Intelligent campus information data analysis method and system

Abstract

The invention discloses a smart campus information data analysis method and system, and belongs to the technical field of data processing. According to the method, the equipment state monitoring data set corresponding to each primary screening internet-of-things equipment monitoring behavior event in the campus internet-of-things operation state log is acquired, the target state monitoring data set is determined in the multiple equipment state monitoring data sets, and then a series of technical means such as state vector mining, energy consumption quantification and combinatorial analysis are combined. Accurate management, effective operation state analysis and energy-saving optimization of the campus Internet of Things equipment can be realized, so that the overall operation efficiency and the energy utilization efficiency of the campus Internet of Things system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and specifically relates to a smart campus information data analysis method and system. Background Art

[0002] With the continuous development of campus informatization, the number and types of campus IoT devices are increasing, such as various devices in smart classrooms (projectors, smart lighting systems, air conditioners, etc.) and environmental sensors (temperature, humidity, light sensors, etc.). These devices play an important role in improving campus teaching, management and service levels, but also bring a series of technical problems.

[0003] In terms of campus IoT device management, traditional management methods often lack accurate judgment and priority distinction of device status. In terms of device operation status analysis, existing technical means usually simply collect device operation data and lack the ability to deeply mine data and conduct focused analysis. In terms of campus energy consumption management, due to the lack of in-depth understanding of the relationship between device operation status and energy consumption, the existing energy consumption management methods are relatively extensive.

[0004] It can be seen that the existing campus Internet of Things equipment management technology is difficult to ensure the overall operation efficiency and energy utilization efficiency of the campus Internet of Things system. Summary of the invention

[0005] The present invention provides a smart campus information data analysis method and system, which can solve or partially solve the technical problems involved in the above-mentioned background technology.

[0006] The present invention provides a method for analyzing smart campus information data. The execution subject of the method is a data analysis system. The method comprises: obtaining a device status monitoring data set corresponding to each pre-screened IoT device monitoring behavior event in a campus IoT operation status log, determining a target status monitoring data set from a plurality of the device status monitoring data sets, wherein the target status monitoring data set is used to indicate a target IoT device monitoring behavior event with the highest priority among a plurality of the pre-screened IoT device monitoring behavior events; and performing status vector analysis on the campus IoT operation status log according to the target status monitoring data set and each of the device status monitoring data sets. Quantity mining is performed to obtain several IoT operation state vectors, and the state monitoring attention vectors corresponding to the target state monitoring data set and each of the device state monitoring data sets are refined; each of the IoT operation state vectors is respectively combined with the corresponding state monitoring attention vector to obtain several global state monitoring vectors; the operation energy consumption of the campus IoT operation state log is identified to obtain a first operation energy consumption quantification vector, and the first operation energy consumption quantification vector and several of the global state monitoring vectors are combined to obtain energy consumption regulation auxiliary information; energy consumption regulation analysis is performed based on the energy consumption regulation auxiliary information to generate a target energy consumption regulation optimization strategy for the target IoT device monitoring behavior event.

[0007] The present invention provides a data analysis system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the above method.

[0008] The present invention provides a readable storage medium, on which a program or instruction is stored, and the program or instruction implements the steps of the above method when executed by a processor.

[0009] The present invention obtains the device status monitoring data set corresponding to each initially screened IoT device monitoring behavior event in the campus IoT operation status log, determines the target status monitoring data set among several device status monitoring data sets, and then combines a series of technical means such as state vector mining, energy consumption quantification and combination analysis to achieve accurate management of campus IoT devices, effective operation status analysis and energy-saving optimization, thereby improving the overall operation efficiency and energy utilization efficiency of the campus IoT system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flow chart of a smart campus information data analysis method provided by the present invention.

[0011] Figure 2 This is a structural schematic diagram of a data analysis system provided by the present invention. DETAILED DESCRIPTION

[0012] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the present invention, "and / or" means at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0014] Figure 1 A smart campus information data analysis method is shown, which is applied to a data analysis system. The method includes the following steps 110-130.

[0015] Step 110: The data analysis system obtains the device status monitoring data set corresponding to each preliminary screened IoT device monitoring behavior event in the campus IoT operation status log, and determines a target status monitoring data set from several of the device status monitoring data sets, wherein the target status monitoring data set is used to indicate the target IoT device monitoring behavior event with the highest priority among several of the preliminary screened IoT device monitoring behavior events.

[0016] In step 110, the campus IoT operation status log is a comprehensive record of the operation status of IoT devices on campus. It contains the device status monitoring data set corresponding to each initially screened IoT device monitoring behavior event. For example, there are IoT devices such as smart classroom equipment (such as projectors, smart lighting systems, air conditioners, etc.), environmental sensors (temperature, humidity, light sensors, etc.) on campus. Each monitoring behavior event of each IoT device will generate a corresponding device status monitoring data set.

[0017] Taking the intelligent lighting system as an example, its equipment status monitoring data set may include the following data: light brightness value (the value range can be 0-1000 lux, for example, the current brightness value is 300 lux), on-time (in minutes, for example, the current on-time is 60 minutes), number of switches (for example, 3 times), etc. For the temperature sensor in the environmental sensor, the equipment status monitoring data set may include the collected temperature value (for example, 25 degrees Celsius), the collection time interval (for example, 10 minutes), etc.

[0018] Among the numerous device status monitoring data sets, a target status monitoring data set needs to be determined. The target status monitoring data set is used to indicate the target IoT device monitoring behavior events with the highest priority among several pre-screened IoT device monitoring behavior events. Priority can be determined based on a variety of factors, such as device importance weight and urgency.

[0019] 1) Equipment importance weight For example, higher weights can be given to equipment that ensures normal teaching, such as projectors and teaching computers in smart classrooms. For example, the weight of the projector is 0.8, the weight of the teaching computer is 0.7, and the weight of the temperature sensor in the environmental sensor is 0.5 (relatively speaking, the projector and teaching computer have a greater direct impact on teaching).

[0020] 2) Urgency If a device fails or is in an abnormal state that may lead to serious consequences, its urgency is high. For example, if the air conditioning system fails (temperature is out of control), it may seriously affect the teaching environment in the hot summer, and the priority of the monitoring behavior events related to the air conditioning system failure will be increased. For example, when the air conditioning system is operating normally, the priority of its equipment status monitoring data set is 0.6. When there is an abnormal temperature fluctuation (such as exceeding the normal range by more than 5 degrees Celsius), the priority is increased to 0.9.

[0021] By comprehensively considering these factors, the target state monitoring data set can be determined from several equipment state monitoring data sets. For example, at a certain moment, if the projector has an early warning that the bulb is about to break (this event has the highest comprehensive priority after calculation), the equipment state monitoring data set related to the projector will be determined as the target state monitoring data set.

[0022] Step 120: The data analysis system mines the state vectors of the campus IoT operation status log based on the target state monitoring data set and each of the device state monitoring data sets to obtain a number of IoT operation state vectors, and extracts the state monitoring attention vectors corresponding to the target state monitoring data set and each of the device state monitoring data sets; and combines each of the IoT operation state vectors with the corresponding state monitoring attention vector to obtain a number of global state monitoring vectors.

[0023] In the present invention, step 120 involves three branch steps: state vector mining, refining state monitoring attention vectors, and combining to obtain a global state monitoring vector.

[0024] 1. State vector mining 1) State vector mining based on target state monitoring data set: The data analysis system mines the state vector of the campus IoT operation status log based on the target state monitoring data set. Taking the previously determined projector as an example, if its target state monitoring data set contains data such as bulb life (for example, the remaining life is 500 hours and the full life is 1000 hours), frequency of use (10 times a week), etc. These data can be quantified and normalized to form a state vector. For example, the bulb life can be quantified as 0.5 (the ratio of the remaining life to the full life), and the frequency of use can be quantified as 0.5 (for example, the maximum number of uses per week is 20 times), then an exemplary state vector of the projector can be [0.5, 0.5].

[0025] 2) Perform state vector mining based on each device status monitoring data set: Similarly, this operation is performed for other devices. For example, for a temperature sensor, the temperature value it collects is 25 degrees Celsius, and the normal comfortable temperature range is 22-26 degrees Celsius. The temperature value can be converted to a value between 0 and 1, and the calculation method is: (25-22) / (26-22)=0.75 (for example). If the collection time interval is 10 minutes and the maximum acceptable time interval is set to 15 minutes, this factor is quantified to 0.67 (10 / 15). The state vector of the temperature sensor can be [0.75, 0.67].

