Building energy consumption deviation diagnosis method

Abstract

A building energy consumption deviation diagnosis method belongs to the technical field of building energy consumption management control, and mainly comprises the following steps: real-time multi-source data acquisition; generating an energy consumption reference value according to the energy consumption historical statistical value; energy consumption deviation quantification is carried out from two aspects of a statistical dimension and a single equipment dimension, and energy consumption abnormity is identified; performing root cause positioning and visualization on the energy consumption abnormity; and performing closed-loop feedback optimization. According to the method, fine diagnosis of energy consumption deviation can be realized, automatic root cause positioning is carried out, and a reliable decision basis is provided for active energy-saving regulation and control.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of energy consumption management control, in particular to a building energy consumption deviation diagnosis method. BACKGROUND

[0002] The current mainstream technology of building energy consumption deviation diagnosis mainly relies on intelligent analysis modules deployed in the cloud or edge platform. It usually collects real-time data of electricity, water, and cold meters through gateways, and compares them with preset static threshold rules (such as fixed period of use upper limit) or simple statistical models based on historical data. When the real-time energy consumption data exceeds the preset threshold or deviates significantly from the historical statistical interval, the system determines that there is an energy consumption anomaly and triggers an alarm.

[0003] The main defect of the prior art is the roughness of the diagnosis logic and the lack of scene adaptability. Static threshold rules cannot dynamically respond to changes in building operating conditions (such as extreme weather, sudden increase in human traffic due to special events, etc.), which can easily cause false positives or false negatives. While the model based on historical statistics can reflect trends, it is difficult to accurately distinguish between normal fluctuations and real anomalies, and it is also difficult to locate the specific equipment or operating link of the anomaly. This results in the fact that the operation and maintenance personnel can only obtain the surface information that "there is an anomaly", and still need to spend a lot of manpower to investigate the root cause, which makes it difficult to support accurate and efficient energy-saving intervention. Therefore, the core technical problem that needs to be solved at present is: how to achieve fine and automatic root cause diagnosis of energy consumption deviation in a complex and variable building operating environment, and accurately locate to a specific device or time period, so as to provide executable decision basis for active energy-saving regulation. SUMMARY

[0004] In view of the above defects or improvement needs of the prior art, the present disclosure provides a building energy consumption deviation root cause diagnosis method, which realizes fine positioning through energy consumption model benchmarking and hierarchical diagnosis engine, and solves the problem of ambiguous abnormal attribution in the prior art.

[0005] The building energy consumption deviation root cause diagnosis method provided by the present disclosure mainly includes the following steps: S1, real-time multi-source data acquisition; S2, generating energy consumption benchmark value according to energy consumption historical statistical value; S3, quantifying energy consumption deviation from statistical dimension and single device dimension, and identifying energy consumption anomaly; S4, root cause positioning and visualization of energy consumption anomaly.

[0006] Further, the step S1 specifically includes: S11, deploying relevant communication protocol analysis module and data format unified conversion processing module for realizing unified collection and basic management of multi-source energy consumption data at data level; S12, combined with the actual building energy consumption equipment category and attribute, through the physical space and business topology mapping relationship, the energy consumption physical model and business model are constructed; S13, energy consumption data collection setting is carried out, which is used for acquiring data required by building energy consumption deviation diagnosis.

[0007] Further, the step S13 specifically comprises: According to the main attribute and the secondary attribute of the equipment, the main level data and the secondary level data are constructed, and the multi-source attribute data of the equipment is acquired through the subscription data; According to the equipment data business type, the energy consumption data to be collected is set, including one or more of the following: regular water, regular electricity, and regular cooling; According to the data statistical type, the energy consumption data collection is set, including one or more of the following: instantaneous value reading, difference calculation, and indication value; The meter data and the physical model data are matched through the rule matching engine, and the basic setting initialization of the energy consumption physical model and the business model is realized.

[0008] Further, the step S2 specifically comprises: S21, multi-dimensional reference values are generated according to energy consumption historical statistical values, including week-on-week values, AI prediction values, and quota management values; S22, the multi-dimensional reference values are synthesized according to the weight rule to obtain the final reference value; S23, dynamic diagnosis rule configuration, including the setting of the time granularity of energy consumption deviation diagnosis, and the setting of the energy consumption deviation alarm threshold in different time periods.

