Park situation assessment method and device based on dynamic weight distribution, medium and equipment

Abstract

The invention relates to the technical field of park management, in particular to a park situation assessment method and device based on dynamic weight distribution, a medium and equipment, and the method comprises the steps: collecting historical and real-time multi-source data, carrying out the data cleaning, and storing the data into a database; constructing a multi-dimensional index, standardizing the data stored in the database according to the multi-dimensional index, and meanwhile, performing exception labeling to obtain a sample set; training a random forest model by using the sample set and generating an initial weight of a multi-dimensional index; based on the initial weight of the multi-dimensional index, dynamic weight adaptive adjustment is carried out according to real-time anomaly detection; and performing weighted summation according to the index real-time weight obtained by self-adaptive adjustment to calculate the park situation index. According to the method, the evaluation accuracy of the park situation can be effectively improved, and the actual park management requirement can be better met.

Description

Technical Field

[0001] This invention relates to the field of park management technology, and in particular to a park situation assessment method, device, medium and equipment based on dynamic weight allocation. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] A park refers to a centrally planned and designated area where enterprises and companies of a specific industry or type are set up and managed in a unified manner. Typical examples include industrial parks, free trade parks, industrial parks, and animation parks.

[0004] Park safety management is a crucial component of park construction, encompassing aspects such as security monitoring, accident prevention, and emergency response. It is vital for ensuring the safety of personnel and property within the park and maintaining normal production and operational order. Intelligent technologies can effectively prevent and reduce safety accidents, thereby improving the overall safety level of the park. Park situation assessment is a key method in park safety management.

[0005] Traditional park situation assessment methods mainly include the following: Option 1: Fixed-weight evaluation model The Analytic Hierarchy Process (AHP) is used to pre-determine the weights of indicators (e.g., safety indicators account for 40%, energy consumption accounts for 30%), and the weights are manually updated periodically (e.g., monthly). A typical system is a comprehensive management platform for a certain park, but the weight adjustments lag behind actual changes in risk.

[0006] The scheme uses static indicator weights and employs expert scoring or fixed weights (such as the analytic hierarchy process), which cannot adapt to changes in park operations (such as the addition of new high-energy-consuming equipment or frequent safety incidents), resulting in lagging assessment results. Furthermore, the scheme's weight allocation lacks dynamic adaptability, and fixed weights or simple rule adjustments cannot reflect changes in the park's situation in real time (such as the weight of personnel mobility indicators should be significantly increased during the pandemic), causing the assessment results to deviate from the actual risks.

[0007] Option 2: Single-Dimensional Evaluation System The evaluation indicators are designed only for specific areas (such as safety or energy consumption) and are judged by thresholds (such as "if energy consumption exceeds the historical average by 10%, an early warning will be issued"), without forming a comprehensive evaluation system across dimensions.

[0008] The data in this solution is fragmented, relying only on data from a single domain (such as security sensors or energy meters), and does not integrate multi-dimensional data, making it difficult to fully reflect the situation in the park. Furthermore, the solution lacks multi-dimensional data integration and has not established correlations between cross-domain indicators (such as correlation analysis between security incidents and equipment operating status), making it difficult to discover potential risks and hazards (such as equipment overheating potentially causing a fire).

[0009] Option 3: Simple Dynamic Weight Algorithm The weights are adjusted based on historical data statistics (e.g., "if the number of abnormal occurrences of a certain indicator increases, the weight increases by 5%), but machine learning is not introduced. The weight adjustment rules are fixed and cannot handle non-linear correlations (e.g., the coupling effect between equipment failure and personnel gathering).

[0010] The scheme lacks real-time capability, the weight adjustment relies on manual intervention (such as quarterly updates), it lacks dynamic response capability to real-time emergencies (such as fire hazards and cyberattacks), and the assessment results cannot support emergency decision-making.

[0011] Therefore, this invention proposes a park situation assessment method based on dynamic weight allocation to solve the problems existing in the prior art. Summary of the Invention

[0012] To overcome the shortcomings of the prior art, the present invention provides a method, apparatus, medium and equipment for assessing the status of a park based on dynamic weight allocation, which can effectively improve the accuracy of park status assessment and better meet the actual needs of park management.

[0013] To achieve the above objectives, the present invention provides the following technical solution: Firstly, a method for assessing the situation of a park based on dynamic weight allocation is provided, the method comprising: Collect historical and real-time multi-source data, clean the data, and store it in the database; Construct multidimensional indicators, standardize the data stored in the database based on the multidimensional indicators, and perform anomaly labeling to obtain a sample set; The random forest model is trained using the sample set, and initial weights for the multidimensional index are generated. The initial weights are based on multidimensional indicators, and the weights are dynamically and adaptively adjusted according to real-time anomaly detection. The park status index is calculated by weighting and summing the indicators in real time based on the adaptively adjusted indicators.

