A full life cycle intelligent operation and maintenance method based on BIM model

Abstract

The present invention discloses a full life cycle intelligent operation and maintenance method based on a BIM model, which relates to the technical field of intelligent operation and maintenance of buildings, including collecting voiceprint data and preprocessing the voiceprint data; setting adaptive weights for the preprocessed voiceprint data according to the credibility of the equipment and the data update frequency, and fusing them according to the adaptive weights to generate a comprehensive voiceprint feature vector; performing real-time analysis on the comprehensive voiceprint feature vector through an anomaly detection model to identify abnormal voiceprint information; generating an operation and maintenance plan based on the abnormal voiceprint information and combining the operation and maintenance data of the building's full life cycle, and feeding back the operation and maintenance decision to the BIM model to update the operation and maintenance plan. The present invention realizes efficient fusion and accurate anomaly recognition of multi-device voiceprint data through two core steps: dynamic adaptive weight setting and fusion, and anomaly detection model based on RNN and GRU.

Description

Technical Field

[0001] The present invention relates to the technical field of building intelligent operation and maintenance, and in particular to a full life cycle intelligent operation and maintenance method based on a BIM model. Background Art

[0002] With the development of Building Information Modeling (BIM) technology, the intelligent management of the entire life cycle of buildings has become one of the core research directions of the construction industry. BIM technology can provide accurate digital models in the stages of building design, construction and operation and maintenance, greatly improving the management efficiency of buildings. However, with the continuous increase in the scale and complexity of buildings, traditional operation and maintenance methods are difficult to effectively cope with the dynamic changes in the entire life cycle of buildings. In recent years, intelligent operation and maintenance solutions have gradually become an important means to solve the problems of building operation and maintenance management. These solutions achieve real-time monitoring and operation and maintenance optimization of building status by combining advanced technologies such as the Internet of Things (IoT), big data analysis, and deep learning. However, most of the existing technologies focus on analyzing sensor data or equipment operation status, and rarely involve data collection and analysis of human participation in the operation and maintenance process, especially voiceprint data. Voiceprint data, as a unique biometric information, can reflect the operation status of equipment and the work behavior of operation and maintenance personnel, but its application in the existing BIM intelligent operation and maintenance system is still in its initial stage.

[0003] Although the existing intelligent operation and maintenance solutions have improved the maintenance efficiency of buildings to a certain extent, there are still some shortcomings that cannot be ignored. First, the processing capacity of existing technologies in data fusion is limited. The integration of multi-source heterogeneous data usually relies only on the data input of sensors, and it is difficult to effectively utilize the various types of data generated during the operation and maintenance process, especially voiceprint data. Since voiceprint data has certain real-time and individual differences, when analyzing it, existing technologies often cannot dynamically adjust it according to the credibility of the equipment and the update frequency of the data, resulting in low data fusion efficiency. Secondly, most of the existing operation and maintenance decisions are based on static analysis, lacking a real-time discovery and response mechanism for dynamic changes in the operation and maintenance process. Although the application of deep learning technology has developed, it is usually only used for the prediction and detection of equipment failures, and has not fully considered the changes in voiceprint characteristics throughout the life cycle of the building and its impact on operation and maintenance decisions. Therefore, how to better integrate voiceprint data and improve the real-time and accuracy of operation and maintenance decisions is an urgent problem to be solved in the current intelligent operation and maintenance methods. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a full life cycle intelligent operation and maintenance method based on the BIM model to solve the problems of low efficiency of voiceprint data fusion and insufficient real-time operation and maintenance decision-making in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides an intelligent operation and maintenance method for the entire life cycle based on a BIM model, which includes collecting voiceprint data and preprocessing the voiceprint data; setting adaptive weights for the preprocessed voiceprint data according to the credibility of the equipment and the data update frequency, and fusing them according to the adaptive weights to generate a comprehensive voiceprint feature vector; performing real-time analysis on the comprehensive voiceprint feature vector through an anomaly detection model to identify abnormal voiceprint information; generating an operation and maintenance plan based on the abnormal voiceprint information and the operation and maintenance data of the entire life cycle of the building, and feeding back the operation and maintenance decisions to the BIM model to update the operation and maintenance plan.

[0008] As a preferred solution of the intelligent operation and maintenance method for the entire life cycle based on the BIM model described in the present invention, the voiceprint data includes the sound generated during the operation of the building equipment, the voice commands of the operation and maintenance personnel, and the environmental noise.