[0026] 2. Extracting the state monitoring attention vector For the target state monitoring data set: For the target state monitoring data set (projector related), the state monitoring attention vector is refined according to its importance and current state in the entire campus Internet of Things system. For example, if the projector is very critical in the current teaching activity (such as an important teaching demonstration is in progress), and the situation of its bulb about to be damaged requires high attention, then its state monitoring attention vector can be [0.9, 0.8] (the first value represents the attention weight to the bulb life, and the second value represents the attention weight to the frequency of use). The weight in the present invention is determined according to the current state and importance of the device. For example, if the bulb life is close to the critical value, the attention weight to it will be higher.

[0027] For each device status monitoring data set: For the temperature sensor, considering that temperature has a certain impact on the teaching environment, but it is not the most urgent situation, its status monitoring attention vector can be [0.6, 0.5] (the first value represents the attention weight of the temperature value, and the second value represents the attention weight of the collection time interval).

[0028] (III) Combine to get the global state monitoring vector Combination process: Combine each IoT operation state vector with the corresponding state monitoring attention vector. For the projector, its global state monitoring vector is calculated as follows: the state vector is [0.5, 0.5], the state monitoring attention vector is [0.9, 0.8], and the combined global state monitoring vector is [0.5*0.9, 0.5*0.8]=[0.45, 0.4]. For the temperature sensor, the state vector is [0.75, 0.67], the state monitoring attention vector is [0.6, 0.5], and the combined global state monitoring vector is [0.75*0.6, 0.67*0.5]=[0.45, 0.335].

[0029] Step 130: The data analysis system identifies the operating energy consumption of the campus IoT operating status log to obtain a first operating energy consumption quantification vector, combines the first operating energy consumption quantification vector and a number of the global status monitoring vectors to obtain energy consumption regulation auxiliary information; performs energy consumption regulation analysis based on the energy consumption regulation auxiliary information, and generates a target energy consumption regulation optimization strategy for the target IoT device monitoring behavior event.

[0030] In detail, step 130 involves three sections: operating energy consumption identification, combining to obtain energy consumption regulation auxiliary information, and generating a target energy consumption regulation optimization strategy.

[0031] (I) Operation energy consumption identification: The data analysis system identifies the operation energy consumption of the campus IoT operation status log. Taking the smart lighting system as an example, its energy consumption is related to factors such as light brightness and on-time. For example, the power of the smart lighting system is 100 watts and the on-time is 60 minutes (1 hour). According to the formula: energy consumption = power × time, its energy consumption is 100 watts × 1 hour = 0.1 kWh. If there are multiple lighting areas, add up the energy consumption of each area to get the total energy consumption of the lighting system. For the air-conditioning system, for example, its power is 1500 watts and the operating time is 30 minutes (0.5 hours), and the energy consumption is 1500 watts × 0.5 hours = 0.75 kWh. Arrange the energy consumption of all IoT devices on campus in a certain order (such as by device type or area) to obtain the first operation energy consumption quantification vector. For example, if there are intelligent lighting systems, air conditioning systems and projectors on campus (the projector has a power of 300 watts and consumes 0.3 kWh of electricity when running for 1 hour), the first operating energy consumption quantization vector can be [0.1, 0.75, 0.3].

[0032] (ii) Combine to obtain energy consumption regulation auxiliary information: Combine the first operation energy consumption quantization vector and several global state monitoring vectors. For example, in addition to the vectors related to the intelligent lighting system, air conditioning system and projector mentioned above, there are other devices (such as humidity sensors, whose global state monitoring vector is [0.3, 0.2]). The combined energy consumption regulation auxiliary information may be a matrix or vector group, for example: [[0.1, 0.75, 0.3], [0.45, 0.4], [0.45, 0.335], [0.3, 0.2]] (the first row is the first operation energy consumption quantization vector, and the following rows are the global state monitoring vectors of each device).

[0033] 3. Energy consumption regulation analysis and generation of target energy consumption regulation optimization strategy Energy consumption regulation analysis algorithm example - linear regression model: A linear regression model can be used for energy consumption regulation analysis. For example, there is a linear relationship between energy consumption (Y) and the state factors of the device (such as the brightness of the smart lighting system, the temperature setting of the air conditioning system, etc., represented by X1, X2, etc.), that is, Y=a1X1+a2X2+…+b (where a1, a2, etc. are coefficients and b is a constant). Taking the smart lighting system as an example, the relationship between brightness (X1) and energy consumption (Y) may be Y=0.01X1+0.05 (for example, the coefficient obtained by fitting historical data). According to the data in the energy consumption regulation auxiliary information, the state factors of each device are substituted into the linear regression model to analyze the impact of different device states on energy consumption. For example, for the smart lighting system, if the current brightness is 300 lux, the current energy consumption and energy consumption forecasts at different brightness are calculated according to the model.

[0034] Generate target energy consumption adjustment optimization strategy: Based on the above analysis, generate target energy consumption adjustment optimization strategy for monitoring behavior events of target IoT devices (such as the use of projectors). If the energy consumption of a projector is related to the usage time and brightness (for example, a projector also has a brightness adjustment function), it is found through analysis that reducing the usage time of the projector or reducing its brightness can reduce energy consumption without affecting the teaching effect. For example, reducing the brightness of the projector from the current 800 lumens to 600 lumens (based on the analysis of the visibility of teaching content, 600 lumens can still meet the viewing needs), and shortening the usage time from 2 hours to 1.5 hours, can effectively reduce energy consumption. Then the target energy consumption adjustment optimization strategy can be: adjust the brightness of the projector to 600 lumens, and control the usage time of each time to no more than 1.5 hours. For other devices, corresponding energy consumption adjustment optimization strategies can also be generated based on similar analysis, such as the intelligent lighting system adjusting the brightness and opening time according to the natural light intensity and classroom usage, and the air conditioning system adjusting the temperature setting according to the classroom temperature and personnel density.

[0035] In the above technical scheme, the IoT operation state vector, state monitoring attention vector, global state monitoring vector, first operation energy consumption quantization vector, and energy consumption regulation auxiliary information are the key technical features involved in the process of generating the target energy consumption regulation optimization strategy. These technical features are exemplarily introduced and explained below.

[0036] The IoT operation state vector is a quantitative representation of the operation state of campus IoT devices, which reflects the various operation state characteristics of the devices in the form of feature vectors. It is mined from the device state monitoring data set.

[0037] Taking the air conditioning equipment in the smart classroom as an example, its IoT operation state vector may contain multiple features. For example, the operation mode of the air conditioner (cooling, heating, ventilation, which can be represented by 0, 1, and 2 respectively), the set temperature (for example, the value range is 16-30 degrees Celsius, the actual set temperature is 22 degrees Celsius, which can be normalized to the interval [0, 1], and the present invention is (22-16) / (30-16)≈0.43), the current power (for example, the air conditioner power range is 500-2000 watts, the current power is 1000 watts, and the normalized value is (1000-500) / (2000-500)=0.33), the fan speed (for example, the speed range is 500-1500 rpm, the current speed is 800 rpm, and the normalized value is (800-500) / (1500-500)=0.3), etc.

[0038] Then the IoT operation state vector of the air-conditioning device can be expressed as [vec{v}_{run}=[0, 0.43, 0.33, 0.3]]. Each element in the present invention is a feature, which together describe the current operation state of the air-conditioning device.

[0039] The IoT operation state vector can fully reflect the operation state of the device and help with subsequent analysis. For example, in terms of energy management, the current energy consumption level of the device can be understood by analyzing the power characteristics in the IoT operation state vector; in equipment fault diagnosis, the combination of the operation mode and other characteristics can help determine whether the device is in an abnormal state. If the air conditioner is in cooling mode (the first characteristic is 0), but the set temperature is very high and the power is very low, it may indicate that the cooling system is faulty.

[0040] The condition monitoring attention vector is a quantitative representation of the degree of attention paid to different features in equipment condition monitoring. It reflects which equipment condition features require more attention in a specific situation.

