Multi-dimensional evaluation system and method for integrated energy operation and maintenance

Abstract

The application discloses a multi-dimensional evaluation system and method for comprehensive energy operation and maintenance, and particularly relates to the field of comprehensive energy operation and maintenance, and comprises the following steps: extracting a measurement value change, a state change and a prediction change formed by a target operation and maintenance object in a current evaluation period, and writing the changes corresponding to the occurrence time into a same time axis to form an object sequence of the target operation and maintenance object; and the measurement value change, the state change, the prediction change and the evolution process of the associated object of the same operation and maintenance object are continuously associated according to time and are determined in stages, so as to solve the problem that it is difficult to accurately identify whether the same operation and maintenance object has evolved from a local abnormality into a healthy deterioration or a fault formation state requiring intervention in the comprehensive energy operation and maintenance process.

Description

Technical Field

[0001] This invention relates to the field of integrated energy operation and maintenance technology, and more specifically, to a multi-dimensional evaluation system and method for integrated energy operation and maintenance. Background Technology

[0002] In the integrated energy operation and maintenance of industrial parks, public buildings and industrial sites, existing technologies mainly focus on the centralized acquisition of decentralized operation information, early detection of anomalies and provision of a basis for subsequent handling. In engineering, the data of sensors, instruments and devices in the power distribution, heating, cooling, lighting, fire protection, water supply and drainage systems are usually connected to a unified platform. Then, combined with over-limit alarms, SOE event records, power flow analysis and fault prediction and health management results based on historical data, the equipment status and system operation are judged. However, under continuous operation conditions where multiple energy sources such as wind, solar, thermal, cooling, and electricity participate in energy supply and consumption simultaneously, energy coupling relationships exist between subsystems, field access protocols are inconsistent, sampling rhythms differ, and edge node processing resources are limited, even if the same maintenance object has continuously exhibited abnormal fluctuations, accompanied by state shifts of related objects or changes in energy distribution relationships, the system side often still presents this information as several isolated alarms, single event records, or local prediction deviations. Therefore, it is impossible to confirm whether the object is currently in a short-term disturbance, continuous degradation, or fault formation stage, and it is also difficult to give a stable and consistent health status conclusion, resulting in late intervention or improper handling. Because the existing processing methods have not yet unified the judgment of heterogeneous monitoring data, alarm events, prediction results, and multi-energy system operation relationships into the continuous evolution process of the same maintenance object; Therefore, how to correlate and process multi-source operation information, alarm information, event information and prediction information around the same operation and maintenance object during the integrated energy operation and maintenance process, so as to accurately identify whether the object has evolved from a local anomaly to a state of health deterioration or failure that requires intervention, has become an urgent problem to be solved. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-dimensional evaluation system and method for integrated energy operation and maintenance. This system continuously correlates and determines the changes in measured values, state changes, predicted changes, and evolution processes of related objects of the same operation and maintenance object over time. This addresses the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a multi-dimensional evaluation method for integrated energy operation and maintenance, comprising: S1. Extract the measured changes, status changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same timeline according to the occurrence time of each change to form the object sequence of the target operation and maintenance object. S2. Extract the state change time and predicted change time from the object sequence and arrange them in chronological order to form a boundary time sequence. Then, use the adjacent times in the boundary time sequence as boundaries to continuously divide the measured value change to obtain candidate segments arranged in chronological order. S3. Calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. Retain the candidate segments that satisfy the corresponding relationship to obtain the abnormal evolution segment of the target operation and maintenance object. S4. Based on the energy exchange relationship, determine the associated objects directly connected to the target maintenance object, and calculate the change direction between the first time measurement and the last time measurement of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. S5. First, check whether the abnormal evolution segment has caused the related objects to change synchronously based on the object evolution record. If it has, determine that the target operation and maintenance object is in a fault formation state. If it has not, determine whether the target operation and maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and the change direction is consistent, and output multi-dimensional evaluation results.

[0005] In a preferred embodiment, S1 includes: S1-1. Extract measurement records, status records, and prediction records from the unified access records corresponding to the target operation and maintenance object. Then, merge the measurement records by measurement points and arrange them continuously by time. Merge the status records by status items and arrange them continuously by time. Merge the prediction records by prediction items and arrange them continuously by time to obtain the original sequence of the object. S1-2. Calculate the difference between two adjacent measurement records in the original sequence of the object and generate a measurement change when the difference is not zero. Compare the states before and after two adjacent state records and generate a state change when the states before and after are different. Compare the prediction results before and after two adjacent prediction records and generate a prediction change when the prediction results before and after are different. Obtain the occurrence time, the value before the change and the value after the change corresponding to each change. S1-3. Write the measured changes, state changes and predicted changes that occur at the same time or consecutively before and after into the same time axis, and keep the previous value of the change at the same time position connected with the subsequent value of the change at the previous position to form an object sequence of the target operation and maintenance object.

[0006] In a preferred embodiment, S2 includes: S2-1. Extract all state change moments and predicted change moments from the object sequence, sort them in chronological order and remove duplicate moments, and take the first moment and the last moment after sorting as the starting and ending boundary points, respectively, to obtain the boundary moment sequence. S2-2. Use two adjacent boundary times in the boundary time sequence to form a segmentation interval, and write the changes of each measured value in the object sequence into the corresponding segmentation interval according to the time of occurrence to obtain the interval measurement change set corresponding to each segmentation interval; S2-3. Arrange the measured value changes in each interval consecutively according to their order of occurrence, and independently retain the measured value change sets of two adjacent segmented intervals that are connected end to end, forming candidate segments arranged according to their order of occurrence.

[0007] In a preferred embodiment, S3 includes: S3-1. Extract the first time measurement and the last time measurement for each candidate segment, and calculate the difference between the last time measurement and the first time measurement. Determine the direction of change of the corresponding candidate segment based on whether the difference is positive or negative. S3-2. Extract the state changes for each candidate segment, compare the occurrence time of each state change with the start and end times of the corresponding candidate segment, and retain the state changes whose occurrence time is within the corresponding candidate segment to obtain the state change record of the corresponding candidate segment.

[0008] In a preferred embodiment, S3 further includes: S3-3. Extract the predicted changes for each candidate segment and compare the magnitude of the changed value with the original value. When the changed value is greater than the original value and the change direction of the corresponding candidate segment is increasing, or when the changed value is less than the original value and the change direction of the corresponding candidate segment is decreasing, determine the corresponding predicted change as a consistent predicted change and obtain the consistent predicted change record of the corresponding candidate segment. S3-4. For each candidate segment, check the state change record and the consistent prediction change record respectively. If both the state change record and the consistent prediction change record exist, retain the candidate segment to obtain the abnormal evolution segment.