[0009] As a preferred solution of the intelligent operation and maintenance method for the entire life cycle based on the BIM model of the present invention, the specific steps of preprocessing the voiceprint data are as follows:

[0010] Wavelet denoising is used to filter the noise of the initial voiceprint data;

[0011] Use blind source separation algorithm to separate the device operation sound and the operation and maintenance personnel's voice commands;

[0012] Process the voiceprint data by framing to obtain the time series characteristics of the voiceprint data;

[0013] Use the threshold judgment method to identify and filter abnormal signals in voiceprint data.

[0014] As a preferred solution of the intelligent operation and maintenance method for the entire life cycle based on the BIM model of the present invention, wherein: according to the credibility of the equipment and the data update frequency, an adaptive weight is set for the pre-processed voiceprint data, and fusion is performed according to the adaptive weight. The specific steps are as follows:

[0015] Based on the failure rate, maintenance timeliness rate and performance deviation of the comprehensive equipment, the three indicators are nonlinearly combined by geometric mean, so that the low value of a single indicator constrains the overall credibility. The natural logarithm function is added to amplify the impact of the maintenance timeliness rate. The expression is:

[0016] ;

[0017] in, is the normalized failure rate of the ith device, is the normalized maintenance timeliness rate of the i-th equipment, is the normalized performance deviation of the i-th device, is the credibility of the i-th device;

[0018] Trustworthiness of the equipment , data update frequency and signal strength are transformed nonlinearly, and the absolute value penalty term of the difference between devices is added to dynamically balance the weights. The expression is:

[0019] ;

[0020] in, is the adaptive weight of the ith device, is the data update frequency of the i-th device, is the signal strength of the ith device, is the power parameter of the device credibility, is the weight coefficient of device credibility, is the weight coefficient of data update frequency, is the weight coefficient of signal strength, N is the total number of devices, is the difference between the i-th device and the j-th device, is the credibility of the jth device used for normalization, is the model strength of the jth device used for normalization, is the data update frequency of the jth device used for normalization.

[0021] As a preferred solution of the intelligent operation and maintenance method for the whole life cycle based on the BIM model of the present invention, the fusion is performed according to the adaptive weight, and the specific steps are as follows:

[0022] Adaptive weighting based on device , adaptively adjust the weight of the voiceprint data collected by each device at the current time point t through the time decay factor;

[0023] When the current time point t is close to the observation time of the device, the weight of the voiceprint data collected by the current device is automatically increased;

[0024] When the current time point t is far away from the observation time of the device, the weight of the voiceprint data collected by the current device is automatically reduced;

[0025] The device voiceprint data is dynamically fused in the time dimension, and the expression is:

[0026] ;

[0027] in, is the comprehensive voiceprint feature vector at time t, is the voiceprint data of the i-th device at time point t, is the parameter that controls the time decay rate, Indicates the time point when the i-th device observes the voiceprint data, represents the adaptive weight of the jth device used for normalization, Indicates the time point when the j-th device observes the voiceprint data for normalization.

[0028] As a preferred solution of the intelligent operation and maintenance method for the entire life cycle of the BIM model described in the present invention, the specific steps are as follows: the comprehensive voiceprint feature vector is analyzed in real time by the anomaly detection model to identify abnormal voiceprint features.

[0029] Build an anomaly detection model based on RNN and GRU;

[0030] The input layer of the anomaly detection model inputs each sliding window into the GRU and extracts the time series dependency of the comprehensive voiceprint feature vector through the RNN;

[0031] The hidden layer of the anomaly detection model adopts a multi-layer GRU stacking structure to output high-dimensional feature representation of the time series;

[0032] The comprehensive voiceprint feature vector Input anomaly detection model for real-time analysis, real-time analysis of comprehensive voiceprint feature vector The reconstruction error of the sliding window is expressed as:

[0033] ;

[0034] in, represents the reconstruction error of the comprehensive voiceprint feature vector at time point t, represents the number of consecutive time points used to calculate the reconstruction error, k represents the time position index within the current sliding window, Indicates time The actual comprehensive voiceprint feature vector at express The comprehensive voiceprint feature value reconstructed at each moment.