[0009] Further, in the step S21, the calculation method of the reference value specifically comprises: Week-on-week value: taking the average value of the same day and the same period in the past 4 weeks; AI prediction value: predicting future 24h energy consumption based on LSTM, and the input of LSTM includes temperature + week feature + behavior feature; Quota management value: according to the unit area energy consumption standard x building area.

[0010] Further, the configuration method in the step S23 specifically comprises: Time granularity setting: diagnosis is performed by hour, day, or month; Deviation alarm threshold setting: different deviation alarm thresholds are set for working hours on weekdays, non-working hours, and night hours, and the threshold is set by the percentage of exceeding the reference value.

[0011] Further, the step S3 specifically comprises: Business energy consumption classification statistical dimension: numerical anomaly analysis is performed according to the business unit proportion intensity to identify business energy anomaly; Single device dimension: real-time monitoring of the state of a single device, basic anomaly detection, including: data interruption, zero value anomaly, jump anomaly, continuous over-limit; wherein: The monitoring logic of the numerical interruption is that the data of the continuous N1 periods is 0 or empty; The detection logic of the zero value anomaly is that the active power reading value of the device = 0, but the device state is running; The detection logic of the jump anomaly is that the change rate of the adjacent two points > historical maximum fluctuation rate x k; The monitoring logic of the continuous over-limit is that the continuous N2 periods exceed the reference threshold; Wherein, N1, k, N2 are natural numbers.

[0012] Further, the step S4 specifically comprises: According to the abnormal service unit / device output by step S3, the deviation diagnosis result is used to reversely locate the abnormal device and the energy use anomaly classification and energy use anomaly period, and the complete energy use curve and energy use situation are observed in the form of a chart.

[0013] Further, the method further comprises the following steps: S5, closed-loop feedback optimization: output diagnosis report, assist operation and maintenance personnel to identify energy consumption anomalies, locate abnormal nodes, and find inefficient links in energy use, provide decision support for users, including: adjusting energy use timing strategy and linkage strategy, processing abnormal fault devices.

[0014] Compared with the prior art, the beneficial effects of the present disclosure are: ①realize fine diagnosis of energy consumption deviation; ②automated root cause diagnosis, accurate positioning to specific devices or time periods; ③provide executable decision basis for active energy saving regulation; ④assist operation and maintenance personnel to quickly identify energy consumption anomalies, locate abnormal nodes, and timely replace and process abnormal fault devices; ⑤find inefficient links in energy use through diagnosis, make adjustment measures, which can reduce energy waste and reduce energy cost. BRIEF DESCRIPTION OF DRAWINGS

[0015] The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description of exemplary embodiments of the present disclosure taken in conjunction with the accompanying drawings, in which like reference characters refer to the like parts throughout the several views.

[0016] Figure 1 is a whole flowchart; Figure 2 is a real-time multi-source data acquisition flowchart; Figure 3 is a flowchart of energy consumption reference value generation, diagnosis rule configuration, and diagnosis; Figure 4A flowchart for energy consumption deviation diagnosis result root cause positioning and closed-loop feedback optimization is provided. Figure 5 An example of abnormal period positioning is provided. DETAILED DESCRIPTION

[0017] Preferred embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although preferred embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0018] The present disclosure provides a building energy consumption deviation root cause diagnosis method, which realizes fine positioning through energy consumption model benchmarking and hierarchical diagnosis engine, and solves the problem of ambiguous abnormal attribution in the prior art. In an exemplary embodiment, the flow is as shown in FIG. 1, and the main steps include: Figure 1 Step 1: Real-time multi-source data acquisition The specific process is shown in FIG. 2. Figure 2 Based on the local edge side deployment protocol analysis engine, the matching protocol docking script is uploaded to the cloud, and the adaptive protocol analysis module (supporting Modbus / BACnet / MQTT / Internet of Things standard protocols) is deployed on the edge side, which uniformly converts different manufacturer device data formats for database processing, realizes the unified collection and basic management of energy consumption multi-data sources at the data level, and realizes the unified collection and basic management of energy consumption multi-data sources at the data level. Combined with the actual energy consumption equipment categories and attributes of the building, the energy consumption physical model and the business model are constructed through the physical space and business topology mapping relationship; Relying on the main attributes and secondary attributes of the equipment, the main data and secondary data are constructed, and the multi-source attribute data of the equipment is collected by subscribing to the data; Relying on the equipment data business type for configuration: divided into regular water consumption, regular electricity consumption, regular cooling, energy consumption diagnosis, etc. Relying on the data statistical type for configuration: divided into instantaneous value reading, difference calculation and analog value; Through the rule matching engine, the meter data and the physical model data are matched to realize the basic configuration initialization of the energy consumption physical model and the business model.