[0009] In a preferred embodiment, S4 includes: S4-1. For each abnormal evolution segment, read the energy exchange record corresponding to the target operation and maintenance object in the abnormal evolution segment, and extract the associated objects according to the rule that one end of the energy exchange record is the target operation and maintenance object and the other end of the exchange does not pass through the intermediate object, so as to obtain the associated object set corresponding to each abnormal evolution segment. S4-2. For each associated object in each associated object set, extract the first and last time measurements of the associated object in the corresponding abnormal evolution segment, calculate the object difference by subtracting the first time measurement from the last time measurement, and determine the direction of change of the associated object based on whether the object difference is positive or negative. At the same time, extract the moment when the associated object first shows a change in measurement in the corresponding abnormal evolution segment to obtain the associated object change record.

[0010] In a preferred embodiment, S4 further includes: S4-3. For each related object change record, when the exchange direction is from the target maintenance object to the related object, compare the time when the measured value of the related object first changes with the start time of the abnormal evolution segment, and compare the change direction of the related object with the change direction corresponding to the abnormal evolution segment. If the time when the measured value first changes is later than the start time and the two change directions are consistent, retain the related object; otherwise, delete the related object. When the exchange direction changes from the associated object to the target maintenance object, the time when the measured value of the associated object first changes is compared with the start time of the abnormal evolution segment, and the change direction of the associated object is compared with the change direction corresponding to the abnormal evolution segment. If the time when the measured value first changes is earlier than the start time and the two change directions are consistent, the associated object is retained; otherwise, the associated object is deleted.

[0011] In a preferred embodiment, S4 further includes: S4-4. Sort the retained related objects in ascending order according to the time of the first change in measurement value. Compare the two adjacent related objects after sorting. If the time of the first change in measurement value of the later related object is later than that of the previous related object and the exchange direction is the same or the beginning and end are connected, write the later related object into the current record. Otherwise, start a new record. If the change direction of the later related object is opposite to that of the previous related object, delete the later related object and keep the record of the previous related object unchanged. Otherwise, retain the later related object. S4-5. Write the associated objects in each record into the object evolution record of the corresponding abnormal evolution section in the order of writing, and write the exchange direction, change direction and writing order of each associated object into the object evolution record, so as to obtain the object evolution record of each abnormal evolution section.

[0012] In a preferred embodiment, S5 includes: S5-1. Read the object evolution records and change directions of each abnormal evolution segment corresponding to the target operation and maintenance object in chronological order. If there are related objects in the object evolution records and the change direction of each related object is consistent with the change direction of the corresponding abnormal evolution segment, the corresponding abnormal evolution segment is determined as the driving segment; otherwise, the corresponding abnormal evolution segment is determined as the undriven segment. S5-2. For each driven segment, check the exchange direction of adjacent related objects according to the writing order in the object evolution record. When the exchange direction of the next related object follows the exchange direction of the previous related object, determine that the target maintenance object is in a fault formation state. Otherwise, change the driven segment to an undriven segment. For each undriven segment, extract the previous abnormal evolution segment and compare the change direction of the current abnormal evolution segment with that of the previous abnormal evolution segment. When the previous abnormal evolution segment exists and the change direction of the two abnormal evolution segments is consistent, determine that the target maintenance object is in a continuous deterioration state. Otherwise, determine that the target maintenance object is in a short-term disturbance state. S5-3. Output the judgment result of the target operation and maintenance object as the fault formation state, continuous deterioration state or short-term disturbance state, and obtain multi-dimensional evaluation results.

[0013] In a preferred embodiment, a multi-dimensional evaluation system for integrated energy operation and maintenance includes: The time-series construction module is used to extract the measured changes, state changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same time axis according to the occurrence time of each change to form the object sequence of the target operation and maintenance object; The segmentation module is used to continuously divide the measured value changes by using the state change time and the predicted change time in the object sequence as the dividing points, and obtain candidate segments arranged in chronological order of occurrence. The anomaly identification module is used to calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and to check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. The candidate segments that satisfy the corresponding relationship are retained to obtain the anomaly evolution segment of the target operation and maintenance object. The correlation assessment module determines the associated objects directly connected to the target maintenance object based on the energy exchange relationship, and calculates the change direction between the first and last time measurements of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. The status output module first checks whether the abnormal evolution segment has caused related objects to change synchronously based on the object evolution record. If it has, it determines that the target maintenance object is in a fault formation state. If it has not, it determines whether the target maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and changes in the same direction, and outputs multi-dimensional evaluation results.

[0014] The technical effects and advantages of this invention are as follows: 1. By writing measured changes, state changes, and predicted changes into the same time axis and continuously associating them around the same maintenance object, and then combining them with the screening of abnormal evolution segments, isolated alarms, single events, and local prediction deviations can be converged into a unified judgment process, thereby making the identification of short-term disturbances, continuous degradation, and fault formation states more accurate. 2. By extracting the time of state change and the time of predicted change to form a boundary time sequence, and continuously segmenting the measured value change accordingly, the change content in different time periods can be separated and processed, reducing the impact of cross-time period mixing on the segment judgment, and making the abnormal segment extraction results relatively clear. 3. By calculating the direction of change between the measured values ​​at the first and last moments of the candidate segment, and retaining only the candidate segments that simultaneously have state change records and consistent prediction change records, general fluctuations can be distinguished from abnormal segments with evolutionary significance, thus suppressing misjudgments and omissions. 4. By extracting the associated objects directly connected to the target maintenance object based on the energy exchange relationship, and combining the exchange direction, change direction and the time of the first change in the measured value to form an object evolution record, the propagation relationship of anomalies among associated objects can be reflected, providing a basis for cross-object status judgment. 5. By first checking whether the abnormal evolution section causes synchronous changes in related objects, and then distinguishing the exchange direction and the direction relationship between the preceding and following abnormal evolution sections, the fault formation state, continuous deterioration state and short-term disturbance state can be classified and judged in layers, making the determination of subsequent operation and maintenance intervention timing and disposal objects more targeted. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method steps of the present invention.