[0035] As a preferred solution of the intelligent operation and maintenance method for the entire life cycle based on the BIM model of the present invention, the specific steps of identifying abnormal voiceprint information are as follows:

[0036] Based on the reconstruction error distribution in historical data, define the abnormal judgment threshold τ;

[0037] like >τ, it is considered that the comprehensive voiceprint information at the current time point t is abnormal and the current device is abnormal;

[0038] like When τ, it is considered that the comprehensive voiceprint information at the current time point t is normal and the current device is normal.

[0039] As a preferred solution of the intelligent operation and maintenance method for the whole life cycle based on the BIM model of the present invention, wherein: according to the abnormal voiceprint information, combined with the operation and maintenance data of the whole life cycle of the building, an operation and maintenance plan is generated, and the operation and maintenance decision is fed back to the BIM model to update the operation and maintenance plan. The specific steps are as follows:

[0040] According to the recognition results, the start and end time periods of the abnormal voiceprint information and the corresponding devices are stored in the abnormal database, and the operation and maintenance history data of the building life cycle are also associated;

[0041] Retrieve historical operation and maintenance records related to abnormal equipment from the operation and maintenance data in the BIM model, and identify abnormal patterns by analyzing historical data and current abnormal voiceprint information;

[0042] Allocate operation and maintenance resources based on abnormal patterns and combine them with the building's operation and maintenance resources, and generate an operation and maintenance plan based on the equipment status, operation and maintenance resources, and the building's overall operation plan;

[0043] The generated operation and maintenance plan and abnormal voiceprint information are fed back to the BIM model and stored in the knowledge base of the BIM model to update future anomaly detection and operation and maintenance plans. The abnormal voiceprint information and updated operation and maintenance plan are pushed to relevant operation and maintenance personnel through the interface of the BIM model.

[0044] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent operation and maintenance method for the entire life cycle based on the BIM model as described in the first aspect of the present invention.

[0045] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent operation and maintenance method for the entire life cycle based on the BIM model as described in the first aspect of the present invention.

[0046] The beneficial effects of the present invention are as follows: through the two core steps of dynamic adaptive weight setting and fusion and anomaly detection model based on RNN and GRU, efficient fusion of multi-device voiceprint data and accurate anomaly recognition are achieved, solving the problem of feature distortion and insufficient real-time anomaly detection caused by the differences in multi-device data in traditional methods; combined with the preprocessing of voiceprint data, abnormal pattern analysis and operation and maintenance plan generation and dynamic update based on the BIM model, a closed-loop management system from anomaly detection to operation and maintenance decision-making is constructed, which greatly improves the intelligence level, operation efficiency and decision-making accuracy of the operation and maintenance of building equipment throughout the life cycle. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0048] Figure 1 This is a flow chart of the intelligent operation and maintenance method for the entire life cycle based on the BIM model in Example 1.

[0049] Figure 2 This is a flowchart for identifying abnormal voiceprint information in Example 1. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0051] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0052] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0053] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a full life cycle intelligent operation and maintenance method based on a BIM model, comprising the following steps:

[0054] S1: Collect voiceprint data and pre-process the voiceprint data.

[0055] S1.1: Voiceprint data includes the sounds generated by the operation of construction equipment, the voice commands of operation and maintenance personnel, and environmental noise.

[0056] It should be noted that collecting the operating sounds of construction equipment, voice commands of operation and maintenance personnel, and environmental noise can comprehensively obtain key voiceprint information of the equipment operating status and the operation and maintenance process, and provide multi-dimensional data support for subsequent anomaly detection and operation and maintenance decisions.

[0057] S1.2: Use wavelet denoising to filter the noise of the initial voiceprint data;

[0058] For example, the wavelet denoising algorithm is used to remove the ambient wind noise from the collected operating sounds of construction equipment to improve the purity of the voiceprint data.

[0059] S1.3: Use the blind source separation algorithm to separate the device operation sound and the operation and maintenance personnel's voice commands;

[0060] For example, blind source separation technology is used to distinguish the mechanical vibration sound during equipment operation from the voice commands of maintenance personnel to avoid data confusion.

[0061] S1.4: Process the voiceprint data by framing to obtain the time series characteristics of the voiceprint data;

[0062] For example, the continuous voiceprint data during the operation of the device is divided into time segments of fixed length to extract voiceprint features at different time points.

[0063] S1.5: Use the threshold determination method to identify and filter abnormal signals in voiceprint data.

[0064] For example, abnormal sharp sounds that exceed the set threshold in the device operation voiceprint are identified and eliminated to avoid abnormal signals interfering with the analysis results.