[0041] For the above air-conditioning equipment, the state monitoring attention vector will be different in different scenarios. For example, in the energy management scenario, the set temperature and power may be the key features, while the fan speed is relatively less important. For example, if the attention weight for the set temperature is 0.8, the attention weight for the power is 0.7, the attention weight for the fan speed is 0.3, and the attention weight for the operating mode is 0.2 (because in energy management, the operating mode is relatively fixed and has little impact on energy consumption), then the state monitoring attention vector is [vec{v}_{attn}=[0.2, 0.8, 0.7, 0.3]].

[0042] The state monitoring attention vector is used to guide the analysis of the equipment state. When performing data analysis, by combining the IoT operation state vector with the state monitoring attention vector (e.g., element-by-element multiplication), the impact of important features on the overall analysis can be highlighted. In energy management, this combination can more accurately evaluate the state related to equipment energy consumption so that effective energy consumption regulation measures can be taken.

[0043] The global state monitoring vector is the result of combining the IoT operation state vector and the state monitoring attention vector. It integrates the operation state of the device and the degree of attention paid to each state feature.

[0044] For air conditioning equipment, the global state monitoring vector is calculated as: [vec{v}_{global}=vec{v}_{run}odotvec{v}_{attn}]; Where (odot) represents the element-wise multiplication operation.

[0045] That is, [vec{v}_{global} =[0*0.2, 0.43*0.8, 0.33*0.7, 0.3*0.3] =[0, 0.344, 0.231, 0.09]].

[0046] The global state monitoring vector provides a more comprehensive and focused description of the device status. It can be used to conduct a comprehensive assessment of the device status at the level of the entire campus IoT system. For example, in the status comparison of multiple devices, the global state monitoring vector can determine which devices need more attention or adjustment in terms of energy consumption, and which devices are operating relatively normal and in line with expectations.

[0047] The first operation energy consumption quantification vector is a quantitative representation of the operation energy consumption of campus IoT devices. It contains the energy consumption information of each device or device group.

[0048] For example, there are three main energy-consuming devices on campus: intelligent lighting system, air conditioning equipment and projector. The energy consumption of the intelligent lighting system is 0.2 kWh, the energy consumption of the air conditioning equipment is 1.5 kWh, and the energy consumption of the projector is 0.3 kWh. Then the first operation energy consumption quantization vector is [vec{v}_{energy}=[0.2, 1.5, 0.3]]. Each element in the present invention corresponds to the energy consumption of a device.

[0049] The first operation energy consumption quantification vector is an important basis for campus energy consumption management. By analyzing the first operation energy consumption quantification vector, we can intuitively understand the contribution of different equipment in energy consumption. For example, we can find that air conditioning equipment is a large energy consumer, so in the energy consumption regulation strategy, we can focus on optimizing air conditioning equipment, such as adjusting temperature settings, optimizing operating time, etc.

[0050] The energy consumption regulation auxiliary information is an information set that combines the first operation energy consumption quantization vector and the global state monitoring vector (of multiple devices).

[0051] For example, in addition to the above-mentioned air-conditioning equipment, there are also related vectors of intelligent lighting systems and projectors.

[0052] The global state monitoring vector of the intelligent lighting system is: [vec{v}_{global-light}=[0.1, 0.2, 0.15, 0.05]].

[0053] The global state monitoring vector of the projector is: [vec{v}_{global-proj}=[0.3, 0.1, 0.2, 0.1]]. Then the energy consumption regulation auxiliary information can be expressed as a matrix form: [ begin{bmatrix} 0.2&1.5&0.3 0&0.344&0.231&0.09 0.1&0.2&0.15&0.05 0.3&0.1&0.2&0.1 end{bmatrix} ] The first row in the above matrix form is the first operation energy consumption quantization vector, and the following rows are the global state monitoring vectors of the air conditioner, the intelligent lighting system and the projector.

[0054] Energy consumption regulation auxiliary information provides a comprehensive data foundation for energy consumption regulation analysis. It combines energy consumption information with comprehensive status information of the equipment, and can use various data analysis methods (such as data mining and machine learning algorithms) to mine the relationship between equipment energy consumption and status, thereby generating effective energy consumption regulation optimization strategies. For example, cluster analysis can be used to classify equipment according to energy consumption and status characteristics, and different energy consumption regulation strategies can be formulated for different categories of equipment.

[0055] The present invention first indicates the target IoT device monitoring behavior events with the highest priority by determining the target status monitoring data set, and can accurately focus on the most critical or urgent device conditions in the campus IoT system. This helps to quickly locate the part that has the greatest impact on the overall campus IoT operation among many device monitoring events, avoiding indiscriminate treatment of all device states, thereby improving the pertinence and efficiency of campus IoT device management. For example, in teaching activities, it can give priority to ensuring the normal operation of core teaching equipment (projectors, teaching computers, etc.) in smart classrooms, reducing the risk of teaching interruptions due to equipment failures or abnormalities.

[0056] Secondly, the process of mining the state vector and refining the state monitoring attention vector, and then combining it to obtain the global state monitoring vector, realizes a more comprehensive, detailed and focused quantitative description of the equipment operation status. On the one hand, state vector mining can deeply mine the hidden information in the equipment operation status data, convert the complex equipment status data into a quantifiable and analyzable vector form, and provide a unified mathematical model basis for subsequent analysis. On the other hand, the extraction of the state monitoring attention vector enables different attention weights to be given to different equipment status characteristics according to actual needs during the analysis process, such as the importance of the equipment, the current task requirements, or the focus of energy consumption management. The combined global state monitoring vector integrates the equipment status and focus, which can not only more accurately reflect the actual operation status of the equipment, but also make more reasonable comparisons and evaluations between different devices, providing a more reliable basis for the optimization management of the overall campus Internet of Things system.

[0057] Furthermore, the operation energy consumption is identified to obtain the first operation energy consumption quantification vector, and it is combined with the global status monitoring vector to obtain energy consumption regulation auxiliary information. This process provides comprehensive and effective data support for the energy consumption management of campus Internet of Things devices. The first operation energy consumption quantification vector intuitively reflects the energy consumption of each device, so that energy consumption management can be targeted. After combining it with the global status monitoring vector, the energy consumption regulation auxiliary information can closely link the operation status of the device with the energy consumption, and fully consider the impact of various operation status characteristics of the device on energy consumption. This helps to discover the potential relationship between device energy consumption and operation status, such as the energy consumption change law of the device under different power, operation time, working mode and other states.

[0058] Finally, based on the auxiliary information of energy consumption regulation, energy consumption regulation analysis is performed and the target energy consumption regulation optimization strategy is generated, which can realize the precise energy consumption control of campus IoT devices. Through the comprehensive analysis of a large amount of data, the generated target energy consumption regulation optimization strategy can formulate personalized and precise energy-saving measures according to the actual operating status and energy consumption of different devices. This not only helps to reduce campus energy consumption and reduce operating costs, but also improves energy utilization efficiency and achieves the goal of energy conservation and emission reduction under the premise of ensuring the normal operation of campus IoT devices, which is in line with the pursuit of green and sustainable development in modern campuses.

[0059] In a possible embodiment, the first operating energy consumption quantization vector and several global state monitoring vectors are combined to obtain energy consumption regulation auxiliary information, including: respectively determining the data involvement weights of each of the device state monitoring data sets, and using the data involvement weights to combine the global state monitoring vectors corresponding to each of the device state monitoring data sets to obtain a first state monitoring combination vector; combining the first operating energy consumption quantization vector, the first state monitoring combination vector, and the global state monitoring vector corresponding to the target state monitoring data set to obtain energy consumption regulation auxiliary information.

[0060] Furthermore, the target energy consumption regulation optimization strategy is generated by a first energy consumption regulation decision algorithm, and the first operating energy consumption quantification vector, the first state monitoring combination vector, and the global state monitoring vector corresponding to the target state monitoring data set are combined to obtain energy consumption regulation auxiliary information, including: obtaining regulation requirement information for guiding the first energy consumption regulation decision algorithm to generate an energy consumption regulation optimization strategy; refining the regulation requirement vector of the regulation requirement information, combining the first operating energy consumption quantification vector, the regulation requirement vector, the first state monitoring combination vector, and the global state monitoring vector corresponding to the target state monitoring data set to obtain energy consumption regulation auxiliary information.