[0019] Step 2: Energy consumption benchmark value generation The specific process of steps 2-5 is shown in FIGS. 3-6. Figure 3 Figure 4

[0020] (1) According to the energy consumption historical statistical value, the week-over-year value, the AI prediction value and the quota management value are generated ​​​​The final benchmark value is synthesized according to the historical week same period value, the AI prediction value and the quota management value three benchmark values and the weight rule. The historical week same period value takes the average value of the same day and time period in the past 4 weeks, the AI prediction value is based on the LSTM to predict the energy consumption in the next 24 hours (input: temperature + week feature + behavior feature), and the quota management value is the unit area energy consumption standard multiplied by the building area.

[0021] (2) Custom configuration of diagnosis rules, including diagnosis time dimension and numerical percentage dimension, for comparison and analysis by the diagnosis engine.

[0022] In this embodiment, the diagnosis rules are dynamically configured, and the deviation alarm threshold is configured according to the time dimension. The time dimension supports diagnosis by hour / day / month granularity; the deviation alarm threshold is set as a segmented percentage threshold, and different deviation alarm thresholds are set for different behavior times such as working hours on weekdays, non-working days and night hours, for example: 10% alarm above the benchmark value during 8:00-18:00 on weekdays, and 30% alarm above the benchmark value during night hours.

[0023] Step three: deviation quantification and decomposition In this embodiment, deviation quantification is performed in two aspects of energy consumption statistical dimension and single device dimension: abnormal business energy consumption is identified according to the business energy consumption classification statistical dimension; and abnormal states of single devices are monitored in real time, including: data collection is 0, abnormal jump and other abnormal conditions.

[0024] (1) The statistical dimension performs numerical anomaly analysis according to the business unit proportion intensity.

[0025] (2) Single device basic anomaly detection includes four types of data interruption, zero value anomaly, jump anomaly and continuous overrun. Among them: The monitoring logic of numerical interruption is that the data of the last three periods is 0 or empty; the detection logic of zero value anomaly is that the active power reading value of the device is 0 but the device state is running; the detection logic of jump anomaly is that the change rate of adjacent two points is greater than the historical maximum fluctuation rate multiplied by 3; the monitoring logic of continuous overrun is that the last six periods exceed the benchmark threshold.

[0026] Step four: root cause positioning and visualization The reverse positioning engine positions the abnormal devices and energy consumption abnormality classification and energy consumption abnormality period according to the abnormal business unit / device output in step three through the diagnosis deviation result, and observes the complete energy consumption curve and energy consumption situation in the form of a chart.

[0027] Step five: closed-loop feedback optimization The diagnosis report is output to assist the operation and maintenance personnel to quickly identify energy consumption abnormalities, locate abnormal nodes and provide decision support for users, including: adjusting energy consumption timing strategy and linkage strategy, timely replacing abnormal fault devices, etc.

[0028] In this embodiment, the low efficiency link in energy use is found through diagnosis, adjustment measures can reduce energy waste, thereby reducing energy cost; and find the problems of the equipment in operation, prolong the service life of the equipment through operation and maintenance adjustment, and reduce the maintenance cost.

[0029] Application examples Mainly include the following steps: 1. Device source data unified access; 2. Initialization of related device data information; 3. Device attribute data subscription; 4. Create energy consumption physical model; 5. Build energy consumption physical topology structure; 6. Build energy consumption business topology structure; 7. Configure diagnosis rules; 8. Energy consumption anomaly diagnosis analysis result; 9. Abnormal period positioning, as shown in the accompanying Figure 5 ; 10. Comprehensive diagnosis report.