[0014] Furthermore, Collect historical and real-time multi-source data by connecting to IoT sensors, business systems and external data sources; The IoT sensors include security cameras, temperature and humidity sensors, electricity meters, water meters, gas meters, equipment temperature sensors, infrared detectors, noise sensors, and air quality sensors; The business system includes an equipment management platform, a security management system, an energy consumption management system (energy consumption abnormal fluctuation record), an OA system, and an access control system; The external data sources include weather forecast interfaces, traffic monitoring, and public opinion monitoring.

[0015] Furthermore, The construction of multidimensional indicators, and the standardization of data stored in the database based on these multidimensional indicators, includes: The constructed multidimensional indicators include primary indicators and secondary sub-indicators. The primary indicators include safety indicators, equipment indicators, energy consumption indicators, personnel indicators, and environmental indicators. Each primary indicator includes several secondary sub-indicators. The multidimensional indicators are divided into numerical indicators and event-based indicators; Numerical indicator data is normalized, and event-type indicator data is one-hot encoded to generate feature vectors in a unified format.

[0016] Furthermore, The anomaly labeling includes: Historical anomaly events are obtained from multi-source historical data. Based on the historical abnormal events, the feature vectors in a unified format are labeled with the corresponding time periods. A label of 1 indicates an abnormal time period, and a label of 0 indicates a normal time period.

[0017] Furthermore, The initial weights for training the random forest model and generating multidimensional metrics using the sample set include: The sample set is input into a random forest model to learn the information gain of each indicator for abnormal events. An initial weight matrix for the multidimensional index is generated based on the information gain.

[0018] Furthermore, The initial weights based on multidimensional indicators are dynamically and adaptively adjusted according to real-time anomaly detection, including: Real-time standardized data is analyzed by using a sliding window of preset duration. The data in the sliding window is compared with preset thresholds to count the number of consecutive occurrences of outliers for each indicator. If the number of consecutive occurrences of an outlier in a certain indicator is greater than or equal to a preset number, the weight adjustment magnitude is calculated using a frequency sensitivity function. And adjust the magnitude according to the weight. Update the real-time weight of this metric:

[0019] in, Indicates the real-time weight of the indicator. This indicates the weight of the indicator before the update, and i represents the indicator number. If the number of consecutive occurrences of an outlier for a certain indicator is less than the preset number, the weight of that indicator will be adjusted back to its initial value.

[0020] Furthermore, The frequency sensitivity function is in exponential form:

[0021] Where a=1.5, and freq is the frequency of anomalies within the sliding window; Weight adjustment range The calculation formula is:

[0022] Where step is the step size for weight adjustment.

[0023] Furthermore, While dynamically adjusting weights based on real-time anomaly detection, the Spearman correlation coefficient is also used to calculate the correlation between indicators and eliminate redundant indicators.

[0024] Furthermore, The calculation of the park situation index based on the real-time weighted summation of indicators obtained through adaptive adjustment includes: Based on the physical zones of the park, the indicator data of each zone are weighted and summed according to the real-time weight of the indicators to obtain the regional situation index of each zone. The park's situation index is obtained by weighting the situation index of all regions according to their area proportion.

[0025] Furthermore, The formula for calculating the regional situation index is as follows:

[0026] in, This indicates the regional situation index. , Indicates the real-time weight of the indicator. This represents the standardized value of the indicator, where i represents the indicator number and n represents the number of indicators.

[0027] Furthermore, After calculating the park's situation index, a visual output is also provided, including: It provides heatmaps to display regional situation indices for each area, drill-down analysis of the contribution of each indicator dimension, and supports risk event tracing.

[0028] Secondly, a campus situation assessment device based on dynamic weight allocation is also provided, the device comprising: The data acquisition module is used to collect historical and real-time multi-source data, clean the data, and then store it in the database. The data preprocessing module is used to construct multidimensional indicators, standardize the data stored in the database based on the multidimensional indicators, and perform anomaly labeling to obtain a sample set. An initial weight generation module is used to train a random forest model using the sample set and generate initial weights for multidimensional indicators. The dynamic weight adjustment module is used to dynamically and adaptively adjust the initial weights based on multi-dimensional indicators and real-time anomaly detection. The situation index calculation module is used to calculate the park situation index by real-time weighted summation of indicators obtained through adaptive adjustment.

[0029] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing one or more programs, which, when executed, can realize the campus situation assessment method based on dynamic weight allocation as described above.

[0030] Based on the same inventive concept, the present invention also provides an electronic device, including a processor, a communication interface, a computer-readable storage medium as described above, and a communication bus; wherein the processor, the communication interface, and the computer-readable storage medium communicate with each other through the communication bus; the processor is used to execute a program stored in the computer-readable storage medium.

[0031] Compared with the prior art, the beneficial effects of the present invention are as follows: By collecting data from multiple sources, the data becomes more comprehensive. After constructing multidimensional indicators based on the collected data and performing data standardization and anomaly labeling, a sample set is obtained. The fusion of multidimensional indicators can more comprehensively reflect the park's situation. The initial weights of the multidimensional indicators are generated by training a random forest model. The real-time weights of the indicators are obtained by dynamically and adaptively adjusting the weights based on real-time anomaly detection. The park situation index is generated by combining the real-time weights and weighted summation, which effectively improves the accuracy of the park situation assessment and can better meet the actual needs of park management.