[0061] In the campus IoT system, different equipment status monitoring data sets have different impacts on the overall energy consumption regulation. For example, for the smart lighting system, the two data features of brightness value and opening time may have a greater impact on energy consumption, while for the temperature sensor, although the collected temperature data has a certain correlation with energy consumption, the correlation is relatively small. Therefore, it is necessary to determine the data involvement weight of each equipment status monitoring data set.

[0062] Take the intelligent lighting system as an example. For example, it contains three data features: brightness value (range: 0-1000 lux), on-time (in hours) and number of switches. Through the analysis of historical energy consumption data and equipment operation data, it is found that the brightness value is linearly related to energy consumption, with a coefficient of 0.6 (that is, for every 100 lux increase in brightness, energy consumption increases by 0.6 kWh. The present invention is a relationship obtained by fitting a large amount of data, for example); the on-time and energy consumption are in an exemplary proportional relationship, with a coefficient of 1 (that is, for every 1 hour of on-time, energy consumption increases by 1 kWh); the number of switches has a relatively small impact on energy consumption, with a coefficient of 0.1 (each switch consumes an additional 0.1 kWh of electricity). In order to convert these coefficients into weights, normalization processing is required. For example, the data obtained by calculation that the brightness value involves a weight of 0.4, the data obtained by calculation that the on-time involves a weight of 0.5, and the data obtained by calculation that the number of switches involves a weight of 0.1.

[0063] For other equipment, such as air conditioning systems, data features such as set temperature (range of 16-30 degrees Celsius), operating power (range of 500-2000 watts) and fan speed (range of 500-1500 rpm) also have their own data involvement weights. Set temperature has a greater impact on energy consumption. Through analysis of historical data, its coefficient is 0.5 (for every 1 degree Celsius decrease in temperature, energy consumption decreases by 0.5 kWh, for example). Operating power is directly proportional to energy consumption, with a coefficient of 1 (1 kWh consumed for every 1000 watts of power running for 1 hour). Fan speed has a smaller impact on energy consumption, with a coefficient of 0.1 (for every 100 rpm increase in speed, energy consumption increases by 0.1 kWh). After normalization, the data involvement weight of set temperature is 0.45, the data involvement weight of operating power is 0.5, and the data involvement weight of fan speed is 0.05.

[0064] For example, the global state monitoring vector of the intelligent lighting system is (vec{v}_{global-light}=[0.3, 0.2, 0.1]) (the elements of the present invention correspond to the processed state values ​​of the brightness value, the opening time and the number of switching times respectively). According to the previously determined data involvement weights, the combined calculation method is as follows: New combination value: (v_{new-light}=0.4*0.3+0.5*0.2+0.1*0.1=0.23) For the air conditioning system, the global state monitoring vector is: (vec{v}_{global-air}=[0.2, 0.3, 0.1]) (the elements of the present invention correspond to the processed state values ​​of the set temperature, the operating power and the fan speed respectively), and the combined calculation is: (v_{new-air}=0.45*0.2+0.5*0.3+0.05*0.1=0.245) After all devices are combined in this way, the first state monitoring combination vector is obtained. For example, if there are only intelligent lighting systems and air conditioning systems in the campus, the first state monitoring combination vector may be ([0.23, 0.245]).

[0065] There are different regulation requirements in the energy consumption regulation process. For example, from the perspective of energy cost control, it may be desirable to reduce overall energy consumption by 30% without affecting basic teaching needs; from the perspective of equipment life, it may be desirable to reduce the high-power operation time of the equipment. These regulation requirement information needs to be quantified.

[0066] For example, the regulation demand information from the perspective of energy cost control is to reduce the overall energy consumption by 30%, and the regulation demand information from the perspective of equipment service life is to reduce the operating time of high-power equipment (such as air conditioners) by 20%. For energy cost control requirements, they can be quantified as a vector element. For example, if the energy consumption reduction ratio is set between 0-1 (1 means 100% reduction), the element is 0.3. Similar quantification is also performed for the requirement to reduce the operating time of equipment. For example, the operating time reduction ratio of high-power equipment is between 0-1, and the present invention is 0.2. In summary, the regulation demand vector is (vec{v}_{req}=[0.3, 0.2]).

[0067] For another example, the first operating energy consumption quantization vector (vec{v}_{energy}=[0.2, 1.5]) has been obtained (for example, the energy consumption of the intelligent lighting system is 0.2 kWh, and the energy consumption of the air conditioning system is 1.5 kWh), the first state monitoring combination vector (vec{v}_{first-comb}=[0.23, 0.245]), and the global state monitoring vector corresponding to the target state monitoring data set (for example, the target device is an air conditioning system, and its global state monitoring vector is: (vec{v}_{global-air-target}=[0.2,0.3,0.1])). And, the adjustment demand vector is (vec{v}_{req}=[0.3,0.2]).

[0068] The energy consumption regulation auxiliary information can be combined by arranging these vectors in a certain order to form a new vector or matrix. For example, a matrix (M) is formed: [ M = begin{bmatrix} 0.2&1.5 0.23&0.245 0.2&0.3&0.1 0.3&0.2 end{bmatrix} ] This matrix (M) is the auxiliary information for energy consumption regulation, in which each row represents different information (the first row is the first operating energy consumption quantification vector, the second row is the first state monitoring combination vector, the third row is the global state monitoring vector corresponding to the target state monitoring data set, and the fourth row is the regulation demand vector). Through this combination method, energy consumption-related information, equipment state information, and regulation demand information are integrated together, providing a comprehensive data basis for subsequent energy consumption regulation analysis.

[0069] Furthermore, the first energy consumption regulation decision algorithm can be based on a linear programming algorithm. For example, the goal is to optimize the operating parameters of the equipment (such as the brightness and opening time of the intelligent lighting system, the set temperature and operating power of the air conditioning system, etc.) under the premise of meeting the requirements in the regulation demand vector (such as reducing energy consumption by 30% and reducing the high-power operating time of the equipment by 20%).

[0070] Assume that the brightness of the intelligent lighting system is (x_1) (the value range is 0-1000 lux), and the opening time is (x_2) (in hours); the set temperature of the air-conditioning system is (y_1) (the value range is 16-30 degrees Celsius), and the operating power is (y_2) (the value range is 500-2000 watts).

[0071] According to the energy consumption relationship, the energy consumption function of the intelligent lighting system is: (E_{light}=a_1x_1+b_1x_2); ((a_1) and (b_1) are coefficients obtained based on the previous analysis, for example (a_1=0.006) kWh / lux, (b_1=1) kWh / hour), the energy consumption function of the air-conditioning system is (E_{air}=a_2y_1+b_2y_2) (for example (a_2=-0.5) kWh / degree Celsius, (b_2=0.001) kWh / watt, the positive and negative coefficients of the present invention represent the increase or decrease relationship with energy consumption).

[0072] The regulation demand constraints include: the overall energy consumption (E=E_{light}+E_{air}) must satisfy (Eleqslant(1-0.3)(E_{light-original}+E_{air-original})) (where (E_{light-original}) and (E_{air-original}) are the original energy consumption), and the high-power operation time of the air-conditioning system (T_{air-high}) must satisfy (T_{air-high}leqslant(1-0.2)T_{air-high-original}).

[0073] At the same time, there are constraints on the operating status of the equipment. For example, the brightness of the intelligent lighting system (x_1) must meet the teaching visibility requirements (for example, the minimum brightness is 300 lux), and the set temperature of the air-conditioning system (y_1) must meet the human comfort requirements (for example, 18-26 degrees Celsius).

[0074] By solving this linear programming problem, we can obtain the optimal operating parameters of the intelligent lighting system and air-conditioning system, such as (x_1=400) lux, (x_2=1.5) hours, (y_1=22) degrees Celsius, (y_2=1000) watts. These parameter combinations are the target energy consumption regulation optimization strategy.