[0016] Figure 2 This is a schematic diagram of the system modules of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1-2 The multi-dimensional evaluation method for integrated energy operation and maintenance of the present invention includes: S1. Extract the measured changes, status changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same timeline according to the occurrence time of each change to form the object sequence of the target operation and maintenance object. In this implementation, S1 first organizes the actual changes in measured values, state transitions, and changes in predicted results that occur within the current evaluation period around the target maintenance object, and then puts these changes on the same timeline so that the change process of the same maintenance object within the evaluation period can be continuously represented; the subsequent processing of candidate segments, abnormal evolution segments, and object evolution records are all based on the changes on this timeline. The implementation process includes the following steps: First, records whose object identifiers match the target maintenance object are filtered from the unified access records, and then categorized into measurement records, status records, and prediction records based on their content. Measurement records represent the measurement results of the same measurement point at different times; status records represent the status values ​​of the same status item at different times; and prediction records represent the prediction results of the same prediction item at different times. After categorization, measurement records are merged by measurement point, with records under the same measurement point arranged in ascending order of occurrence time, and records with the same occurrence time arranged in order of access sequence. Status records are then merged by status item, with records under the same status item arranged in ascending order of occurrence time. Sort the records in ascending order by occurrence time, and sort those with the same occurrence time by access order. Merge the prediction records by prediction item, and sort the prediction records under the same prediction item by occurrence time, and sort those with the same occurrence time by access order, thus obtaining the original sequence of objects. Taking a heating unit in a comprehensive energy station as an example, the measurement records can be taken from the supply water temperature measurement point, return water temperature measurement point, and input power measurement point. The status records can be taken from the start / stop status item and the operation mode status item. The prediction records can be taken from the heat load prediction item and the fault prediction item. After the above records are uniformly accessed, they can be merged and sorted according to the corresponding measurement point, status item, and prediction item. Subsequently, the locations where changes actually occurred are located within the original sequence of objects. For two consecutive measurement records at the same measurement point, the difference between the value of the latter record and the value of the former record is calculated. If the difference is not equal to zero, a measurement change is generated, and the time when the latter measurement record occurs is recorded as the time of occurrence of the measurement change. The value of the former measurement record is recorded as the value before the change, and the value of the latter measurement record is recorded as the value after the change. If the difference is equal to zero, no measurement change is generated. For two consecutive state records under the same state item, the state value of the latter state record is compared with the state value of the former state record. If the two are different, a state change is generated, and the time when the latter state record occurs is recorded as the time of occurrence of the state change. At any given moment, the state value of the previous state record is recorded as the value before the change, and the state value of the next state record is recorded as the value after the change. If the two are the same, no state change is generated. For two adjacent prediction records under the same prediction item, the prediction result of the next prediction record is compared with the prediction result of the previous prediction record. If the two are different, a prediction change is generated, and the occurrence time of the next prediction record is recorded as the occurrence time of the prediction change. The prediction result of the previous prediction record is recorded as the value before the change, and the prediction result of the next prediction record is recorded as the value after the change. If the two are the same, no prediction change is generated. Here, adjacent comparisons are limited to the same measurement point, the same state item, or the same prediction item, and cross-comparisons are not performed between different record items. After obtaining various changes, the measured changes, state changes, and predicted changes are written into the same timeline to form an object sequence. During writing, all changes are first arranged in ascending order of their occurrence time. When different types of changes occur at the same time, they are written in the order of measured changes, state changes, and predicted changes. For changes occurring at different times but consecutively, the connection is based on whether the value before the change and the value after the change at the previous position can be connected. Specifically: if the value before the change of the current change is the same as the value after the change of the previous position, and both originate from the same measurement point, the same state item, or the same prediction item, the current change is written into the next position. When the value before a change is different from the value after a change at the previous position, it is still written according to the time of occurrence, but the connection relationship between the two changes is recorded again from that position. For example, if a measurement point changes from 35 to 40 at 10:00 and from 40 to 45 at 10:05, then the value before the change of the later measurement is the same as the value after the change of the previous measurement, and the two are processed as consecutive. If a status item changes from running to stopped at 10:07, then the status change is inserted into the corresponding position on the time axis according to the time of occurrence, and its own value before and after the change is retained for direct use when extracting the time of status change and predicting the time of change later. After the above processing, the measured changes, status changes and predicted changes of the target operation and maintenance object in the current evaluation period are organized on the same time axis. Each change corresponds to a clear occurrence time, the value before the change and the value after the change. There is also a clear sequential relationship between adjacent positions. Therefore, the boundary time sequence can be directly extracted, candidate segments can be segmented and abnormal evolution segments can be identified. In practical applications: When the target maintenance object is a heating unit in an integrated energy station, the access records of the corresponding temperature measurement point, power measurement point, start-up and shutdown status item, operation mode status item, heat load prediction item, and fault prediction item of the unit can be extracted first. Then, adjacent comparisons are performed within the same measurement point, the same status item, and the same prediction item to generate measured value changes, status changes, and prediction changes. Finally, these changes are written into the same time axis according to the time of occurrence, and the connection between the previous change and the subsequent change is maintained when the value after the change of the previous change is consistent with the value before the change of the next change, thereby forming the object sequence of the heating unit in the current evaluation cycle.

[0019] S2. Extract the state change time and predicted change time from the object sequence and arrange them in chronological order to form a boundary time sequence. Then, use the adjacent times in the boundary time sequence as boundaries to continuously divide the measured value change to obtain candidate segments arranged in chronological order. In this embodiment, the function of S2 is to divide the measured changes in the object sequence according to comparable time boundaries, so that subsequent determinations can be made about whether there is abnormal evolution around each continuous measured change. Here, the positions that can change the segment boundaries are first identified in the object sequence, and then the measured changes on the same time axis are assigned to their respective time ranges using these positions. Finally, the measured changes in each time range are organized into candidate segments that can be directly determined later. After this processing, the measured changes no longer exist as a mixture of continuous time segments, but are assigned to a group of local segments with clear connections between them. The implementation process includes the following steps: First, read the state changes and predicted changes one by one from the object sequence, extract the occurrence time corresponding to each state change and each predicted change, and sort all the extracted times in ascending order. When two or more times are the same, only one time is kept, and the remaining duplicate times are deleted. After deduplication, the first time in the sorted result is determined as the starting boundary point, and the last time in the sorted result is determined as the ending boundary point. The remaining times in between are retained in their original sorting order, thus forming the boundary time sequence. Here, the state change time and the predicted change time are directly taken as... The occurrence time at the corresponding position in the object sequence is not recalculated. If both state change and predicted change exist at the same time in the object sequence, then that time is retained only once in the boundary time sequence. Taking the object sequence of a certain heating unit as an example, if the state change times are 10:05 and 10:20, and the predicted change times are 10:05, 10:15, and 10:25, then after sorting and deduplication, we get 10:05, 10:15, 10:20, and 10:25. Among them, 10:05 is used as the starting boundary point, 10:25 is used as the ending boundary point, and the two intermediate times are retained as subsequent dividing boundaries. After obtaining the boundary time sequence, a segmentation interval is formed by two adjacent boundary times in the sequence. Specifically, the preceding boundary time is taken as the start time of the segmentation interval, and the following boundary time is taken as the end time. Then, all measured value changes in the object sequence are read, and the occurrence time of each measured value change is compared with the start and end times of each segmentation interval. If the occurrence time of a measured value change is later than or equal to the start time of the corresponding segmentation interval, and earlier than or equal to the end time of the corresponding segmentation interval, the measured value change is written into the segmentation interval. If the occurrence time of a measured value change does not fall within the current segmentation interval... If an interval is divided, it continues to be compared with the next interval until all intervals are written. In this way, each interval obtains a corresponding set of interval measurement changes, and intervals that do not contain measurement changes form an empty set. Taking the heating unit mentioned above as an example, if the boundary time sequence is 10:05, 10:15, 10:20, and 10:25, then three intervals are formed: 10:05 to 10:15, 10:15 to 10:20, and 10:20 to 10:25. If a measurement change in the object sequence occurs at 10:17, then that measurement change is written into the interval measurement change set corresponding to 10:15 to 10:20. After writing the measured value changes, the interval measured value change sets corresponding to each segmented interval are processed separately. Specifically, the measured value changes in each interval measured value change set are arranged in ascending order of occurrence time, and those occurring at the same time are arranged in the original writing order in the object sequence, so that the measured value changes within the same interval remain continuous. Then, the interval measured value change sets are checked according to the time sequence of the segmented intervals. When an interval measured value change set is empty, no candidate segment is generated; when an interval measured value change set is not empty, it is retained as an independent candidate segment. For two temporally adjacent segmented intervals, even if the changed value of the last measured value change in the previous interval is contiguous with the changed value of the first measured value change in the next interval, they are still separated. Do not retain two candidate segments, and do not merge across intervals; after this processing, each candidate segment corresponds to the measured value change content within only one segmented interval, and its start and end range is directly limited by the boundary time sequence, which facilitates subsequent checks on whether state changes and predicted changes exist simultaneously within the candidate segment; taking the aforementioned heating unit as an example, if there are two measured value changes in the interval from 10:05 to 10:15, no measured value changes in the interval from 10:15 to 10:20, and one measured value change in the interval from 10:20 to 10:25, then two candidate segments are formed, where the first candidate segment corresponds to the two consecutive measured value changes in the previous interval, and the second candidate segment corresponds to one measured value change in the next interval, and the empty interval in between does not form a candidate segment; After the above processing, the state change time and the predicted change time are organized into a boundary time sequence, and the measured value changes are allocated to the adjacent segmentation intervals according to the time of occurrence, and further formed into candidate segments arranged in chronological order of occurrence. In this way, when checking the state change and predicted change in each candidate segment, the scope of inspection has been limited to a clear time boundary, and the measured value changes in different time periods will not be mixed together. Therefore, the identification of abnormal evolution segments can be based on clear segments. In practical applications: When the target maintenance object is the heating unit in the integrated energy station, the start-up and shutdown status change time, the operation mode status change time, the heat load prediction change time, and the fault prediction change time can be extracted from the object sequence to form a boundary time sequence; then, each segmentation interval is formed by two adjacent boundary times, and the changes in the supply water temperature measurement value, the return water temperature measurement value, and the input power measurement value are written into the corresponding segmentation interval according to the occurrence time; then, the measurement changes in each segmentation interval are arranged continuously in time, and the non-empty segmentation intervals are retained as independent candidate segments, thereby obtaining a set of candidate segments arranged in time order for the heating unit in the current evaluation cycle, which can be used for subsequent steps.