[0032] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings.

[0033] The invention will now be further described with reference to the accompanying drawings. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart illustrating a method for assessing the situation of a park based on dynamic weight allocation, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a heat map according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating dimensional contribution drilling and risk event tracing in one embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of a park situation assessment device based on dynamic weight allocation according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0037] like Figure 1 As shown, an embodiment of the present invention provides a method for assessing the situation of a park based on dynamic weight allocation. The method includes the following steps: S1. Collect historical and real-time multi-source data, clean the data, and store it in the database; S2. Construct multi-dimensional indicators, standardize the data stored in the database based on the multi-dimensional indicators, and simultaneously perform anomaly labeling to obtain a sample set; S3. Use the sample set to train a random forest model and generate initial weights for the multidimensional index; S4. Based on the initial weights of multi-dimensional indicators, the weights are dynamically and adaptively adjusted according to real-time anomaly detection; S5. Calculate the park status index by weighting and summing the indicators in real time based on the adaptively adjusted indicators.

[0038] The above technical solution collects historical and real-time multi-source data, resulting in more comprehensive data. Based on the collected data, multi-dimensional indicators are constructed, and after data standardization and anomaly labeling, a sample set is obtained. The fusion of multi-dimensional indicators can more comprehensively reflect the park's situation. The initial weights of the multi-dimensional indicators are generated by training a random forest model, and the real-time weights are obtained by dynamically and adaptively adjusting the weights based on real-time anomaly detection. The comprehensive evaluation index (park situation index) is generated by combining the real-time weights and weighted summation, which effectively improves the accuracy of park situation assessment and can better meet the actual needs of park management.

[0039] In this embodiment, the collected multi-source data is cleaned using an ETL tool and then stored in a data warehouse (i.e., a database).

[0040] As a preferred technical solution, historical and real-time multi-source data is collected by accessing IoT sensors, business systems, and external data sources; the IoT sensors include security cameras, temperature and humidity sensors, electricity meters, water meters, gas meters, equipment temperature sensors, infrared detectors, noise sensors, and air quality sensors; the business systems include equipment management platforms, security management systems, energy consumption management systems (energy consumption abnormal fluctuation records), OA systems, and access control systems; the external data sources include weather forecast interfaces, traffic monitoring, and public opinion monitoring.

[0041] This embodiment collects multi-dimensional historical and real-time data within the park, with data sources including IoT sensors, business systems, and external data sources.

[0042] IoT sensors mainly include security cameras (perimeter intrusion detection, people gathering detection), temperature and humidity sensors (indoor environmental parameters), electricity / water / gas meters (energy consumption data), equipment temperature sensors (equipment operating status), infrared detectors (dangerous area intrusion detection), noise / air quality sensors (environmental quality), etc.

[0043] The business systems mainly include an equipment management platform (equipment failure and maintenance records), a security management system (fire alarm and perimeter intrusion logs), an energy consumption management system (energy consumption abnormal fluctuation records), an OA system (personnel attendance and emergency drill records), and an access control system (unauthorized personnel intrusion records).

[0044] External data sources mainly include weather forecast interfaces (extreme weather warnings, such as heavy rain and high temperatures), traffic monitoring, and public opinion monitoring.

[0045] Collecting data from multiple sources, as mentioned above, can help improve the comprehensiveness of the park's situation assessment.

[0046] As a preferred technical solution, the construction of multi-dimensional indicators and the standardization of data stored in the database based on the multi-dimensional indicators include: The constructed multidimensional indicators include primary indicators and secondary sub-indicators. The primary indicators include safety indicators, equipment indicators, energy consumption indicators, personnel indicators, and environmental indicators. Each primary indicator includes several secondary sub-indicators. The multidimensional indicators are divided into numerical indicators and event-based indicators; Numerical indicator data is normalized, and event-type indicator data is one-hot encoded to generate feature vectors in a unified format.

[0047] The following section will further explain the multidimensional indicators constructed in this embodiment from the aspects of safety indicators, equipment indicators, energy consumption indicators, personnel indicators and environmental indicators.

[0048] 1. Safety Indicators: Core Assessment Dimensions of the Park's "Risk Prevention and Control Bottom Line" Safety indicators include secondary sub-indicators (such as "perimeter intrusion count" and "fire alarm response time"). Actual assessments focus on safety risks to personnel, property, and operational order within the park, specifically covering: Physical security risks: such as the frequency and response efficiency of incidents like perimeter intrusion, unauthorized entry, fire alarms (smoke / open flame detection), and security camera malfunctions (such as obstruction or offline status); Emergency response capabilities: such as the arrival time of security personnel after a fire alarm and the handling time for abnormal events (such as personnel injury).