[0075] With this design, firstly, the global state monitoring vector is combined by determining the data involvement weight to obtain the first state monitoring combination vector, which can more reasonably integrate the impact of each device state on energy consumption, avoid equal treatment of different device state characteristics, and make the integration of device state information and energy consumption relationship more in line with the actual situation. For example, by accurately quantifying the weights of each state characteristic of the intelligent lighting system and the air-conditioning system, their different contributions to energy consumption can be accurately reflected. Secondly, energy consumption regulation auxiliary information is obtained by combining the regulation demand information, and energy consumption related, device state and regulation demand are fully integrated, providing rich and targeted data basis for energy consumption regulation decision-making. Finally, the first energy consumption regulation decision algorithm is used to generate the target energy consumption regulation optimization strategy. For example, based on the linear programming algorithm, it can accurately optimize the equipment operation parameters under the conditions of meeting multiple constraints (energy consumption reduction, equipment life protection, etc.), effectively reduce energy consumption, improve equipment operation efficiency, and realize intelligent and precise energy consumption management of campus Internet of Things equipment.

[0076] Under some preferred design ideas, the campus IoT operation status log is subjected to state vector mining based on the target state monitoring data set and each of the device state monitoring data sets to obtain a number of IoT operation status vectors, including: performing pyramid state vector mining on the campus IoT operation status log to obtain the pyramid operation status vector of the campus IoT operation status log; performing feature downsampling on the pyramid operation status vector based on the target state monitoring data set and each of the device state monitoring data sets to obtain a number of pyramid downsampling vectors; and interacting with each of the pyramid downsampling vectors to obtain a number of IoT operation status vectors.

[0077] In the next possible specific steps, each of the pyramid downsampling vectors is interacted respectively to obtain a number of IoT operation state vectors, including: for any of the pyramid downsampling vectors, the local state vectors of each feature depth in the pyramid downsampling vector are adjusted respectively to obtain a number of implicit state vectors of consistent size, and each of the implicit state vectors is interacted to obtain an interaction state vector; and each of the interaction state vectors is fully connected to obtain a number of IoT operation state vectors.

[0078] To facilitate understanding of the above technical solution, the principle of pyramid state vector mining is first introduced. When processing campus IoT operation status logs, pyramid state vector mining is an effective data processing method. It is similar to the pyramid structure in image processing, and obtains comprehensive operation status information by analyzing data at different levels.

[0079] For example, the campus IoT operation status log contains the status data of multiple devices at different time points. For example, for the smart lighting system, there is data such as the brightness and switch status of each lamp, and for the air conditioning system, there is data such as temperature setting and operating power. These data are recorded in chronological order to form a large-scale data set.

[0080] Pyramid state vector mining first processes this large-scale data set in layers. At the bottom layer, there is the most original and detailed device status data. As the layers go up, the abstraction level of the data gradually increases. For example, the bottom layer data may be the lighting brightness value every minute, the upper layer may be the average brightness value every ten minutes, and the upper layer may be the brightness trend data every hour, etc.

[0081] Through this hierarchical processing, data at different levels can capture the characteristics of equipment operation status at different time scales and abstract levels. For example, the bottom-level data can reflect the instantaneous changes in equipment status, while the high-level data can reflect the long-term trend of equipment operation.

[0082] The resulting pyramid operation state vector is a vector that integrates information at different levels, which can fully describe the operation state of campus IoT devices. For example, this vector may contain different levels of information such as the long-term trend value of the brightness of the intelligent lighting system and the power fluctuation range of the air-conditioning system, expressed in numerical form, such as [0.3 (long-term trend value of lighting brightness), 0.5 (power fluctuation range of air-conditioning), ...].

[0083] Based on the above content, for the target state monitoring data set, for example, taking the air-conditioning system as an example, if the target state monitoring data set focuses on the energy consumption-related state of the air conditioner (such as operating power, operating time, etc.), then the pyramid operating state vector is feature downsampled based on this data set.

[0084] Feature downsampling is a method of data dimensionality reduction. For example, the part of the pyramid operation state vector related to the air conditioning system contains 10 features (such as power values ​​at different time scales, operation time in different modes, etc.). Through feature downsampling, it is possible to select the three most relevant features (such as average operation power, operation time during peak hours, and total operation time) according to the requirements of the target state monitoring data set, and resample these three features to make them more in line with the requirements of subsequent analysis in terms of numerical range and data density.

[0085] For other equipment status monitoring data sets, such as the data set of intelligent lighting systems, if the focus is on the status related to lighting efficiency (such as brightness utilization, the relationship between switching frequency and energy consumption, etc.), similar feature downsampling operations are also performed. For example, the intelligent lighting system has 8 relevant features in the pyramid operation state vector. After downsampling, 2 key features (such as the average brightness utilization, the relationship between switching frequency and energy consumption) are selected and resampled.

[0086] In this way, several pyramid downsampling vectors are obtained based on the target state monitoring data set and each equipment state monitoring data set. Each pyramid downsampling vector focuses on the key features of the corresponding data set and has undergone data dimension reduction and resampling for subsequent analysis.

[0087] Furthermore, for any pyramid downsampling vector, for example, the pyramid downsampling vector of the air conditioning system obtained previously, it contains features such as average operating power, peak operating time, total operating time, etc. These features are at different feature depths in the vector.

[0088] For example, the position of the average operating power in the vector is represented as depth 1, the operating time during peak hours is represented as depth 2, and the total operating time is represented as depth 3. These local state vectors of different depths are adjusted.

[0089] Taking the average operating power as an example, if its value range is 500-2000 watts, in order to better interact with other features, it may be adjusted to a value range of 0-1, through the formula (x'=(x-min) / (max-min)) (where (x) is the original value, (x') is the adjusted value, (min) is the minimum value, and (max) is the maximum value). For example, if the average operating power is 1000 watts, then the adjusted value is ((1000-500) / (2000-500)=0.33).

[0090] Similar adjustments are made to the peak-hour running time and the total running time to obtain several implicit state vectors of the same size (same value range and data type). For example, after adjustment, the value of the implicit state vector corresponding to the peak-hour running time is 0.4, and the value of the implicit state vector corresponding to the total running time is 0.5.

[0091] Next, the implicit state vectors are interacted. An exemplary interaction method may be to connect them in a certain order. For example, the implicit state vectors corresponding to the adjusted average operating power, peak operating time, and total operating time of the air-conditioning system are connected to obtain the interactive state vector [0.33, 0.4, 0.5]. This interactive state vector integrates the information of different key features of the air-conditioning system after downsampling, and through adjustment and connection, this information can be represented and analyzed in a unified vector.

[0092] Finally, the features of each interaction state vector are fully connected. For example, for the interaction state vector [0.33, 0.4, 0.5] of the air conditioning system and the interaction state vector (for example, [0.2, 0.3]) of the intelligent lighting system, there are many ways to perform feature full connection.

[0093] One way is to directly concatenate them into a new vector to obtain [0.33, 0.4, 0.5, 0.2, 0.3], which is the IoT operation state vector. This vector comprehensively contains the key state information of different devices (the present invention is the air conditioning system and the intelligent lighting system) after pyramid downsampling, local state vector adjustment, interaction and other operations, and can provide a comprehensive data basis for subsequent energy consumption analysis, equipment status evaluation, etc.

[0094] With this design, firstly, the pyramid operation state vector is obtained through pyramid state vector mining, which can analyze the operation status of campus IoT devices from different levels and capture a variety of information from instantaneous changes to long-term trends, such as the brightness change level information of the intelligent lighting system and the power fluctuation level information of the air conditioning system. Secondly, the pyramid downsampling vector is obtained by feature downsampling based on different data sets. By focusing on key features and performing data dimension reduction and resampling, the amount of data processing is reduced, and the important information related to the status of specific devices is highlighted, which improves the efficiency and accuracy of subsequent analysis. Finally, the interactive operation of the pyramid downsampling vector, including local state vector adjustment, implicit state vector interaction and feature full connection to obtain the IoT operation state vector, enables the key state information of different devices to be effectively integrated and fully represented in a unified vector, which is convenient for energy consumption analysis, equipment status evaluation and other operations, and helps to achieve refined management and optimized operation of campus IoT devices.

[0095] Under some technical schemes, the device status monitoring data set corresponding to each preliminary screening IoT device monitoring behavior event in the campus IoT operation status log is obtained, and a target status monitoring data set is determined among several of the device status monitoring data sets, including: obtaining the campus IoT operation status log and index information of the campus IoT operation status log, wherein the campus IoT operation status log includes several preliminary screening IoT device monitoring behavior events, and the index information is used to indicate a target IoT device monitoring behavior event with the highest priority among the several preliminary screening IoT device monitoring behavior events; disassembling the campus IoT operation status log to obtain device status monitoring data sets corresponding to each of the preliminary screening IoT device monitoring behavior events, and determining the target status monitoring data set among each of the device status monitoring data sets based on the index information.