[0065] S2: According to the credibility of the device and the data update frequency, set adaptive weights for the preprocessed voiceprint data, and fuse them according to the adaptive weights to generate a comprehensive voiceprint feature vector.

[0066] S2.1: Based on the failure rate, maintenance timeliness rate and performance deviation of the comprehensive equipment, the three indicators are nonlinearly combined by geometric mean, so that the low value of a single indicator constrains the overall credibility, and the influence of the natural logarithm function on the maintenance timeliness rate is amplified. The expression is:

[0067] ;

[0068] in, is the normalized failure rate of the ith device, is the normalized maintenance timeliness rate of the i-th equipment, is the normalized performance deviation of the i-th device, is the credibility of the i-th device;

[0069] Furthermore, the three indicators are first normalized to have the same dimension and range; then, the three indicators are nonlinearly combined using the geometric mean method to ensure that the low value of a single indicator has a constraining effect on the overall credibility, that is, when any indicator performs poorly, the overall credibility will be significantly lowered; in addition, in order to further highlight the impact of maintenance timeliness, the natural logarithm function is used to amplify it, and finally a credibility indicator that can comprehensively reflect the operating status of the equipment is generated to ensure that the credibility evaluation is more scientific and comprehensive.

[0070] S2.2: Trustworthiness of the equipment , data update frequency and signal strength are transformed nonlinearly, and the absolute value penalty term of the difference between devices is added to dynamically balance the weights. The expression is:

[0071] ;

[0072] in, is the adaptive weight of the ith device, is the data update frequency of the i-th device, is the signal strength of the ith device, is the power parameter of the device credibility, is the weight coefficient of device credibility, is the weight coefficient of data update frequency, is the weight coefficient of signal strength, N is the total number of devices, is the difference between the i-th device and the j-th device, is the credibility of the jth device used for normalization, is the model strength of the jth device used for normalization, is the data update frequency of the jth device used for normalization.

[0073] It should be noted that Source: Determined through statistical analysis and experimental verification of equipment operation data to determine the optimal value that can reflect the actual status of the equipment and meet specific operation and maintenance needs.

[0074] Function: Dynamically adjust the nonlinear response degree of equipment credibility to each indicator, enhance the constraint effect of low-value indicators on the overall credibility, and thus improve the accuracy of status assessment.

[0075] It should also be said that i means that the adaptive weight is currently being calculated. target device.

[0076] i is the focus of weight calculation, and the credibility of a specific device (the i-th device) is calculated. , Data update frequency , signal strength The weighted contribution of the device is calculated by overall normalization. .

[0077] i represents the number of the specific device, and the weight at this time It is calculated separately for the i-th device.

[0078] j represents the ID of other devices in the device set, which is used for normalization and calculation of inter-device differences.

[0079] j traverses the total number of devices N, which is used to calculate the parameters of all devices (such as , , ), thereby normalizing the weight of the i-th device to ensure that the sum of all weights is 1. At the same time, j is also used to calculate the penalty term for differences between devices , that is, the difference between the i-th device and the j-th device, further dynamically balances the weights.

[0080] S2.3: Adaptive weights based on device , adaptively adjust the weight of the voiceprint data collected by each device at the current time point t through the time decay factor;

[0081] S2.3.1: When the current time point t is close to the observation time of the device, the weight of the voiceprint data collected by the current device is automatically enhanced;

[0082] For example, when the voiceprint data collected by the device 10 seconds ago is very close to the current time point t, the weight is automatically increased to give priority to the latest voiceprint data to reflect the current operating status of the device.

[0083] S2.3.2: When the current time point t is far away from the observation time of the device, the weight of the voiceprint data collected by the current device is automatically reduced;

[0084] For example, when the voiceprint data collected by the device 5 hours ago is far away from the current time point t, the weight is automatically reduced, thereby reducing the impact of expired data on the current analysis results.

[0085] S2.4: Dynamically fuse the device voiceprint data in the time dimension, the expression is:

[0086] ;

[0087] in, is the comprehensive voiceprint feature vector at time t, is the voiceprint data of the i-th device at time point t, is the parameter that controls the time decay rate, Indicates the time point when the i-th device observes the voiceprint data, represents the adaptive weight of the jth device used for normalization, Indicates the time point when the j-th device observes the voiceprint data for normalization.