[0020] S3. Calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. Retain the candidate segments that satisfy the corresponding relationship to obtain the abnormal evolution segment of the target operation and maintenance object. In this embodiment, the function of S3 is to screen out segments that truly have abnormal evolutionary characteristics from the candidate segments. Specifically, the overall direction of the measured value change is first calculated for each candidate segment. Then, the state changes and predicted changes falling within the segment are extracted around the candidate segment. Subsequently, it is determined whether the direction of the predicted change is consistent with the change direction of the candidate segment. Finally, only candidate segments that have both state change records and consistent predicted change records are retained, thereby distinguishing candidate segments that only exhibit general fluctuations from candidate segments with evolutionary characteristics. The implementation process includes the following steps: For each candidate segment, the measured values ​​are first read according to the time of occurrence of the changes within that segment. The value before the change corresponding to the earliest measured value change is determined as the first measured value, and the value after the change corresponding to the latest measured value change is determined as the last measured value. After obtaining the first and last measured values, the segment difference is calculated by subtracting the first measured value from the last measured value. When the segment difference is greater than zero, the change direction of the candidate segment is determined to be increasing; when the segment difference is less than zero, the change direction of the candidate segment is determined to be decreasing. Decrease; Here, the initial and final measured values ​​are directly taken from the retained measured changes within the candidate segment, without returning to the original measured value record, thus ensuring that the direction of change is consistent with the formation process of the candidate segment itself; Taking a candidate segment of a heating unit as an example, if the earliest measured change in the candidate segment is a change in water supply temperature from 60 to 62, and the latest measured change is a change in water supply temperature from 64 to 66, then the initial measured value is 60, the final measured value is 66, the segment difference is 6, and the corresponding direction of change is increase; After determining the direction of change, the state changes are extracted for each candidate segment. Specifically, all state changes in the object sequence are first read, and the occurrence time of each state change is obtained. Then, the start and end times of the current candidate segment are read, where the start time is the occurrence time of the first measurement change in the candidate segment, and the end time is the occurrence time of the last measurement change in the candidate segment. Subsequently, the occurrence time of each state change is compared with the start and end times of the candidate segment one by one. If the occurrence time of a state change is later than or equal to the start time and earlier than or equal to the end time, the state change is retained in the current segment. In the state change records corresponding to the candidate segment, if the occurrence time of the state change is earlier than the start time or later than the end time, it will not be written into the current candidate segment. After the above processing, each candidate segment will have a set of state change records that only occur within the time range of the candidate segment. Taking the heating unit as an example, if the start time of a certain candidate segment is 10:15 and the end time is 10:22, and the start / stop state change occurs at 10:18 and the operation mode state change occurs at 10:24, then the state change corresponding to 10:18 will be written into the state change record of the candidate segment, and the state change corresponding to 10:24 will not be written into the candidate segment. Subsequently, predicted changes are extracted for each candidate segment, and it is determined whether the predicted changes are consistent with the change direction of that candidate segment. Specifically, all predicted changes in the object sequence are first read, obtaining the occurrence time, pre-change value, and post-change value for each predicted change. Then, the occurrence time of the predicted change is compared with the start and end times of the current candidate segment. If the occurrence time of the predicted change is later than or equal to the start time and earlier than or equal to the end time, this predicted change is considered an intra-segment predicted change for the current candidate segment and participates in subsequent judgments. For intra-segment predicted changes, the magnitude of the post-change value and the pre-change value are compared. If the post-change value is greater than the pre-change value and the change direction of the current candidate segment is increasing, this predicted change is determined to be a consistent predicted change. If the changed value is less than the original value and the current candidate segment's change direction is decreasing, the predicted change is determined to be a consistent prediction change; otherwise, it is not considered a consistent prediction change. After this processing, each candidate segment corresponds to a set of consistent prediction change records. Here, the determination of consistent prediction changes is directly based on the original value, the changed value, and the change direction of the candidate segment, without introducing any additional judgment factors. For example, if the change direction of a candidate segment is increasing, and there is a fault prediction change within that candidate segment whose original value is general risk and whose changed value is rising risk, then this prediction change can be retained as a consistent prediction change. If another prediction change within the candidate segment changes from rising risk to general risk, then this prediction change is not written into the consistent prediction change record. After extracting the state change records and consistent prediction change records, a retention decision is made for each candidate segment. Specifically, the state change records and consistent prediction change records corresponding to the current candidate segment are checked one by one. If there is at least one state change in the state change record and at least one consistent prediction change in the consistent prediction change record, the candidate segment is retained and written into the abnormal evolution segment. If there is no state change record or no consistent prediction change record, the candidate segment is not retained. Through this process, only when both state switching and segment change occur simultaneously within a segment will the candidate segment be retained. When the predicted change is consistent with the direction, the candidate segment is confirmed as an abnormal evolution segment, thus distinguishing simple numerical fluctuations from segments with evolutionary significance. Taking the aforementioned heating unit as an example, if a candidate segment has both a 10:18 start-up / shutdown state change and a fault prediction change in the same direction as the segment change, then the candidate segment is retained as an abnormal evolution segment. If another candidate segment has continuously increasing measured values ​​but no state change, or has a state change but the predicted change direction is inconsistent with the candidate segment change direction, then the candidate segment does not enter the abnormal evolution segment. After the above processing, each candidate segment has completed the determination of change direction, screening of state change, screening of predicted change, and consistency judgment. Finally, it can screen out the segments that truly have abnormal evolution characteristics from all candidate segments, so that the subsequent construction of associated objects and object evolution records around abnormal evolution segments has clear segment boundaries and segment basis. In practical applications: When the target maintenance object is the heating unit in the integrated energy station, the difference between the first and last measured values ​​of the water supply temperature or input power in the candidate segments formed in chronological order can be calculated to determine whether the candidate segment is in the direction of increase or decrease. Then, the start-stop status changes and operating mode status changes are screened into each candidate segment according to the time of occurrence. Subsequently, the heat load prediction changes and fault prediction changes are screened into each candidate segment according to the time of occurrence, and the prediction result is compared with the direction of change of the candidate segment to determine whether it is an increase or decrease. Only the prediction changes with the same direction are retained. Finally, for the candidate segment that has both status change records and consistent prediction change records, it is identified as an abnormal evolution segment for continued use in subsequent steps.