[0096] In detail, determining the target state monitoring data set in each of the device state monitoring data sets based on the index information includes: when the index information is a first-category index, determining the target state monitoring data set in each of the device state monitoring data sets based on the relative distribution characteristics between the first-category index and each of the device state monitoring data sets; or, when the index information is a second-category index, determining the target state monitoring data set in each of the device state monitoring data sets based on the association between the second-category index and the key-value characteristics of each of the device state monitoring data sets; or, when the index information is a third-category index, determining the target state monitoring data set in each of the device state monitoring data sets based on the association between the third-category index and each of the device state monitoring data sets.

[0097] In the present invention, the detailed implementation steps of the above technical solution are as follows.

[0098] 1. Obtain campus IoT operation status log and index information and disassemble the log to obtain the device status monitoring data set 1) Obtain campus IoT operation status log and index information The campus IoT operation status log is a detailed record of the operation of the entire campus IoT equipment. It contains many preliminary screening IoT equipment monitoring behavior events. For example, in the campus IoT environment, the equipment includes equipment in smart classrooms (such as projectors, smart lighting systems, air conditioners, etc.), environmental sensors (temperature, humidity, light sensors, etc.), etc. Various monitoring behavior events of these devices during operation, such as the opening and closing of projectors, brightness adjustment, changes in light brightness of smart lighting systems, switch operations, changes in temperature settings of air conditioners, switching of operation modes, etc., will be recorded in the campus IoT operation status log.

[0099] At the same time, there is also index information related to the log. This index information plays a key role. It can indicate the target IoT device monitoring behavior event with the highest priority among several preliminary screening IoT device monitoring behavior events. For example, the index information may determine the priority based on factors such as the importance of the device and the current urgency. For teaching activities, projector-related events when important teaching demonstrations are in progress may have a higher priority; in hot weather, events related to refrigeration failures in the air-conditioning system will also have a high priority.

[0100] 2) Disassemble the campus IoT operation status log to obtain the device status monitoring data set The purpose of disassembling the campus IoT operation status log is to obtain the device status monitoring data set corresponding to each initially screened IoT device monitoring behavior event. Taking the smart lighting system as an example, its device status monitoring data set may contain multiple data elements. For example, light brightness value (the value range may be 0-1000 lux), turn-on time (in minutes), number of switches, etc. For example, in a certain time period, the light brightness value is 300 lux, the turn-on time is 60 minutes, and the number of switches is 3 times. Then the device status monitoring data set of this smart lighting system in this time period is {brightness: 300 lux, duration: 60 minutes, number of switches: 3 times}.

[0101] For an air conditioning system, the equipment status monitoring data set may include the set temperature (e.g., the range is 16-30 degrees Celsius), the operating power (e.g., the range is 500-2000 watts), the operating mode (cooling, heating, ventilation, etc., which can be represented by numerical values, such as 0 for cooling, 1 for heating, and 2 for ventilation), etc. If the air conditioner set temperature is 22 degrees Celsius, the operating power is 1000 watts, and the operating mode is cooling, then its equipment status monitoring data set is {temperature: 22 degrees Celsius, power: 1000 watts, mode: 0}.

[0102] 2. Determining the target state monitoring dataset based on different types of indexes 1) When the index information is the first type of index There is a relationship between the first-class index and the relative distribution characteristics of each device status monitoring data set. For example, the first-class index determines the priority based on the frequency of use of the device. For example, in the campus Internet of Things, the frequency of use of the device can be determined by counting the number of times it is used within a certain time range (such as one day, one week, or one month).

[0103] Taking the smart lighting system and projector as an example, if the smart lighting system is used 100 times in a week and the projector is used 30 times, the usage frequency can be normalized. For example, if the maximum usage frequency of all devices on campus is 200 times, the normalized usage frequency of the smart lighting system is 0.5, and the normalized usage frequency of the projector is 0.15.

[0104] Each device status monitoring data set has a relative distribution characteristic with this first type of index based on usage frequency. For example, some data elements in the device status monitoring data set may have a direct or indirect relationship with the usage frequency. For smart lighting systems, the on-time and on-off times in its device status monitoring data set may have a strong correlation with the usage frequency. If the on-time is long and the on-off times are high, it usually means that the usage frequency is high.

[0105] The target state monitoring data set is determined based on this relative distribution feature. For example, if there are multiple device state monitoring data sets, and through analysis it is found that the distribution characteristics of the data elements related to the frequency of use in the device state monitoring data set of the intelligent lighting system best match the first type of index (based on the priority of the frequency of use), then the device state monitoring data set of the intelligent lighting system will be determined as the target state monitoring data set.

[0106] 2) When the index information is the second type of index The second type of index is associated with the key value features of each device status monitoring data set. For example, the second type of index is to determine the priority based on the failure risk level of the device. The failure risk level may be determined based on factors such as the historical failure data of the device and the current operating status.

[0107] Taking the air conditioning system and environmental sensors as examples, the key features of the equipment status monitoring data set of the air conditioning system may include temperature setting, operating power, etc. The key features of the equipment status monitoring data set of the environmental sensor may include the collected data value, collection frequency, etc.

[0108] For example, the air conditioning system is judged to have a high fault risk level based on its historical fault data and current operating status (such as large operating power fluctuations, frequent set temperature adjustments, etc.). The key features of its equipment status monitoring data set (such as operating power fluctuations, temperature setting adjustment range, etc.) are associated with the second type of index based on the fault risk level. If this correlation is the strongest among all devices, for example, the operating power fluctuation of the air conditioning system exceeds 30% of the normal range (such as the normal fluctuation range is 10%), and the temperature setting adjustment range exceeds 5 degrees Celsius in a short period of time (such as the normal adjustment range is 1-2 degrees Celsius), and other devices do not have similar obvious key feature changes related to the fault risk level, then the equipment status monitoring data set of the air conditioning system will be determined as the target status monitoring data set based on this correlation.

[0109] 3) When the index information is the third type of index The third type of index is associated with each device status monitoring data set. For example, the third type of index determines the priority based on the impact of the device on the overall campus IoT operation.

[0110] Consider the smart lighting system, air conditioning system and network server in the campus IoT (the network server is also part of the IoT device and is responsible for data transmission and other functions). The smart lighting system mainly affects the lighting environment of the classroom, the air conditioning system affects the indoor temperature environment, and once a problem occurs in the network server, it may affect the data transmission and interaction between the IoT devices in the entire campus.

[0111] The device status monitoring data set of a network server may contain data elements such as network bandwidth utilization and data transmission delay. If the network bandwidth utilization of a network server reaches 90% (for example, the normal utilization is 60%-70%), and the data transmission delay increases by 50% (for example, the normal delay fluctuation range is 10%-20%), this indicates that its device status monitoring data set has a strong correlation with the third type of index based on the degree of impact on the overall campus IoT operation. In contrast, the impact of the smart lighting system and air conditioning system on the overall IoT operation is not as great as that of the network server in the current state, so the device status monitoring data set of the network server will be determined as the target status monitoring data set based on this correlation.

[0112] In this way, by obtaining the campus IoT operation status log and its index information, the priority of the device monitoring behavior events can be clarified, providing a basis for subsequent accurate processing. For example, the priority is determined based on factors such as the frequency of device use, the level of fault risk, or the degree of impact on the overall operation, such as determining its high priority based on the network bandwidth utilization and data transmission delay of the network server. Secondly, the log is disassembled to obtain the device status monitoring data set, which can carefully analyze the operating status of each device, such as the brightness, duration and other data elements of the intelligent lighting system. Finally, the target status monitoring data set is determined based on different types of indexes, and the data set of key equipment can be accurately found according to the specific index type (frequency of use, fault risk, overall impact, etc.), which helps to concentrate resources to handle key equipment-related matters and improve the efficiency and pertinence of campus IoT equipment management.