[0088] It should be noted that by dynamically fusing the device voiceprint data in the time dimension, the voiceprint data of each device is adjusted according to its adaptive weight and the distance between the acquisition time and the current time, so that the voiceprint data close to the current time point contributes more to the comprehensive features, while the influence of expired data far from the current time point is weakened. This can not only effectively highlight the real-time and timeliness of the device voiceprint data, but also balance the differences in data quality and weight among multiple devices, thereby generating more accurate and reliable comprehensive voiceprint features, and providing high-quality data support for device status monitoring and anomaly detection.

[0089] S3: The anomaly detection model is used to analyze the comprehensive voiceprint feature vector in real time to identify abnormal voiceprint information.

[0090] S3.1: Build an anomaly detection model based on RNN and GRU;

[0091] It should be noted that the anomaly detection model is built based on RNN and GRU because they can efficiently capture the time series characteristics and long-term and short-term dependencies of voiceprint data. At the same time, GRU improves computational efficiency by simplifying the structure and is suitable for processing real-time anomaly detection tasks.

[0092] S3.2: The input layer of the anomaly detection model inputs each sliding window into the GRU and extracts the time series dependency of the comprehensive voiceprint feature vector through the RNN;

[0093] The time dependency of the comprehensive voiceprint feature vector refers to the changing relationship of the comprehensive voiceprint feature vector at different time points, that is, how the feature vector at the current time point is affected by the feature vector at the past time point. This dependency helps to capture the changing pattern of the device operating status over time.

[0094] S3.3: The hidden layer of the anomaly detection model adopts a multi-layer GRU stacking structure to output a high-dimensional feature representation of the time series;

[0095] The high-dimensional feature representation of time series refers to a more abstract, complex and information-rich feature set extracted from the comprehensive voiceprint feature vector after being processed by a multi-layer GRU stacking structure. These features can better reflect the essential attributes of the device's operating status and provide strong support for subsequent anomaly detection.

[0096] S3.4: Comprehensive voiceprint feature vector Input anomaly detection model for real-time analysis, real-time analysis of comprehensive voiceprint feature vector The reconstruction error of the sliding window is expressed as:

[0097] ;

[0098] in, represents the reconstruction error of the comprehensive voiceprint feature vector at time point t, represents the number of consecutive time points used to calculate the reconstruction error, k represents the time position index within the current sliding window, Indicates time The actual comprehensive voiceprint feature vector at express The comprehensive voiceprint feature value reconstructed at each moment.

[0099] It should be noted that by inputting the comprehensive voiceprint feature vector into the anomaly detection model for real-time analysis, the change of the equipment operation status can be effectively monitored by calculating the difference between the actual voiceprint features and the model-reconstructed voiceprint features over a period of time. This method can detect abnormal conditions of the equipment in a timely manner, because when the equipment operation status changes, the gap between the actual voiceprint features and the voiceprint features predicted by the model will become larger. In this way, abnormal conditions of the equipment can be quickly identified and early warning can be given, thereby reducing the impact of equipment failures and improving reliability and safety.

[0100] S3.5: Based on the reconstruction error distribution in the historical data, define the abnormal judgment threshold τ;

[0101] S3.5.1: If >τ, the comprehensive voiceprint information at the current time point t is considered abnormal, and the current device abnormal;

[0102] For example, if at a certain point in time, the reconstruction error of the comprehensive voiceprint feature vector exceeds the abnormal judgment threshold τ, this means that the voiceprint information at that point in time is significantly different from the normal pattern, indicating that the current device may have an abnormal situation and requires further inspection or maintenance.

[0103] S3.5.2: If When τ, it is considered that the comprehensive voiceprint information at the current time point t is normal and the current device is normal.

[0104] For example, if the reconstruction error of the comprehensive voiceprint feature vector remains below the abnormality judgment threshold τ, it means that the voiceprint information at that time point is consistent with the pattern when the device is operating normally. It is considered that the current device is in good operating condition and no obvious abnormality is detected.

[0105] S4: Generate an operation and maintenance plan based on abnormal voiceprint information and combined with the operation and maintenance data of the entire life cycle of the building, and feed back the operation and maintenance decisions to the BIM model to update the operation and maintenance plan.