[0049] 2. Equipment Indicators: Stability Assessment Dimensions of the Park's "Operational Hardware Infrastructure" Equipment indicators target the operational status of the park's core operating equipment. Secondary sub-indicators may include "Equipment Operating Status (Normal / Faulty)," "Equipment Fault Logs (e.g., inverter failure, air conditioning compressor malfunction)," "Equipment Load Rate (e.g., power equipment load, elevator operating load)," and "Equipment Maintenance Cycle Compliance Rate," etc. The evaluation focuses on: Equipment health status: such as the number of failures and duration of failures of critical equipment (power supply, air conditioning, elevators, production equipment, etc.); Equipment operating efficiency: such as whether the equipment load rate exceeds the safety threshold, equipment response speed (such as the door opening delay of the access control system), etc.

[0050] 3. Energy Consumption Indicators: Efficiency Evaluation Dimensions of the Park's "Green Operation and Cost Control" Energy consumption indicators focus on the rationality and efficiency of energy consumption in the park. Secondary sub-indicators may include "electricity consumption (kWh per unit time)," "water resource consumption (daily / hourly water consumption)," "natural gas / heat consumption," and "abnormal energy consumption fluctuations (such as a sudden increase in energy consumption in a certain area exceeding the historical average by 10%)." The assessment will focus on: Total energy consumption and trends: such as whether daily / hourly energy consumption exceeds the budget threshold, and the year-on-year / month-on-month growth rate of energy consumption; Abnormal energy consumption points: such as a sudden increase in energy consumption on a certain floor (possibly due to equipment failure or forgetting to turn off appliances), or excessive energy consumption during non-working hours (such as office air conditioners not being turned off).

[0051] 4. Personnel Indicators: A dimension for assessing the rationality of "personnel flow and management order" in the park. Personnel indicators target the activity status and management efficiency of personnel within the park. Secondary sub-indicators may include "personnel attendance compliance rate," "personnel density (e.g., whether the number of people in meeting rooms / lobby exceeds safe capacity)," "abnormal personnel movement patterns (e.g., unauthorized personnel entering restricted areas)," and "visitor registration compliance rate," etc. The evaluation will focus on: Personnel management order: such as the number of employees who are not on attendance, and the number of times unregistered visitors have entered; Personnel safety risks: such as the population density in a certain area exceeding the fire protection capacity, or unauthorized personnel entering core areas (such as computer rooms and finance offices).

[0052] 5. Environmental Indicators: Comfort Assessment Dimensions of the Park in "Personal Experience and Ecological Adaptation" Environmental indicators focus on the impact of the park's internal and external environment on people. Secondary sub-indicators may include "temperature and humidity (whether indoor and outdoor temperature and humidity are within the comfortable range)," "air quality (PM2.5 and CO2 concentrations)," "noise levels (decibels)," and "external weather impacts (such as heavy rain, high temperatures, see point 2 of the relevant document, "external data source weather forecasts")," etc. The assessment will focus on: Indoor environmental comfort: such as whether the office temperature and humidity exceed the range of human comfort (e.g., summer temperature > 26℃), and whether the CO2 concentration in the conference room is too high (affecting people's concentration); External environmental adaptability: such as water accumulation in the park due to heavy rain, and the risk of heatstroke for outdoor workers due to high temperatures (cross-validated by combining external weather forecast data).

[0053] In practice, the multidimensional indicators constructed in this embodiment include 5 major categories of primary indicators and 20+ secondary sub-indicators, as shown in the table below.

[0054]

[0055] This embodiment effectively solves the data silo problem by constructing the aforementioned multi-dimensional indicator system.

[0056] This invention integrates the above 5 categories and 20+ indicators, which can discover potential risk correlations (such as "personnel gathering + high equipment load" may cause safety hazards), and improve the coverage of assessment dimensions by 80%. By constructing the above multi-dimensional indicators and integrating them, the situation of the park can be reflected more comprehensively.

[0057] As a preferred technical solution, the anomaly labeling includes: obtaining historical anomaly events based on historical multi-source data; labeling the corresponding time period of the feature vector after uniform format according to the historical anomaly events, where a label of 1 indicates an abnormal time period and a label of 0 indicates a normal time period.

[0058] In this embodiment, data preprocessing mainly standardizes the collected multi-source data. When the system is initialized, the data standardization also includes anomaly labeling, such as outputting a sample set of "standardized feature vector + anomaly event label".

[0059] The following sections will provide further explanation of data standardization and anomaly labeling. 1. Data standardization: Numerical indicators (such as equipment temperature and energy consumption) are converted to values ​​in the [0,1] interval using Min-Max normalization, as shown in the formula: ; Event-based indicators (such as equipment failure, perimeter intrusion): are converted into binary values ​​of "0-1" using one-hot encoding (1 = abnormal occurrence, 0 = normal).

[0060] To facilitate the standardization of the collected multi-source data, this embodiment divides the multidimensional indicators into numerical indicators and event indicators. The numerical indicators (such as energy consumption values) are normalized, and the event indicators (such as equipment failures) are one-hot encoded, thereby generating feature vectors in a unified format.

[0061] In this embodiment, standardized data is used for initializing system weight matrix parameters, real-time anomaly detection (calculation), and weight iteration.