[0113] In actual application, the target energy consumption regulation optimization strategy can be generated by a first energy consumption regulation decision algorithm. Based on this, the debugging steps of the first energy consumption regulation decision algorithm include the following contents: (1) obtaining a historical IoT operation status log and a second monitoring data set sample corresponding to a historical IoT device monitoring behavior event in the historical IoT operation status log, and disassembling the historical IoT operation status log to obtain a first monitoring data set sample corresponding to each frequent IoT device behavior event in the historical IoT operation status log, wherein the historical IoT device monitoring behavior event is one IoT device monitoring behavior event among a plurality of frequent IoT device behavior events; (2) extracting the second operation energy consumption quantization vector of the historical IoT operation status log, mining the state vectors of the historical IoT operation status log according to the second monitoring data set samples and each of the first monitoring data set samples, obtaining a plurality of historical operation state vectors, and extracting the historical monitoring attention vectors corresponding to the second monitoring data set samples and each of the first monitoring data set samples; (3) Combining the second operation energy consumption quantization vector, the historical operation state vector, and the historical monitoring attention vector, and inputting them into the first energy consumption regulation decision algorithm for energy consumption regulation analysis to generate an energy consumption regulation heat map, wherein the energy consumption regulation heat map is used to determine the energy consumption regulation optimization strategy of the historical IoT device monitoring behavior event; (4) Obtaining a first energy consumption regulation optimization strategy for a historical energy consumption regulation task configured by the historical IoT operation status log, determining an algorithm training error based on the energy consumption regulation heat map and the first energy consumption regulation optimization strategy, and debugging the first energy consumption regulation decision algorithm based on the algorithm training error, wherein the historical energy consumption regulation task is used to indicate the historical IoT device monitoring behavior events.

[0114] Exemplarily, the debugging step of the first energy consumption regulation decision algorithm is implemented as follows.

[0115] 1. Obtain historical data and disassemble logs to obtain monitoring data set samples 1) Obtain historical IoT operation status logs and second monitoring data set samples and disassemble the logs The historical IoT operation status log is a record of the operation status of campus IoT devices over a period of time. It contains rich information, such as various monitoring behavior events of the device. For example, for the devices in the smart classroom, it includes the projector's power-on time, the number of brightness adjustments, the light brightness changes of the smart lighting system, the switch time, etc., as well as the temperature setting and operating power of the air conditioning system.

[0116] A second monitoring data set sample corresponding to the historical IoT device monitoring behavior event is obtained from the historical IoT operation status log. Taking the air conditioning system as an example, this sample may include a set of data within a specific time period, such as the set temperature (the value range is 16-30 degrees Celsius, for example, the set temperature value within a certain period of time is 22 degrees Celsius), operating power (the value range is 500-2000 watts, for example, the power at a certain moment is 1000 watts), operating mode (such as cooling, heating, ventilation, the present invention is a cooling mode, for example), etc.

[0117] The historical IoT operation status log is disassembled to obtain the first monitoring data set sample corresponding to each frequent IoT device behavior event. Frequent IoT device behavior events are device behavior events that frequently appear in historical records. For example, the frequent on-off behavior of an intelligent lighting system, its first monitoring data set sample may include the time interval between each on-off (for example, 30 minutes), the brightness value when each on-off is turned on (for example, 500 lux), etc. A historical IoT device monitoring behavior event is a specific event among several frequent IoT device behavior events, such as the monitoring event of the cooling behavior of an air-conditioning system within a specific time period.

[0118] 2. Extracting the quantified vector of operating energy consumption and related vectors 1) Extract the energy consumption quantization vector of the second operation Extract the second operation energy consumption quantification vector from the historical IoT operation status log. Take the smart lighting system, air conditioning system and projector in the campus as an example. For example, the power of the smart lighting system is 100 watts and the operation time is 2 hours. According to the energy consumption calculation formula (E=P*t) ((E) is energy consumption, (P) is power, and (t) is time), its energy consumption is 100 watts × 2 hours = 0.2 kWh; the power of the air conditioning system is 1500 watts, and the energy consumption is 1.5 kWh when it runs for 1 hour; the power of the projector is 300 watts, and the energy consumption is 0.45 kWh when it runs for 1.5 hours. Then the second operation energy consumption quantification vector is [0.2, 1.5, 0.45], which represents the energy consumption of different devices during the historical operation process.

[0119] 2) State vector mining and extraction of historical monitoring attention vectors The state vectors of the historical IoT operation status logs are mined based on the second monitoring data set samples and the first monitoring data set samples. Taking the second monitoring data set sample of the air-conditioning system as an example, if it contains three data elements: set temperature, operating power, and operating mode, these data can be quantified to form a state vector. For example, if the set temperature is 22 degrees Celsius, it is quantized to 0.4 (by mapping the actual temperature value to the 0-1 interval, the calculation method is ((22-16) / (30-16))), the operating power is 1000 watts, quantized to 0.33 (the calculation method is ((1000-500) / (2000-500))), and the operating mode is cooling, represented by 0, then the state vector of the air-conditioning system is [0.4, 0.33, 0].

[0120] Similar state vector mining is also performed for each first monitoring data set sample. For example, if the first monitoring data set sample corresponding to the frequent switching behavior of the intelligent lighting system contains two data elements, the switching time interval and the turn-on brightness value, the state vector is obtained after quantization. For example, if the switching time interval is 30 minutes, it is quantized to 0.5 (for example, the maximum time interval is 60 minutes, and the calculation method is (30 / 60)), and the turn-on brightness value is 500 lux, it is quantized to 0.5 (for example, the brightness value range is 0-1000 lux, and the calculation method is (500 / 1000)), then the state vector of the intelligent lighting system is [0.5, 0.5].

[0121] At the same time, the historical monitoring attention vectors corresponding to the second monitoring data set samples and each first monitoring data set sample are refined. For the second monitoring data set sample of the air-conditioning system, if the current focus is on the impact of operating power and set temperature on energy consumption, then the attention weight for operating power is 0.6, the attention weight for set temperature is 0.4, and the attention weight for operating mode is 0.1 (relatively small, because the impact of operating mode in energy consumption analysis is relatively weak), then the historical monitoring attention vector is [0.4, 0.6, 0.1]. For the first monitoring data set sample of the intelligent lighting system, if the focus is on the impact of the switching time interval on energy consumption is greater than the turn-on brightness value, for example, the attention weight for the switching time interval is 0.7, and the attention weight for the turn-on brightness value is 0.3, then the historical monitoring attention vector is [0.7, 0.3].

[0122] 3. Combining data input algorithms to generate energy consumption regulation heat maps 1) Combine data and feed into the algorithm The second operation energy consumption quantization vector, the historical operation state vector, and the historical monitoring attention vector are combined and input into the first energy consumption regulation decision algorithm for energy consumption regulation analysis. For example, the second operation energy consumption quantization vector [0.2, 1.5, 0.45] obtained previously, the historical operation state vector [0.4, 0.33, 0] and the historical monitoring attention vector [0.4, 0.6, 0.1] of the air conditioning system, and the historical operation state vector [0.5, 0.5] and the historical monitoring attention vector [0.7, 0.3] of the intelligent lighting system are combined according to certain rules. One possible combination method is to splice them into a large vector or matrix.

[0123] For example, combine these data into a matrix (M): [ M = begin{bmatrix} 0.2&1.5&0.45 0.4&0.33&0 0.4&0.6&0.1 0.5&0.5 0.7&0.3 end{bmatrix} ] This matrix (M) is input into the first energy consumption regulation decision algorithm. The first energy consumption regulation decision algorithm can be an algorithm based on machine learning, such as a neural network algorithm. In the neural network, the input layer receives the combined matrix data, calculates and processes it through the neurons of the hidden layer, and finally outputs the result for energy consumption regulation analysis.

[0124] 2) Generate energy consumption regulation heat map An energy consumption regulation heat map is generated by calculating the first energy consumption regulation decision algorithm. The energy consumption regulation heat map is a visual representation method used to determine the energy consumption regulation optimization strategy for historical IoT device monitoring behavior events. For example, in the heat map, the horizontal axis may represent different device state parameters (such as the set temperature of the air conditioning system, the brightness of the intelligent lighting system, etc.), and the vertical axis may represent different time intervals or device operation modes. The depth of the color in the heat map indicates the level of energy consumption or the potential for energy consumption regulation. If the color in the heat map is darker at a certain set temperature (such as the set temperature of the air conditioning system is 20 degrees Celsius), it means that the energy consumption is higher at this temperature or there is a greater potential for energy consumption regulation; if the color in the heat map is lighter at a certain brightness value (such as the brightness of the intelligent lighting system is 300 lux), it means that the energy consumption is lower at this brightness or the need for energy consumption regulation is smaller.