[0106] S4.1: Based on the recognition results, the start and end time periods of the abnormal voiceprint information and the corresponding devices are stored in the abnormal database, and the operation and maintenance history data of the building life cycle are associated;

[0107] Furthermore, based on the results of abnormal identification, the detected abnormal voiceprint information, including the start and end time periods of the abnormality and the relevant equipment identification, is stored in a special abnormality database. At the same time, the operation and maintenance history data related to the abnormal equipment is retrieved from the BIM model, such as previous maintenance records, maintenance conditions, etc., and these historical data are associated with the current abnormal information for subsequent analysis and processing. This can not only help the operation and maintenance personnel quickly understand the abnormal conditions of the equipment, but also provide a basis for formulating effective maintenance plans.

[0108] S4.2: Identify abnormal patterns by analyzing historical data and current abnormal voiceprint information;

[0109] For example, abnormal patterns include periodic abnormalities, sudden abnormalities, gradual abnormalities, environmental impact abnormalities, and operational error abnormalities, etc. These patterns are identified by analyzing historical data and current abnormal voiceprint information. For example, by comparing abnormal voiceprint features that appear multiple times in a device within a specific time period, periodic abnormalities can be identified; and by detecting sudden increases in voiceprint changes, sudden abnormalities can be identified.

[0110] S4.3: Allocate operation and maintenance resources based on the abnormal pattern and the operation and maintenance resources of the building, and generate an operation and maintenance plan based on the equipment status, operation and maintenance resources and the overall operation plan of the building;

[0111] The operation and maintenance plan includes equipment maintenance schedule, required spare parts list, external maintenance support requirements, maintenance personnel scheduling arrangements, and emergency response measures. For example, if the equipment needs to be replaced, check whether the spare parts inventory is sufficient. If the spare parts are insufficient, place an order immediately; if external maintenance support is required, generate an outsourcing request and contact a qualified service provider; at the same time, according to the equipment status and the overall operation plan of the building, reasonably arrange maintenance time to ensure that maintenance activities will not affect the normal operation of the building.

[0112] S4.4: Feed the generated operation and maintenance plan and abnormal voiceprint information back to the BIM model, store it in the knowledge base of the BIM model, update future abnormal detection and operation and maintenance plans, and push abnormal voiceprint information and updated operation and maintenance plans to relevant operation and maintenance personnel through the interface of the BIM model.

[0113] It should be noted that the specific process includes storing this information in the knowledge base of the BIM model to update the future anomaly detection standards and operation and maintenance plans. At the same time, through the interface of the BIM model, abnormal voiceprint information and updated operation and maintenance plans are automatically pushed to relevant operation and maintenance personnel to ensure that all relevant personnel can obtain important information in a timely manner. This process not only helps to improve the accuracy of anomaly detection and make future operation and maintenance plans more scientific and reasonable, but also ensures that maintenance work can be carried out efficiently and orderly, reducing delays and errors caused by information asymmetry, and ultimately improving the operation efficiency and service quality of building facilities.

[0114] This embodiment also provides a computer device, which is suitable for the intelligent operation and maintenance method of the whole life cycle based on the BIM model, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent operation and maintenance method of the whole life cycle based on the BIM model as proposed in the above embodiment.

[0115] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0116] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing intelligent operation and maintenance of the entire life cycle based on the BIM model as proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0117] In summary, the present invention achieves efficient fusion of multi-device voiceprint data and accurate anomaly identification through two core steps: dynamic adaptive weight setting and fusion, and anomaly detection model based on RNN and GRU, which solves the problem of feature distortion and insufficient real-time anomaly detection caused by the differences in multi-device data in traditional methods; combined with the preprocessing of voiceprint data, abnormal pattern analysis, and operation and maintenance plan generation and dynamic update based on the BIM model, a closed-loop management system from anomaly detection to operation and maintenance decision-making is constructed, which greatly improves the intelligence level, operation efficiency and decision-making accuracy of the operation and maintenance of building equipment throughout the life cycle.

[0118] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of an intelligent operation and maintenance method for the entire life cycle based on a BIM model are provided.

[0119] In order to verify the effectiveness of the full life cycle intelligent operation and maintenance method based on the BIM model, this example selects a typical commercial building as the research object. Four key devices in the building are selected, namely two air-conditioning units (AC1, AC2), an elevator (EL1), and a lighting system (LT1). The test goal is to evaluate the performance of this method in anomaly detection and operation and maintenance plan generation by collecting and analyzing the voiceprint data of these devices.

[0120] The initial performance indicators of the four key devices are shown in Table 1.