[0021] S4. Based on the energy exchange relationship, determine the associated objects directly connected to the target maintenance object, and calculate the change direction between the first time measurement and the last time measurement of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. In this embodiment, S4 is used to identify associated objects that have a direct energy exchange relationship with the target maintenance object within the abnormal evolution segment, and further determine whether these associated objects form a unidirectional and sequential evolution process around the target maintenance object, thereby providing an object evolution basis for the subsequent determination of fault formation state, continuous deterioration state and short-term disturbance state; during processing, objects directly connected to the target maintenance object are first screened out from the energy exchange records corresponding to the abnormal evolution segment, and then the change direction and the time of the first measurement change of each associated object within the abnormal evolution segment are calculated respectively. Subsequently, the exchange direction, the start time of the abnormal evolution segment and the change direction of the abnormal evolution segment are combined to perform retention or deletion, and finally the retention results are organized into object evolution records according to the time sequence relationship; The implementation process includes the following steps: First, for each abnormal evolution segment, read the energy exchange records corresponding to the target maintenance object within that segment. Using the target maintenance object as the starting point, examine the objects at both ends of each energy exchange record. If one end is the target maintenance object and the other end is another object, and there is no third object acting as a transitional exchange end between this other object and the target maintenance object within the current abnormal evolution segment, then this other object is identified as an associated object. Here, "direct connection" is determined based on the energy exchange record itself; that is, only records that form an exchange relationship with the target maintenance object are retained. The system excludes objects that require transfer through other objects to reach them; after filtering, all associated objects corresponding to each abnormal evolution segment, along with their respective exchange directions, are written into the associated object set; for example, when the target maintenance object is a heating unit, and the energy exchange record shows that the heating unit has a direct heat exchange relationship with the primary side pipe network and a direct energy supply relationship with the circulating water pump, while the terminal heating unit is only connected to the heating unit through the primary side pipe network, then the primary side pipe network and the circulating water pump are included in the associated object set, while the terminal heating unit is not included in the associated object set; Subsequently, for each associated object in each associated object set, its measured value changes within the corresponding abnormal evolution segment are extracted and arranged in ascending order of occurrence time. The earliest measured value change before the change is taken as the first measured value, and the latest measured value change after the change is taken as the last measured value. The object difference obtained by subtracting the first measured value from the last measured value is calculated. When the object difference is greater than zero, the change direction of the associated object is determined to be increasing; when the object difference is less than zero, the change direction of the associated object is determined to be decreasing. At the same time, the earliest time among all the occurrence times of measured value changes of the associated object within the corresponding abnormal evolution segment is taken as the time when the associated object first shows a measured value change. The associated object identifier, exchange direction, change direction, and the time when the measured value first shows a change are written into the associated object change record. Here, the first measured value and the last measured value are both taken from the measured value changes after the associated object has entered the object sequence, without returning to the original measured value record. After obtaining the change records of the associated objects, a retention judgment is performed based on the exchange direction. Specifically, when the exchange direction changes from the target maintenance object to the associated object, the time when the first measurement change of the associated object occurs is compared with the start time of the corresponding abnormal evolution segment, and the change direction of the associated object is compared with the change direction corresponding to the abnormal evolution segment. If the time when the first measurement change of the associated object occurs is later than the start time of the abnormal evolution segment, and the two change directions are the same, the associated object is retained; otherwise, the associated object is deleted. Correspondingly, when the exchange direction changes from the associated object to the target maintenance object, the time when the first measurement change of the associated object occurs is compared with the start time of the corresponding abnormal evolution segment. The change direction of the associated object is compared with the change direction corresponding to the abnormal evolution segment. If the time when the first measurement change of the associated object occurs is earlier than the start time of the abnormal evolution segment, and the two change directions are the same, then the associated object is retained; otherwise, the associated object is deleted. Here, the start time of the abnormal evolution segment is taken as the time when the first measurement change occurs within the abnormal evolution segment, and the change direction corresponding to the abnormal evolution segment is taken from the judgment result of S3 above. The meaning of the aforementioned retention judgment is that when the target maintenance object propagates outward, the associated object should change in the same direction after the target maintenance object enters the abnormal evolution, while when the associated object propagates inward, the associated object should change in the same direction before the target maintenance object. For the retained associated objects, they are sorted in ascending order according to the time of the first change in measured value, and adjacent associated objects are compared sequentially. If the time of the first change in measured value of the later associated object is later than that of the earlier associated object, and their exchange directions are the same, or if the start or end point of the exchange of the later associated object in the energy exchange record is consecutive with the exchange relationship of the earlier associated object, then the later associated object is continued in the current record; otherwise, a new record is started. After completing the above continuation judgment, the change directions of the later associated object and the earlier associated object are compared again; if the two... If the directions of change are opposite, the latter associated object is deleted while the record containing the former associated object remains unchanged; if the directions of change are the same, the latter associated object is retained. Here, "connection between beginning and end" is determined according to the connection relationship in the energy exchange record. For example, if the former associated object is the direct upstream or downstream object in the exchange relationship corresponding to the latter associated object, the two are considered to be connected. Through this process, associated objects that are connected in the same direction within the same abnormal evolution segment can be retained in the same record, while associated objects that are not connected in time or have opposite directions are removed or separated. After sorting and filtering, the associated objects in each record are written sequentially to the object evolution record of the corresponding abnormal evolution segment in the writing order. The exchange direction, change direction and writing order of each associated object are written synchronously in the object evolution record. The writing order is determined by the position of the associated object in the current record. The first written object is recorded as the first order, the second written object is recorded as the second order, and so on. After writing is completed, each abnormal evolution segment will have one or more object evolution records. The record content can reflect how the associated objects directly related to the target operation and maintenance object change in time sequence, whether the change direction is consistent, and along which exchange direction these changes occur, thus obtaining the object evolution record of each abnormal evolution segment. After the above processing, the associated objects in each abnormal evolution segment are directly connected and filtered, the direction of change is calculated, the occurrence time is compared and the time sequence is sorted out, and finally an object evolution record that can be directly used for subsequent state determination is formed. The object evolution record obtained in this way not only retains the energy exchange relationship between the target operation and maintenance object and the associated objects, but also retains the order and direction of the changes of the associated objects. Therefore, it can distinguish between general accompanying changes and the evolution process formed around the target operation and maintenance object. In practical applications: When the target maintenance object is the heating unit in the integrated energy station, the heat exchange record and energy supply record corresponding to the heating unit in a certain abnormal evolution section can be read first, and objects with an exchange relationship with it, such as the primary side pipeline and circulating water pump, can be screened out as associated objects; then, the object difference and change direction of these associated objects in the abnormal evolution section can be calculated respectively, and the time when the first measured value change occurs can be extracted; then, based on the exchange direction, it can be determined whether these associated objects should change in the same direction before or after the start time of the abnormal evolution section, and objects that do not conform to the time sequence relationship or direction relationship can be deleted; then, the objects are retained in ascending order according to the time when the first measured value change occurs, and objects with the same time sequence and direction are continuously written into the same record; finally, all records form the object evolution record of the abnormal evolution section, which can be used for subsequent steps to continue to determine whether the heating unit has driven the associated objects to change synchronously.