[0062] 2. Anomaly labeling (i.e., anomaly event labeling): Extract abnormal events (such as equipment failures and perimeter intrusions) from the historical logs of the business system, and label them with the corresponding time period y∈{0,1} (1=abnormal time period, 0=normal time period); finally forming a training sample set: ,in, The standardized feature vector of the secondary sub-indicators (n≥20) is where n is the number of secondary sub-indicators and m is the total number of samples.

[0063] As a preferred technical solution, the step of training a random forest model using the sample set and generating initial weights for multidimensional indicators includes: inputting the sample set into the random forest model, learning the information gain of each indicator for abnormal events, and generating an initial weight matrix for multidimensional indicators based on the information gain.

[0064] The following sections will provide further explanation of model training and the calculation of the initial weight matrix.

[0065] 1. Model Training Input the sample set obtained in step 2 (i.e. S2) into the random forest model, and the training objective is to "learn the information gain (IG) of each secondary sub-indicator for abnormal events (such as fire, equipment downtime)"; The random forest model learns through ensemble learning of multiple decision trees, statistically analyzing the contribution of each decision tree to reducing the uncertainty of anomaly event prediction when a certain sub-index is used as a split node, and taking the average value as the information gain of that sub-index. ; The higher the information gain, the greater the contribution of the sub-indicator to the identification of abnormal events (e.g., the information gain of equipment temperature to equipment failure is higher than that of personnel attendance). 2. Calculation of the initial weight matrix Based on the information gain, initial weights are assigned to ensure that the sum of all weights is 1. The calculation formula is as follows: ; Output the initial weight matrix ,in This represents the initial weight of the i-th secondary sub-indicator.

[0066] As a preferred technical solution, the initial weights based on multi-dimensional indicators, and the dynamic adaptive adjustment of weights according to real-time anomaly detection, include: analyzing real-time standardized data through a sliding window of a preset duration, comparing the data within the sliding window with a preset threshold, and counting the consecutive occurrences of anomalies for each indicator; if the consecutive occurrences of an anomaly for a certain indicator are greater than or equal to a preset number, then calculating the weight adjustment magnitude using a frequency-sensitive function. And adjust the magnitude according to the weight. Update the real-time weight of this metric: ,in, Indicates the real-time weight of the indicator. The value represents the weight of the indicator before the update, and i represents the indicator index. If the number of consecutive occurrences of an outlier for a certain indicator is less than a preset number, the weight of that indicator is adjusted back to its initial value. Furthermore, the frequency sensitivity function adopts an exponential form: Where a=1.5, freq is the frequency of anomalies within the sliding window; weight adjustment range. The calculation formula is: , where step is the weight adjustment step size.

[0067] In this embodiment, a value of 1.5 is preferred for 'a'. In specific implementation, the value of 'a' can be adjusted according to actual needs. 'a' and 'step' together determine the magnitude of the weight adjustment. 'a' controls the sensitivity of the impact of abnormal frequency on the magnitude of the weight adjustment. Their values ​​are verified through simulation using historical data to achieve the best results.

[0068] It should be noted that the number of occurrences (i.e., the number of consecutive occurrences of an outlier) and frequency (the frequency of anomalies within a sliding window) have different units of measurement, but in reality, since the window is a unit of time, the value of frequency is consistent with the number of occurrences.

[0069] This embodiment uses the initial weight as a benchmark, analyzes real-time data through a sliding window (preset duration: 1 hour), and triggers weight adjustment (i.e., dynamic weight adaptive adjustment).

[0070] The following section will further explain the dynamic weight adaptive adjustment process from three aspects: real-time anomaly detection, adjustment of triggering conditions, and weight iterative update.

[0071] 1. Real-time anomaly detection: Compare the real-time standardized data in the sliding window with a preset threshold (e.g., if the standardized value of the device temperature is >0.6, it is judged as an anomaly), and count the number of consecutive occurrences of an anomaly value of a certain sub-indicator; 2. Adjust trigger conditions: If the number of consecutive outlier occurrences of a certain sub-indicator is greater than or equal to the preset number (e.g., 3 times), the weight adjustment range will be calculated using a frequency-sensitive function. ; The frequency sensitivity function is in exponential form: (freq is the abnormal frequency within the sliding window; in this example, a=1.5), the adjustment range formula is: (step is the weight adjustment step size; in this embodiment, step=0.05). 3. Weight Iterative Update: If the abnormal frequency meets the target (abnormal frequency ≥ preset threshold), then update the real-time weight of the sub-indicator: (i.e., the weight adaptive algorithm); If the anomaly is mitigated (anomaly frequency < preset threshold), the weights are adjusted to return to near the initial value to ensure weight stability. Output real-time weight matrix .

[0072] As a preferred technical solution, while dynamically adjusting the weights based on real-time anomaly detection, the Spearman correlation coefficient is also used to calculate the correlation between indicators and eliminate redundant indicators.