[0121] Table 1 Equipment initial performance index table

[0122]

[0123] The reliability of the equipment is calculated by normalizing the failure rate, maintenance timeliness rate, and performance deviation and then combining the three nonlinearly. After normalization, the geometric mean combined with the natural logarithm function is used to calculate the credibility of each device, as follows:

[0124] ;

[0125] ;

[0126] ;

[0127] ;

[0128] ;

[0129] ;

[0130] ;

[0131] ;

[0132] The credibility results are shown in Table 2:

[0133] Table 2 Equipment credibility evaluation table

[0134]

[0135] Calculate the adaptive weight of each device based on the device's credibility, data update frequency, and signal strength The weights are used to balance the feature contributions between devices and dynamically adjust the differences between devices, as follows:

[0136] in, ;

[0137] Data update frequency ;

[0138] Signal Strength ;

[0139] Differences ;

[0140] The molecular part is as follows:

[0141] ;

[0142] The normalized weight part is as follows:

[0143] ;

[0144] ;

[0145] The adaptive weight results are shown in Table 3.

[0146] Table 3 Adaptive weight allocation table

[0147]

[0148] The reconstruction error of the comprehensive voiceprint feature vector is calculated by averaging the squared error between the actual value and the reconstructed value in the sliding window, as follows:

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] The reconstruction error results are shown in Table 4:

[0154] Table 4 Comprehensive voiceprint feature vector reconstruction error table

[0155]

[0156] Combined with the above process, the final equipment status assessment result table is generated, as shown in Table 5:

[0157] Table 5 Equipment status comprehensive evaluation results

[0158]

[0159] Through the data analysis of the above table, it can be clearly seen that there are differences in the operating status of different equipment. For example, LT1 has the highest credibility (0.876), but its failure rate is high (0.041) and its maintenance timeliness rate is the lowest (0.857), indicating that its performance is relatively stable but needs to strengthen fault response and maintenance optimization; while EL1 has the lowest failure rate (0.018) and the highest maintenance timeliness rate (0.973), but the lowest credibility (0.574), indicating that it may have long-term performance risks. The present invention calculates the credibility of the equipment by comprehensively analyzing indicators such as failure rate, maintenance timeliness rate, and performance deviation. and adaptive weights , combined with voiceprint data reconstruction error , dynamically adjust the contribution of equipment characteristics, accurately evaluate the operating status, and provide a decision-making basis for equipment operation and maintenance. For example, for EL1 with low credibility, long-term performance monitoring can be strengthened, while for LT1 with high failure rate, the maintenance plan needs to be optimized. This method effectively improves the accuracy of equipment anomaly detection and the intelligence level of operation and maintenance plan, and verifies the feasibility of the full life cycle intelligent operation and maintenance method based on BIM model.

[0160] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A full life cycle intelligent operation and maintenance method based on BIM model, characterized by: include, Collect voiceprint data and pre-process the voiceprint data; According to the credibility of the device and the frequency of data update, an adaptive weight is set for the pre-processed voiceprint data, and fusion is performed according to the adaptive weight to generate a comprehensive voiceprint feature vector; The anomaly detection model is used to analyze the comprehensive voiceprint feature vector in real time to identify abnormal voiceprint information; Generate an operation and maintenance plan based on abnormal voiceprint information and operation and maintenance data of the building's entire life cycle, and feed the operation and maintenance decision back to the BIM model to update the operation and maintenance plan; According to the credibility of the device and the data update frequency, the pre-processed voiceprint data is set with an adaptive weight, and fused according to the adaptive weight. The specific steps are as follows: Based on the failure rate, maintenance timeliness rate and performance deviation of the comprehensive equipment, the three indicators are nonlinearly combined by geometric mean, so that the low value of a single indicator constrains the overall credibility. The natural logarithm function is added to amplify the impact of the maintenance timeliness rate. The expression is: ； in, is the normalized failure rate of the ith device, is the normalized maintenance timeliness rate of the i-th equipment, is the normalized performance deviation of the i-th device, is the credibility of the i-th device; Trustworthiness of the equipment , data update frequency and signal strength are transformed nonlinearly, and the absolute value penalty term of the difference between devices is added to dynamically balance the weights. The expression is: ； in, It is Adaptive weights for each device, It is The data update frequency of each device, It is The signal strength of each device, is the power parameter of the device credibility, is the weight coefficient of device credibility, is the weight coefficient of data update frequency, is the weight coefficient of signal strength, is the total number of devices, It is Device and The difference between the devices, is used for normalization The reliability of the equipment, is used for normalization The signal strength of each device, is used for normalization The data update frequency of each device; The specific steps of performing fusion according to the adaptive weight are as follows: Adaptive weighting based on device , adaptively adjust the current time point through the time decay factor The weight of the voiceprint data collected by each device; Current time point When the observation time of the device is close, the weight of the voiceprint data collected by the current device is automatically increased; Current time point When you are away from the device's observation time, the weight of the voiceprint data collected by the current device will be automatically reduced; The device voiceprint data is dynamically fused in the time dimension, and the expression is: ； in, It's time The comprehensive voiceprint feature vector at the moment, It is Devices at time Voiceprint data at all times, is the parameter that controls the time decay rate, Indicates The time point when the device observed the voiceprint data, Indicates the first Adaptive weights for each device, Indicates the first The time point when a device observes the voiceprint data.