[0022] S5. First, check whether the abnormal evolution segment has caused the related objects to change synchronously based on the object evolution record. If it has, determine that the target operation and maintenance object is in a fault formation state. If it has not, determine whether the target operation and maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and the change direction is consistent, and output multi-dimensional evaluation results. In this embodiment, S5 is used to make a final determination on the current state of the target maintenance object based on the determination of the abnormal evolution segment and the formation of the object evolution record. During processing, the object evolution record and change direction corresponding to each abnormal evolution segment are read in chronological order of occurrence, and it is distinguished which abnormal evolution segments have driven the synchronous change of related objects and which abnormal evolution segments have not yet driven the synchronous change of related objects. Subsequently, for the abnormal evolution segments that have driven the synchronous change of related objects, it is further checked whether the exchange direction between related objects is continuous to determine whether the target maintenance object has entered the fault formation state. For the abnormal evolution segments that have not driven the synchronous change of related objects, it is further combined with the change direction of the previous abnormal evolution segment to determine whether the current abnormal evolution is continuous or short-term. Finally, the multi-dimensional evaluation results corresponding to the fault formation state, continuous deterioration state, or short-term disturbance state are output. The implementation process includes the following steps: First, read all abnormal evolution segments corresponding to the target maintenance object in chronological order of occurrence, and then read the object evolution record and the change direction of each abnormal evolution segment in turn. The order of the abnormal evolution segments is determined by the order of their start times, with those starting earlier being read first and those starting later being read later. For each abnormal evolution segment, first check if there are any associated objects in its object evolution record. If there are no associated objects in the object evolution record, the abnormal evolution segment is directly identified as an unaffected segment. If there are associated objects in the object evolution record, continue to read the change direction corresponding to each associated object one by one, and compare the change direction of each associated object with the change direction of the abnormal evolution segment. The change directions are compared one by one; when the change direction of all related objects is consistent with the change direction of the abnormal evolution segment, the abnormal evolution segment is determined as the driving segment; when at least one related object's change direction is inconsistent with the change direction of the abnormal evolution segment, the abnormal evolution segment is determined as the non-driving segment. Here, "consistent change direction" is uniformly determined by both increasing or both decreasing, and no other directional criteria are introduced. Taking a heating unit as an example, if the change direction of a certain abnormal evolution segment is increasing, and the change direction of the two related objects, the primary side pipe network and the circulating water pump, in its object evolution record is also increasing, then the abnormal evolution segment is determined as the driving segment; if the change direction of any of the related objects is decreasing, then the abnormal evolution segment is determined as the non-driving segment. Subsequently, subsequent determinations are performed on the identified driven and undriven sections. For each driven section, the two adjacent related objects are read sequentially according to the writing order in the object evolution record corresponding to that driven section, and the exchange direction corresponding to these two related objects is extracted. Then, based on the exchange relationship in the energy exchange record, it is checked whether the exchange direction of the subsequent related object follows the exchange direction of the preceding related object. Specifically, when the start or end point of the exchange relationship corresponding to the subsequent related object is the same as the end or start point of the exchange relationship corresponding to the preceding related object, and the writing order of the two in the object evolution record is consecutive, it is determined that the exchange direction of the subsequent related object follows the exchange direction of the preceding related object. If the exchange directions of the adjacent related objects in the driven section satisfy the above-mentioned follow-up relationship, the target maintenance object is determined to be in a fault formation state. If the above-mentioned follow-up relationship is not satisfied, the driven section is reclassified as an undriven section, and no fault formation state determination is made on the driven section. For each undriven section, the following is extracted: The preceding abnormal evolution segment is selected based on its occurrence sequence, being the one preceding the current abnormal evolution segment and being the closest in time to it. The direction of change in the current abnormal evolution segment is then compared with that of the preceding abnormal evolution segment. If a preceding abnormal evolution segment exists and the directions of change in both segments are consistent, the target maintenance object is identified as being in a state of continuous deterioration. If a preceding abnormal evolution segment does not exist or the directions of change in the two segments are inconsistent, the target maintenance object is identified as being in a state of short-term disturbance. Here, the preceding abnormal evolution segment is uniquely determined solely by chronological order and does not involve reselecting from other existing abnormal evolution segments. Taking a heating unit as an example, if the most recent abnormal evolution segment preceding the current one also shows a continuous increase in water supply temperature, the corresponding state of the currently unaffected segment is judged as a state of continuous deterioration. If the preceding abnormal evolution segment shows a decrease, or if there are no other abnormal evolution segments preceding the current one, it is judged as a state of short-term disturbance. After completing the above state determination, the fault formation state, continuous degradation state, or short-term disturbance state are output as the determination result of the target maintenance object under the current abnormal evolution segment. The determination result is written into the multi-dimensional evaluation result in a one-to-one correspondence with the corresponding abnormal evolution segment. When the same target maintenance object has multiple abnormal evolution segments within the current evaluation period, the corresponding determination results are output according to the order of occurrence of each abnormal evolution segment, forming a complete multi-dimensional evaluation result of the target maintenance object within the current evaluation period. When outputting, each multi-dimensional evaluation result includes at least the start time, end time, and determination result of the corresponding abnormal evolution segment. The start time and end time are directly taken from the segment boundary of the corresponding abnormal evolution segment, and the determination result is directly taken from the aforementioned fault formation state, continuous degradation state, or short-term disturbance state. After this processing, the multi-dimensional evaluation result retains both the segment position and the corresponding state determination of the segment, which can be directly used for maintenance prompts, alarm linkage, or further analysis. Through the above processing, the target maintenance object has completed the transformation from object evolution record to state result in each abnormal evolution segment. Among them, the driving segment can further distinguish whether a fault formation state has been formed and continuously propagated along the exchange relationship, while the non-driving segment can distinguish between continuous deterioration state and short-term disturbance state based on the connection between the abnormal evolution before and after. Therefore, abnormal evolution segments of different natures can be assigned to clear state conclusions, avoiding giving a general judgment based on a single abnormal phenomenon. In practical applications: When the target maintenance object is a heating unit in an integrated energy station, the object evolution records and change directions of each abnormal evolution section corresponding to the heating unit can be read in chronological order. If the primary side pipeline and circulating water pump in a certain abnormal evolution section are both written into the object evolution record, and their change direction is consistent with the change direction of the abnormal evolution section, then the abnormal evolution section is first identified as the driving section. Then, check whether the corresponding exchange relationship between the primary side pipeline and the circulating water pump is continuous in the energy exchange record. If it is continuous, then it is determined that... The heating unit is in a fault-forming state. If it is not taken over, the section will be reclassified as an undriven section. For another undriven section, the previous abnormal evolution section will be extracted and the change direction of the two will be compared. If both show an increase, the heating unit will be determined to be in a state of continuous deterioration. If the previous abnormal evolution section does not exist or the direction is different, it will be determined to be a short-term disturbance state. Finally, the above state results will be written into the multi-dimensional evaluation results of the corresponding abnormal evolution sections to form the state output results of the heating unit in the current evaluation period.