[0073] This embodiment also performs correlation analysis: Spearman correlation coefficient is used to calculate the correlation between indicators, and redundant indicators are eliminated (such as highly correlated "equipment temperature" and "equipment load" are retained, only one of them), thus improving model efficiency.

[0074] The dynamic weight engine in this embodiment includes a weight adaptive algorithm and correlation analysis, and provides a dynamic weight allocation mechanism: a real-time adjustment algorithm for indicator weights based on machine learning (i.e., random forest model), which supports dynamic weight updates based on the frequency and correlation of abnormal events.

[0075] The weight adaptive algorithm in this embodiment is a data-driven dynamic weight algorithm that combines a machine learning random forest model (other models can also be used; this embodiment uses a random forest model). It automatically learns the correlation weights between indicators based on historical abnormal event data, and can respond to changes in the park's situation in real time.

[0076] The embodiments of the present invention adopt dynamic weights to effectively improve the accuracy of park situation index assessment. By automatically adjusting the weights through machine learning, the accuracy of abnormal event identification is increased to 95%, the false alarm rate is reduced by 60%, and the "missed judgment in high-risk scenarios" or "false alarm in normal scenarios" caused by traditional fixed weights are avoided.

[0077] As a preferred technical solution, the step of calculating the park situation index by weighted summation of real-time weights of indicators obtained through adaptive adjustment includes: weighting and summing the indicator data for each region according to the real-time weights of the indicators, based on the physical partitioning of the park, to obtain the regional situation index for each region; and then weighting all regional situation indices by their respective area proportions to obtain the park situation index. Further, the formula for calculating the regional situation index is: ,in, This indicates the regional situation index. , Indicates the real-time weight of the indicator. This represents the standardized value of the indicator, where i represents the indicator number and n represents the number of indicators.

[0078] The calculation of the regional situation index and the overall situation index will be explained in more detail below.

[0079] 1. Regional Situation Index Calculation: Based on the physical zones of the park (e.g., the North Zone Administration Building, the South Zone P2 Workshop, and the West Zone Perimeter), the secondary sub-indicator data for each zone are weighted and summed using the formula: "Standardized Value × Real-Time Weight".

[0080] in, The higher the value, the higher the risk of the area (generally, the initial values ​​are 75-100 = high risk, 50-74 = medium risk, and 0-49 = low risk). 2. Calculation of the overall situation index: The overall situation index of the park is obtained by weighting the situation index of all regions according to their area proportion. .

[0081] As a preferred technical solution, after calculating the park's situation index, a visualization output is also provided, including: providing a heat map to display the regional situation index of each area, drill-down analysis of the contribution of each indicator dimension, and supporting risk event tracing.

[0082] This embodiment uses visual output: it provides a heatmap displaying the situation index of each area, drill-down analysis of the contribution of each dimension (e.g., "Equipment dimension contributes 30%, mainly due to air conditioning compressor failure in P2 area"), and supports risk event tracing (e.g., clicking on abnormal indicators to view the real-time status of related equipment), specifically including: 1. Heat map display: Based on the physical layout of the park, "dark red-orange-light green" are used to map "high-medium-low" risks respectively, and the situation index of each area is displayed (e.g., laboratory area S=89, dark red; administration building S=42, light green). 2. Drill down based on dimensional contribution: The first layer displays the contribution percentage of the five primary indicators (safety, equipment, energy consumption, personnel, and environment) to the overall index (e.g., equipment contributes 30%). Second level: Drill down to the second-level sub-indicators under the first-level indicators (such as breaking down the equipment dimension into equipment failure 15% and equipment temperature 10%). 3. Risk event tracing: Click on the abnormal area or sub-indicator to trigger a details pop-up window, which displays "abnormal equipment number, abnormal frequency, related data, and handling progress" (e.g., the air conditioner compressor in P2 area malfunctioned, alarmed 3 times in 1 hour, and maintenance personnel have arrived on site).

[0083] Given the weaknesses in visualization and decision support of existing technology assessment results—such as outputting only a single numerical value or graded result (e.g., "Safety Level: Yellow") without providing contribution analysis of each dimension (e.g., "The decline in the safety index is mainly due to an increase in perimeter intrusion events")—which affects the accuracy of decision-making, this embodiment of the invention adds the aforementioned visualization output scheme. The indicator contribution analysis in this embodiment can quantify the impact of each dimension on the situation index (e.g., "The current weight of the equipment failure indicator is 35%, due to a recent increase in inverter failure events"), supporting rapid risk tracing and decision adjustment.

[0084] The visualization used in this invention can effectively support decision-making efficiency. For example, the contribution analysis function can reduce the time for managers to locate the source of risk from 30 minutes to 5 minutes, and help to quickly formulate targeted measures (such as prioritizing the repair of high-weight abnormal equipment).

[0085] A schematic diagram of a heat map in this embodiment is shown below. Figure 2As shown.

[0086] This embodiment illustrates a dimensional contribution drill-down and risk event tracing diagram, as shown below. Figure 3 As shown.