2. The intelligent operation and maintenance method for the entire life cycle based on the BIM model as claimed in claim 1, characterized in that: The voiceprint data includes the sound generated when the construction equipment is running, the voice instructions of the operation and maintenance personnel, and the environmental noise.

3. The intelligent operation and maintenance method for the entire life cycle based on the BIM model as claimed in claim 2, characterized in that: The specific steps of preprocessing the voiceprint data are as follows: Wavelet denoising is used to filter the noise of the initial voiceprint data; Use blind source separation algorithm to separate the device operation sound and the operation and maintenance personnel's voice commands; Process the voiceprint data by framing to obtain the time series characteristics of the voiceprint data; Use the threshold judgment method to identify and filter abnormal signals in voiceprint data.

4. The intelligent operation and maintenance method for the entire life cycle based on the BIM model as claimed in claim 3, characterized in that: The specific steps of performing real-time analysis on the comprehensive voiceprint feature vector through the anomaly detection model to identify abnormal voiceprint features are as follows: Build an anomaly detection model based on RNN and GRU; The input layer of the anomaly detection model inputs each sliding window into the GRU and extracts the time series dependency of the comprehensive voiceprint feature vector through the RNN; The hidden layer of the anomaly detection model adopts a multi-layer GRU stacking structure to output high-dimensional feature representation of the time series; The comprehensive voiceprint feature vector Input anomaly detection model for real-time analysis, real-time analysis of comprehensive voiceprint feature vector The reconstruction error of the sliding window is expressed as: ； in, Indicates time point The reconstruction error of the comprehensive voiceprint feature vector is represents the number of consecutive time points used in calculating the reconstruction error, Represents the time position index within the current sliding window, Indicates time The actual comprehensive voiceprint feature vector at express The comprehensive voiceprint feature value reconstructed at each moment.

5. The intelligent operation and maintenance method for the entire life cycle based on the BIM model as claimed in claim 4, characterized in that: The specific steps of identifying abnormal voiceprint information are as follows: Based on the reconstruction error distribution in historical data, define the abnormal judgment threshold τ; like >τ, then the current time point is considered The comprehensive voiceprint information of the current device is abnormal. abnormal; like ≤τ, the current time point is considered The comprehensive voiceprint information is normal and the current device is normal.

6. The intelligent operation and maintenance method for the entire life cycle based on the BIM model as claimed in claim 5, characterized in that: The operation and maintenance plan is generated based on the abnormal voiceprint information and the operation and maintenance data of the entire life cycle of the building, and the operation and maintenance decision is fed back to the BIM model to update the operation and maintenance plan. The specific steps are as follows: According to the recognition results, the start and end time periods of the abnormal voiceprint information and the corresponding devices are stored in the abnormal database, and the operation and maintenance history data of the building life cycle are also associated; Retrieve historical operation and maintenance records related to abnormal equipment from the operation and maintenance data in the BIM model, and identify abnormal patterns by analyzing historical data and current abnormal voiceprint information; Allocate operation and maintenance resources based on abnormal patterns and combine them with the building's operation and maintenance resources, and generate an operation and maintenance plan based on the equipment status, operation and maintenance resources, and the building's overall operation plan; The generated operation and maintenance plan and abnormal voiceprint information are fed back to the BIM model and stored in the knowledge base of the BIM model to update future anomaly detection and operation and maintenance plans. The abnormal voiceprint information and updated operation and maintenance plan are pushed to relevant operation and maintenance personnel through the interface of the BIM model.

Images