[0030] The technical solution described above is only an exemplary embodiment of the present application, and for those skilled in the art, on the basis of the application disclosed application method and principle, various types of improvements or modifications can be easily made, and are not limited to the method described in the above embodiment of the present application, therefore the above described method is only preferred, and does not have the meaning of limitation.

Claims

1. A building construction energy consumption deviation diagnosis method, characterized in that, The method comprises the following steps: S1, real-time multi-source data acquisition; S2, generating energy consumption benchmark value according to energy consumption historical statistical value; S3, energy consumption deviation quantification from statistical dimension and single device dimension to identify energy consumption anomaly; S4, root cause positioning and visualization of energy consumption anomaly.

2. The method of claim 1, wherein, The step S1 specifically comprises: S11, deploying relevant communication protocol analysis module and data format unified conversion processing module to realize unified collection and basic management of multi-source energy consumption data at data layer; S12, combining building actual energy consumption device categories and attributes, constructing energy consumption physical model and business model through physical space and business topology mapping relationship; S13, energy consumption data collection setting for obtaining data required for building energy consumption deviation diagnosis.

3. The method of claim 2, wherein, The step S13 specifically comprises: According to device main attribute and secondary attribute, constructing primary data and secondary data to obtain device multi-source attribute data through subscription data; According to device data business type, setting energy consumption data to be collected, including one or more of: regular water, regular electricity, and regular cooling; According to data statistical type, setting energy consumption data collection, including one or more of: instantaneous value reading, difference calculation, and indication value; Through rule matching engine, meter data and physical model data are matched to realize basic setting initialization of energy consumption physical model and business model.

4. The method of claim 1, wherein, The step S2 specifically comprises: S21, generating multi-dimensional benchmark value according to energy consumption historical statistical value, including: week same period value, AI prediction value, and quota management value; S22, synthesizing multi-dimensional benchmark value according to weight rule to obtain final benchmark value; S23, dynamic diagnosis rule configuration, including: setting of time granularity of energy consumption deviation diagnosis, and setting of energy consumption deviation alarm threshold in different time periods.

5. The method of claim 4, wherein, In the step S21, the calculation method of benchmark value specifically comprises: Week same period value: taking average value of same day and same period in the past 4 weeks; AI prediction value: predicting future 24h energy consumption based on LSTM, and the input of LSTM includes: temperature + week feature + behavior feature; Quota management value: unit area energy consumption standard × building area.

6. The method of claim 4, wherein, The configuration method in the step S23 specifically comprises: Setting of time granularity: diagnosing by hour, day, or month; Setting of deviation alarm threshold: setting different deviation alarm thresholds for working day working period, non-working day period, and night period, and the threshold is set by percentage of exceeding benchmark value.

7. The method of claim 1, wherein, The step S3 specifically comprises: Business energy consumption classification statistical dimension: numerical anomaly analysis according to business unit proportion intensity to identify business energy anomaly; Single device dimension: real-time monitoring of single device state for basic anomaly detection, including: data interruption, zero value anomaly, jump anomaly, and continuous overrun; wherein: Monitoring logic of numerical interruption: N1 consecutive period data is 0 or empty; Detection logic of zero value anomaly: device active power reading value = 0, but device state is running; Detection logic of jump anomaly: adjacent two point change rate > historical maximum fluctuation rate × k; Monitoring logic of continuous overrun: N2 consecutive periods exceed benchmark threshold. Wherein, N1, k, N2 are natural numbers.

8. The method of claim 1, wherein, The step S4 specifically comprises: According to the abnormal service unit / equipment output by the step S3, through the deviation diagnosis result, the abnormal equipment is reversely located and the energy abnormality is classified and the energy abnormality period, the complete energy curve and the energy situation are observed in the form of chart.

9. The method of any one of claims 1-8, wherein, Further comprising the following steps: S5, closed-loop feedback optimization: outputting a diagnosis report, assisting operation and maintenance personnel in identifying energy consumption abnormalities, locating abnormal nodes, and finding inefficient links in energy use, to provide decision support for users, including: adjusting energy timing strategy and linkage strategy, and processing abnormal fault equipment.

Images