[0023] Furthermore, a multi-dimensional assessment system for integrated energy operation and maintenance includes: The time-series construction module is used to extract the measured changes, state changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same time axis according to the occurrence time of each change to form the object sequence of the target operation and maintenance object; The segmentation module is used to continuously divide the measured value changes by using the state change time and the predicted change time in the object sequence as the dividing points, and obtain candidate segments arranged in chronological order of occurrence. The anomaly identification module is used to calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and to check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. The candidate segments that satisfy the corresponding relationship are retained to obtain the anomaly evolution segment of the target operation and maintenance object. The correlation assessment module determines the associated objects directly connected to the target maintenance object based on the energy exchange relationship, and calculates the change direction between the first and last time measurements of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. The status output module first checks whether the abnormal evolution segment has caused related objects to change synchronously based on the object evolution record. If it has, it determines that the target maintenance object is in a fault formation state. If it has not, it determines whether the target maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and changes in the same direction, and outputs multi-dimensional evaluation results.

[0024] Working Principle: This solution revolves around fault prediction and health management during integrated energy operation. First, it organizes the measured changes, state changes, and predicted changes of the target maintenance object within the current assessment period onto the same timeline, forming an object sequence. Then, it extracts the state change and predicted change times from the object sequence as boundaries to continuously segment the measured changes, obtaining candidate segments. Next, it calculates the direction of change of the first and last measured values ​​for each candidate segment and checks whether state changes and predicted changes with the same direction exist simultaneously within the segment, screening out abnormal evolution segments. Based on this, it further considers energy exchange relationships... Identify the related objects directly connected to the target maintenance object, determine the direction, sequence, and connection relationship of these related objects within the abnormal evolution segment, and form an object evolution record; finally, based on the object evolution record, determine whether the abnormal evolution segment has driven the related objects to change synchronously. If it has, further determine whether it has entered a fault formation state; if it has not, further distinguish whether it is continuous degradation or short-term disturbance, thereby outputting multi-dimensional evaluation results; the core of the whole process is that it does not look at a single alarm, a single state switch, or a single prediction result, but puts them on the same change chain for continuous judgment. For example, when performing fault prediction and health management on a heating unit in an integrated energy station, the system first reads the changes in measured values ​​such as the unit's supply water temperature, return water temperature, and power, as well as changes in start-up / shutdown status, operating mode, and heat load or fault prediction. These changes are then arranged into a continuous timeline based on their occurrence time. If, within a certain time period, the unit's measured values ​​continuously change in the same direction, and a status switch occurs, and the prediction results also change in the same direction, then that time period is identified as an abnormal evolution segment. Next, the system examines objects such as the primary side piping network and circulating water pumps that directly exchange energy with the unit. Determine whether these objects also show changes in the same direction before and after the same time period, and whether the sequence of changes between them is continuously transmitted along the energy exchange relationship; if such a transmission relationship has already occurred, it means that the anomaly is no longer limited to the unit itself, but is expanding to related objects, which can be judged as a fault formation state; if it has not yet expanded, but the same-direction anomaly occurs repeatedly in adjacent time periods, it is judged as a continuous deterioration state; if it only occurs once and does not expand outward, it is judged as a short-term disturbance state; in this way, the judgment of integrated energy equipment can be further improved from "whether there is an anomaly" to "what stage the anomaly is in and whether immediate intervention is required".

[0025] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional evaluation method for integrated energy operation and maintenance, characterized in that, include: S1. Extract the measured changes, status changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same timeline according to the occurrence time of each change to form the object sequence of the target operation and maintenance object. S2. Extract the state change time and predicted change time from the object sequence and arrange them in chronological order to form a boundary time sequence. Then, use the adjacent times in the boundary time sequence as boundaries to continuously divide the measured value change to obtain candidate segments arranged in chronological order. S3. Calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. Retain the candidate segments that satisfy the corresponding relationship to obtain the abnormal evolution segment of the target operation and maintenance object. S4. Based on the energy exchange relationship, determine the associated objects directly connected to the target maintenance object, and calculate the change direction between the first time measurement and the last time measurement of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. S5. First, check whether the abnormal evolution segment has caused the related objects to change synchronously based on the object evolution record. If it has, determine that the target operation and maintenance object is in a fault formation state. If it has not, determine whether the target operation and maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and the change direction is consistent, and output multi-dimensional evaluation results.

2. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 1, characterized in that: S1 includes: S1-1. Extract measurement records, status records, and prediction records from the unified access records corresponding to the target operation and maintenance object. Then, merge the measurement records by measurement points and arrange them continuously by time. Merge the status records by status items and arrange them continuously by time. Merge the prediction records by prediction items and arrange them continuously by time to obtain the original sequence of the object. S1-2. Calculate the difference between two adjacent measurement records in the original sequence of the object and generate a measurement change when the difference is not zero. Compare the states before and after two adjacent state records and generate a state change when the states before and after are different. Compare the prediction results before and after two adjacent prediction records and generate a prediction change when the prediction results before and after are different. Obtain the occurrence time, the value before the change and the value after the change corresponding to each change. S1-3. Write the measured changes, state changes and predicted changes that occur at the same time or consecutively before and after into the same time axis, and keep the previous value of the change at the same time position connected with the subsequent value of the change at the previous position to form an object sequence of the target operation and maintenance object.

3. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 2, characterized in that: S2 includes: S2-1. Extract all state change moments and predicted change moments from the object sequence, sort them in chronological order and remove duplicate moments, and take the first moment and the last moment after sorting as the starting and ending boundary points, respectively, to obtain the boundary moment sequence. S2-2. Use two adjacent boundary times in the boundary time sequence to form a segmentation interval, and write the changes of each measured value in the object sequence into the corresponding segmentation interval according to the time of occurrence to obtain the interval measurement change set corresponding to each segmentation interval; S2-3. Arrange the measured value changes in each interval consecutively according to their order of occurrence, and independently retain the measured value change sets of two adjacent segmented intervals that are connected end to end, forming candidate segments arranged according to their order of occurrence.

4. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 3, characterized in that: S3 includes: S3-1. Extract the first time measurement and the last time measurement for each candidate segment, and calculate the difference between the last time measurement and the first time measurement. Determine the direction of change of the corresponding candidate segment based on whether the difference is positive or negative. S3-2. Extract the state changes for each candidate segment, compare the occurrence time of each state change with the start and end times of the corresponding candidate segment, and retain the state changes whose occurrence time is within the corresponding candidate segment to obtain the state change record of the corresponding candidate segment.

5. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 4, characterized in that: S3 further includes: S3-3. Extract the predicted changes for each candidate segment and compare the magnitude of the changed value with the original value. When the changed value is greater than the original value and the change direction of the corresponding candidate segment is increasing, or when the changed value is less than the original value and the change direction of the corresponding candidate segment is decreasing, determine the corresponding predicted change as a consistent predicted change and obtain the consistent predicted change record of the corresponding candidate segment. S3-4. For each candidate segment, check the state change record and the consistent prediction change record respectively. If both the state change record and the consistent prediction change record exist, retain the candidate segment to obtain the abnormal evolution segment.

6. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 5, characterized in that: S4 includes: S4-1. For each abnormal evolution segment, read the energy exchange record corresponding to the target operation and maintenance object in the abnormal evolution segment, and extract the associated objects according to the rule that one end of the energy exchange record is the target operation and maintenance object and the other end of the exchange does not pass through the intermediate object, so as to obtain the associated object set corresponding to each abnormal evolution segment. S4-2. For each associated object in each associated object set, extract the first and last time measurements of the associated object in the corresponding abnormal evolution segment, calculate the object difference by subtracting the first time measurement from the last time measurement, and determine the direction of change of the associated object based on whether the object difference is positive or negative. At the same time, extract the moment when the associated object first shows a change in measurement in the corresponding abnormal evolution segment to obtain the associated object change record.

7. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 6, characterized in that: S4 further includes: S4-3. For each related object change record, when the exchange direction is from the target maintenance object to the related object, compare the time when the measured value of the related object first changes with the start time of the abnormal evolution segment, and compare the change direction of the related object with the change direction corresponding to the abnormal evolution segment. If the time when the measured value first changes is later than the start time and the two change directions are consistent, retain the related object; otherwise, delete the related object. When the exchange direction changes from the associated object to the target maintenance object, the time when the measured value of the associated object first changes is compared with the start time of the abnormal evolution segment, and the change direction of the associated object is compared with the change direction corresponding to the abnormal evolution segment. If the time when the measured value first changes is earlier than the start time and the two change directions are consistent, the associated object is retained; otherwise, the associated object is deleted.

8. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 7, characterized in that: S4 further includes: S4-4. Sort the retained related objects in ascending order according to the time of the first change in measurement value. Compare the two adjacent related objects after sorting. If the time of the first change in measurement value of the later related object is later than that of the previous related object and the exchange direction is the same or the beginning and end are connected, write the later related object into the current record. Otherwise, start a new record. If the change direction of the later related object is opposite to that of the previous related object, delete the later related object and keep the record of the previous related object unchanged. Otherwise, retain the later related object. S4-5. Write the associated objects in each record into the object evolution record of the corresponding abnormal evolution section in the order of writing, and write the exchange direction, change direction and writing order of each associated object into the object evolution record, so as to obtain the object evolution record of each abnormal evolution section.

9. The multi-dimensional evaluation method for integrated energy operation and maintenance according to claim 8, characterized in that: S5 includes: S5-1. Read the object evolution records and change directions of each abnormal evolution segment corresponding to the target operation and maintenance object in chronological order. If there are related objects in the object evolution records and the change direction of each related object is consistent with the change direction of the corresponding abnormal evolution segment, the corresponding abnormal evolution segment is determined as the driving segment; otherwise, the corresponding abnormal evolution segment is determined as the undriven segment. S5-2. For each driven segment, check the exchange direction of adjacent related objects according to the writing order in the object evolution record. When the exchange direction of the next related object follows the exchange direction of the previous related object, determine that the target maintenance object is in a fault formation state. Otherwise, change the driven segment to an undriven segment. For each undriven segment, extract the previous abnormal evolution segment and compare the change direction of the current abnormal evolution segment with that of the previous abnormal evolution segment. When the previous abnormal evolution segment exists and the change direction of the two abnormal evolution segments is consistent, determine that the target maintenance object is in a continuous deterioration state. Otherwise, determine that the target maintenance object is in a short-term disturbance state. S5-3. Output the judgment result of the target operation and maintenance object as the fault formation state, continuous deterioration state or short-term disturbance state, and obtain multi-dimensional evaluation results.

10. A multi-dimensional evaluation system for integrated energy operation and maintenance, characterized in that, include: The time-series construction module is used to extract the measured changes, state changes and predicted changes of the target operation and maintenance object within the current evaluation period, and write them into the same time axis according to the occurrence time of each change to form the object sequence of the target operation and maintenance object; The segmentation module is used to continuously divide the measured value changes by using the state change time and the predicted change time in the object sequence as the dividing points, and obtain candidate segments arranged in chronological order of occurrence. The anomaly identification module is used to calculate the direction of change between the measured value at the first time and the measured value at the last time for each candidate segment, and to check whether there is a state change and a predicted change consistent with the direction of change in the candidate segment at the same time. The candidate segments that satisfy the corresponding relationship are retained to obtain the anomaly evolution segment of the target operation and maintenance object. The correlation assessment module determines the associated objects directly connected to the target maintenance object based on the energy exchange relationship, and calculates the change direction between the first and last time measurements of the associated objects in each abnormal evolution segment to obtain the object evolution record for each abnormal evolution segment. The status output module first checks whether the abnormal evolution segment has caused related objects to change synchronously based on the object evolution record. If it has, it determines that the target maintenance object is in a fault formation state. If it has not, it determines whether the target maintenance object is in a continuous deterioration state or a short-term disturbance state based on whether the abnormal evolution segment recurs and changes in the same direction, and outputs multi-dimensional evaluation results.

Images