[0087] The following example, using a device failure scenario, illustrates one implementation process for calculating the park's situation index, including: Step 1: Multi-source data acquisition and preprocessing Collect equipment operation data (current, voltage, load rate), fault logs (historical maintenance records), environmental data (temperature and humidity in equipment room), etc., and generate feature vectors after standardization.

[0088] Step 2: Dynamic Weight Calculation If the number of equipment failures exceeds the threshold (e.g., 3 times) within 1 hour, the dynamic weighting engine will automatically increase the weight of the "equipment load rate" and "equipment temperature" indicators (e.g., from 15% to 25%), while decreasing the weight of irrelevant indicators (e.g., the weight of "personnel attendance" will be reduced from 5% to 3%).

[0089] Step 3: Situation Index Calculation and Early Warning The equipment dimension index is calculated based on the adjusted weights. If the equipment dimension index is greater than 80, a yellow warning is triggered, and a message is displayed that "the risk of equipment failure in P2 area has increased, and inspection is recommended".

[0090] In some embodiments, after calculating the park situation index, real-time early warning and decision support are also provided: a hierarchical early warning mechanism based on the situation index (such as four levels: red / yellow / blue / green) is used to link business systems to achieve automatic risk response.

[0091] The following example of a fire early warning scenario illustrates real-time early warning and decision support.

[0092] When the fire sensor detects that the smoke concentration exceeds the standard (safety indicator abnormality), and the temperature sensor data of the adjacent equipment is abnormal (equipment indicator abnormality), the dynamic weight engine automatically increases the weight of "smoke concentration" and "equipment temperature" (from 10% to 20%), calculates the comprehensive situation index (i.e., the park situation index, such as S=85), triggers a red alert, synchronously activates the fire protection system, and pushes the information to the park security personnel.

[0093] In summary, the embodiments of the present invention have at least the following characteristics: 1. Dynamic weight allocation mechanism + situation index calculation model improves assessment accuracy: The real-time adjustment algorithm of indicator weights based on machine learning supports dynamic updates of weights according to the frequency and correlation of abnormal events, realizing automatic adjustment of indicator weights with real-time data to adapt to complex scenario changes; a comprehensive assessment index (i.e., park situation index) is generated by weighted summation combined with dynamic weights, supporting multi-dimensional contribution analysis and risk tracing.

[0094] 2. A multi-dimensional indicator integration system comprehensively reflects the park's status: Construct a five-layer indicator system (5 major categories and 20+ sub-indicators) that includes safety, equipment, energy consumption, personnel, and environment, and define indicator standardization and correlation analysis methods.

[0095] 3. Visualization, real-time early warning, and decision support improve park management efficiency: Supports visualization (heat maps, contribution analysis of various dimensions, and risk event tracing), a graded early warning mechanism based on situation index (such as red / yellow / blue / green four levels), and linkage with business systems to achieve automatic risk response, assisting managers in implementing precise policies.

[0096] like Figure 4 As shown, an embodiment of the present invention also provides a campus situation assessment device based on dynamic weight allocation, the device comprising: The data acquisition module is used to collect historical and real-time multi-source data, clean the data, and then store it in the database. The data preprocessing module is used to construct multidimensional indicators, standardize the data stored in the database based on the multidimensional indicators, and perform anomaly labeling to obtain a sample set. An initial weight generation module is used to train a random forest model using the sample set and generate initial weights for multidimensional indicators. The dynamic weight adjustment module is used to dynamically and adaptively adjust the initial weights based on multi-dimensional indicators and real-time anomaly detection. The situation index calculation module is used to calculate the park situation index by real-time weighted summation of indicators obtained through adaptive adjustment.

[0097] Regarding the apparatus in the above embodiments, the specific manner in which each unit module performs its operations has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0098] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing one or more programs, which, when executed, can realize the campus situation assessment method based on dynamic weight allocation as described above.

[0099] Based on the same inventive concept, the present invention also provides an electronic device, such as... Figure 5 As shown, it includes a processor, a communication interface, a computer-readable storage medium as described above, and a communication bus; wherein the processor, the communication interface, and the computer-readable storage medium communicate with each other via the communication bus; the processor is used to execute a program stored in the computer-readable storage medium.

[0100] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0101] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0102] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0104] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0105] The parts not mentioned in the above embodiments are the same as or can be implemented using existing technologies, and will not be further described here.

[0106] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A park situation assessment method based on dynamic weight distribution, characterized in that, The method comprises: Collecting historical and real-time multi-source data, and storing the data in a database after data cleaning; Building multi-dimensional indexes, standardizing the data stored in the database according to the multi-dimensional indexes, and simultaneously performing abnormal labeling to obtain a sample set; Training a random forest model using the sample set and generating initial weights of the multi-dimensional indexes; Based on the initial weights of the multi-dimensional indexes, dynamically adjusting the weights according to real-time abnormal detection; Calculating a park situation index by weighted summation according to the real-time weights of the indexes obtained by adaptive adjustment.

