain the correlation weights between the two candidate devices in any candidate device combination.

8. The method according to claim 7, characterized in that, The time-delay Pearson correlation coefficients of the two candidate devices at different time delays within the preset time delay range were calculated, including: The time-delay Pearson correlation coefficient of the two candidate devices at any time delay is calculated using the following formula; ； In the formula, Indicates the time delay of the two candidate devices at any given time. The time-delayed Pearson correlation coefficient, This represents the historical energy-saving deviation value of candidate device p at time t in the historical energy-saving deviation value sequence of the two candidate devices. This represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device p. This represents the first value in the historical energy-saving deviation sequence of candidate device q among two candidate devices. Historical energy saving deviation at time _____ The value q represents the arithmetic mean of the historical energy-saving deviation sequence of candidate device q, and T represents the total duration of the preset historical statistical period. ,and This indicates the preset time delay range; Accordingly, based on the correlation matrix and the energy-saving deviation values ​​of each influencing source device, the correlation impact value of each device is calculated, including: For any given device, obtain all influencing source devices corresponding to that device. Based on the correlation matrix, the correlation weight between any device and its corresponding source devices is determined. The correlation weight between any device and its corresponding source devices is multiplied by the energy-saving deviation value of each source device, and the sum of the products is obtained to obtain the correlation impact value of any device.

9. The method according to claim 1, characterized in that, Based on the aforementioned inspection benefit evaluation model, the inspection benefit index corresponding to each piece of equipment after inspection is calculated, including: For any device, calculate the theoretical energy consumption of the device under the current operating state, and collect the actual energy consumption of the device under the current operating state, so as to calculate the energy loss rate of the device per unit time based on the theoretical energy consumption and actual energy consumption of the device. From the historical maintenance database, filter out historical inspection and maintenance records that match the current operating status of any equipment; For any historical inspection and maintenance record, calculate the first average energy loss rate of any device during a preset statistical period before the historical inspection and maintenance record, and the second average energy loss rate during a preset period after the historical inspection and maintenance record. Based on the first average energy loss rate and the second average energy loss rate, the energy efficiency recovery rate corresponding to any historical inspection and maintenance record is calculated, and after all historical inspection and maintenance records have been traversed, the energy efficiency recovery rate corresponding to each historical inspection and maintenance record is obtained. Take the arithmetic mean or median of all energy efficiency recovery rates as the expected energy recovery rate for any of the devices. Obtain the inspection cost and equipment importance level of any of the aforementioned devices; Based on the energy loss rate per unit time, the expected energy-saving recovery rate, the inspection cost, the equipment importance level, and the energy-saving deviation value of any equipment, and using the inspection benefit evaluation model, the inspection benefit index corresponding to any equipment is calculated. The inspection benefit evaluation model is as follows: ； In the formula, This represents the model value of the inspection benefit evaluation model, and also represents the inspection benefit index corresponding to any of the aforementioned devices. This indicates the energy-saving deviation value of any device. This represents the energy loss rate per unit time. Indicates the importance level of the equipment. This indicates the projected energy recovery rate. This represents the inspection cost.

10. A system for dynamically generating inspection tasks based on energy-saving operation data, characterized in that, include: The acquisition unit is used to acquire real-time operating data and historical operating records of each device in the system to be inspected. The energy-saving deviation calculation unit is used to calculate the energy-saving deviation value of each device based on the real-time operating data of each device and the pre-built energy-saving deviation evaluation model. The energy-saving deviation value of any device is used to characterize the degree of deviation between the actual energy consumption and the theoretical energy consumption of that device under the current operating state. The equipment health risk assessment unit is used to calculate the equipment health risk index of each piece of equipment based on the historical operating records of each piece of equipment and using a pre-built equipment health risk assessment model. The correlation impact assessment unit is used to construct the equipment correlation relationship model of the system to be inspected, and calculate the correlation impact value of each equipment based on the equipment correlation relationship model. The correlation impact value of any equipment is used to characterize the degree of influence of the energy saving deviation of the source equipment on the equipment. The larger the correlation impact value, the greater the degree of influence. The source equipment is all equipment in the system to be inspected that has an influence path with the equipment. The inspection benefit calculation unit is used to establish an inspection benefit evaluation model and, based on the inspection benefit evaluation model, calculate the inspection benefit index obtained by each device after inspection. The inspection strategy generation unit is used to calculate the comprehensive inspection score of each piece of equipment by utilizing the energy-saving deviation value, equipment health risk index, correlation impact value and inspection benefit index of each piece of equipment. The inspection strategy generation unit is also used to generate inspection strategies for the system to be inspected based on the comprehensive inspection scores of each device.

Images