2. The park situation evaluation method based on dynamic weight distribution according to claim 1, characterized in that: The historical and real-time multi-source data are collected by accessing Internet of Things sensors, business systems and external data sources; The Internet of Things sensors include security cameras, temperature and humidity sensors, electricity meters, water meters, gas meters, equipment temperature sensors, infrared detectors, noise sensors and air quality sensors; The business systems include equipment management platforms, security management systems, energy consumption management systems, OA systems and access control systems; The external data sources include weather forecast interfaces, traffic monitoring and public opinion monitoring.

3. The park situation evaluation method based on dynamic weight distribution according to claim 1, characterized in that: The multi-dimensional indexes are built, and the data stored in the database are standardized according to the multi-dimensional indexes, comprising: The built multi-dimensional indexes include primary indexes and secondary sub-indexes, the primary indexes include safety indexes, equipment indexes, energy consumption indexes, personnel indexes and environmental indexes, and each primary index includes a plurality of secondary sub-indexes; The multi-dimensional indexes are divided into numerical indexes and event indexes; The numerical index data are normalized, and the event index data are one-hot encoded to generate a uniform format feature vector.

4. The park situation evaluation method based on dynamic weight distribution according to claim 3, characterized in that: The abnormal labeling comprises: Historical abnormal events are obtained based on historical multi-source data; The labels of the corresponding period are labeled on the uniform format feature vector according to the historical abnormal events, and the label 1 represents an abnormal period and the label 0 represents a normal period.

5. The park situation evaluation method based on dynamic weight distribution according to claim 4, characterized in that: The random forest model is trained using the sample set, and the initial weights of the multi-dimensional indexes are generated, comprising: The sample set is input into the random forest model to learn the information gain of each index on abnormal events; The initial weight matrix of the multi-dimensional indexes is generated based on the information gain.

6. The park situation evaluation method based on dynamic weight distribution according to any one of claims 1-5, characterized in that: The initial weights of the multi-dimensional indexes are dynamically adjusted according to real-time abnormal detection, comprising: The real-time standardized data are analyzed through a preset length sliding window, the data in the sliding window are compared with a preset threshold, and the continuous occurrence number of abnormal values of each index is counted; If the number of continuous abnormal values of a certain index is greater than or equal to a preset number, the weight adjustment range is calculated by a frequency sensitive function , and the real-time weight of the index is updated according to the weight adjustment range ​ wherein, represents the index real-time weight, represents the index pre-update weight, i represents the index serial number; If the continuous occurrence number of abnormal values of an index is < a preset number, the weight of the index is adjusted to fall back to the initial value.

7. The park situation assessment method based on dynamic weight distribution according to claim 6, characterized in that, the frequency sensitive function adopts an exponential form: wherein a = 1.5, and freq is the abnormal frequency in the sliding window; weight adjustment range The calculation formula is: wherein step is the weight adjustment step.

8. The park situation assessment method based on dynamic weight distribution according to claim 1, characterized in that, while dynamically adjusting the weight according to real-time anomaly detection, the correlation between indexes is calculated using the Spearman correlation coefficient to exclude redundant indexes.

9. The park situation assessment method based on dynamic weight distribution according to claim 1, characterized in that, the real-time weight of the indexes obtained by adaptive adjustment is summed to calculate the park situation index, including: according to the real-time weight of the indexes, the index data of each region is weighted and summed to obtain the regional situation index of each region according to the physical zoning of the park; the regional situation indexes are weighted and averaged according to the area proportion of each region to obtain the park situation index.

10. The park situation assessment method based on dynamic weight distribution according to claim 9, characterized in that, the calculation formula of the regional situation index is: wherein, represents the regional situation index, , represents the index real-time weight, represents the index standardized value, i represents the index serial number, and n represents the index number.

11. The park situation assessment method based on dynamic weight distribution according to claim 1, characterized in that, after calculating the park situation index, visualization output is performed, including: providing a heat map to display the regional situation index of each region, drilling down to analyze the contribution of each index dimension, and supporting risk event tracing.

12. A park situation assessment device based on dynamic weight distribution, characterized in that, the device includes: a data acquisition module for acquiring historical and real-time multi-source data, and storing the data in a database after data cleaning; a data preprocessing module for constructing multi-dimensional indexes, standardizing the data stored in the database according to the multi-dimensional indexes, and performing anomaly labeling to obtain a sample set; an initial weight generation module for training a random forest model using the sample set and generating initial weights of the multi-dimensional indexes; a dynamic weight adjustment module for dynamically adjusting the weight based on the initial weights of the multi-dimensional indexes according to real-time anomaly detection; a situation index calculation module for calculating the park situation index by weighting and summing the real-time weights of the indexes obtained by adaptive adjustment.

13. A computer-readable storage medium storing one or more programs, characterized in that, when the one or more programs are executed, the park situation assessment method based on dynamic weight distribution of any one of claims 1-11 can be implemented.

14. An electronic device comprising a processor, a communication interface, the computer readable storage medium of claim 13 and a communication bus; wherein, the processor, the communication interface, and the computer-readable storage medium communicate with each other through a communication bus; characterized in that, the processor is configured to execute the programs stored in the computer-readable storage medium.

Images