[0125] 4. Determine algorithm training error and debug the algorithm 1) Obtain the first energy consumption regulation optimization strategy for historical energy consumption regulation tasks Get the first energy consumption adjustment optimization strategy of the configured historical energy consumption adjustment task from the historical IoT operation status log. For example, for an air conditioning system, the historical energy consumption adjustment task may be to reduce energy consumption while meeting indoor comfort. The first energy consumption adjustment optimization strategy historically adopted may be to adjust the set temperature from 22 degrees Celsius to 24 degrees Celsius (based on experience or previous analysis, it is determined that this adjustment can reduce energy consumption), and adjust the operation mode of the air conditioner (such as from continuous cooling mode to intermittent cooling mode).

[0126] 2) Determine the algorithm training error The algorithm training error is determined based on the energy consumption adjustment heat map and the first energy consumption adjustment optimization strategy. For example, according to the energy consumption adjustment heat map, energy consumption adjustment suggestions under the same conditions (equipment status, environment and other conditions are the same) are obtained. For example, according to the heat map, it is more energy-efficient to adjust the air conditioner set temperature to 23 degrees Celsius than to adjust it to 24 degrees Celsius in history. Then the algorithm training error can be determined by calculating the difference between the actual historical energy consumption (the energy consumption after the execution of the first energy consumption adjustment optimization strategy in history) and the energy consumption recommended by the energy consumption adjustment heat map. For example, if the actual historical energy consumption is 1.2 kWh and the energy consumption recommended by the heat map is 1.0 kWh, then the algorithm training error can be ((1.2-1.0) / 1.2approx0.167).

[0127] 3) Debug the first energy consumption regulation decision algorithm The first energy consumption regulation decision algorithm is debugged based on this algorithm training error. If the algorithm training error is large, it means that there is a large deviation between the algorithm's prediction result and the actual optimal energy consumption regulation strategy. The error can be reduced by adjusting the parameters in the algorithm. For example, if the first energy consumption regulation decision algorithm is a neural network algorithm, the connection weights between neurons can be adjusted. If the error is caused by unreasonable weight allocation of equipment status parameters (such as the set temperature and operating power of the air-conditioning system), the weights of these parameters in the algorithm can be increased or decreased. By continuously adjusting the parameters and re-entering the data for calculation, the algorithm training error is reduced to an acceptable range.

[0128] Therefore, by obtaining historical data and disassembling it to obtain monitoring data set samples, we can deeply explore the equipment status information in historical operation, such as the various parameters of air conditioning and lighting systems. Secondly, by refining the operation energy consumption quantification vector and related vectors, we can accurately quantify energy consumption and pay attention to key status factors, such as determining the energy consumption of different equipment and the attention weights of different parameters. Furthermore, the combined data input algorithm generates an energy consumption regulation heat map, which presents the energy consumption regulation potential in a visual way, making it easier to determine the optimization strategy. Finally, determining the algorithm training error and debugging the algorithm can improve the accuracy of the first energy consumption regulation decision algorithm, so that it can more accurately give energy consumption regulation optimization strategies.

[0129] In some optional embodiments, the historical IoT operation status log, the second monitoring data set sample, and the first energy consumption regulation optimization strategy are all obtained from a training sample pool. Before obtaining the historical IoT operation status log and the second monitoring data set sample corresponding to the historical IoT device monitoring behavior event in the historical IoT operation status log, the smart campus information data analysis method also includes: obtaining a number of IoT operation status training logs and a state monitoring training data set corresponding to each of the IoT operation status training logs, and determining the log mining results corresponding to each of the IoT operation status training logs according to each of the IoT operation status training logs and the corresponding state monitoring training data set, wherein the state monitoring training data set is used to characterize the selected IoT device monitoring behavior event in the corresponding IoT operation status training log obtained by mining; obtaining a number of preliminary screening energy consumption regulation tasks, according to each of the IoT operation status training logs and the corresponding state monitoring training data set. The log mining results are used to determine the associated energy consumption adjustment tasks corresponding to each of the IoT operation status training logs in several of the initial screening energy consumption adjustment tasks; the prior state data sets of each of the IoT operation status training logs are determined based on each of the IoT operation status training logs and the corresponding state monitoring training data sets, wherein the prior state data sets are used to indicate the corresponding selected IoT device monitoring behavior events; each of the IoT operation status training logs, the corresponding prior state data sets and the corresponding associated energy consumption adjustment tasks are synchronously imported into the training sample pool, wherein the historical IoT operation status logs are obtained from several of the IoT operation status training logs, the second monitoring data set sample is the prior state data set corresponding to the historical IoT operation status log, and the historical energy consumption adjustment task is the associated energy consumption adjustment task configured for the historical IoT operation status log.

[0130] In actual application, the exemplary implementation of the above embodiment is as follows.

[0131] 1. Obtain IoT operation status training logs and related data sets and determine log mining results 1) Obtain IoT operation status training log and status monitoring training data set In the IoT environment of smart campuses, there are numerous IoT operation status training logs. These logs record the operation status information of various IoT devices on campus. For example, for devices in smart classrooms, the IoT operation status training logs will include the projector's startup time, usage time, brightness adjustment, the smart lighting system's light switch time, brightness change value, the air conditioning system's temperature setting, operating power fluctuations, and other information.

[0132] At the same time, the state monitoring training data set corresponding to each IoT operation status training log is obtained. Taking the air conditioning system as an example, the state monitoring training data set may contain data used to characterize the selected IoT device monitoring behavior events mined in the IoT operation status training log. For example, if the selected monitoring behavior events are related to the air conditioning cooling effect, then the state monitoring training data set may include the air conditioning set temperature (the value range is 16-30 degrees Celsius, such as the set temperature is 22 degrees Celsius), the indoor and outdoor temperature difference (for example, the indoor temperature is 22 degrees Celsius, the outdoor temperature is 30 degrees Celsius, and the temperature difference is 8 degrees Celsius), the evaporator temperature (for example, 10 degrees Celsius) and other data elements.

[0133] 2) Determine the log mining results According to each IoT operation status training log and the corresponding status monitoring training data set, the log mining results corresponding to each IoT operation status training log are determined respectively. For example, for the IoT operation status training log containing air conditioning system information, the temperature data, power data, etc. are mined and analyzed. If a mining algorithm based on statistical analysis is adopted, the average value, standard deviation of temperature, fluctuation range of power and other statistics are calculated. For example, within a period of time, the average value of the air conditioner set temperature is 22 degrees Celsius, the standard deviation is 1 degree Celsius, and the fluctuation range of the operating power is 500-1000 watts (fluctuation range is 500 watts). These statistics and related data relationships constitute the log mining results of the IoT operation status training log.

[0134] (II) Obtaining the initial screening energy consumption adjustment tasks and determining the associated energy consumption adjustment tasks 1) Obtain the initial screening energy consumption adjustment task In the energy consumption management of campus IoT devices, there are several preliminary screening energy consumption adjustment tasks. These tasks may be based on different goal settings, such as reducing overall energy consumption, optimizing the energy efficiency of specific equipment, etc. For example, one preliminary screening energy consumption adjustment task may be to reduce the overall energy consumption of the smart classroom by 20% without affecting the comfort of the teaching environment; another preliminary screening energy consumption adjustment task may be to increase the energy efficiency of the air conditioning system by 30% (energy efficiency can be measured by the ratio of actual cooling capacity to power consumption).

[0135] 2) Determine the associated energy consumption regulation tasks According to the mining results of each log, the associated energy consumption adjustment tasks corresponding to each IoT operation status training log are determined in several preliminary screening energy consumption adjustment tasks. Taking the IoT operation status training log containing air conditioning system information as an example, if the log mining results show that the operating power of the air conditioner fluctuates greatly and the set temperature is high, then the energy consumption adjustment task associated with it may be to reduce the operating power of the air conditioner to improve energy efficiency while meeting the indoor temperature comfort level (for example, the human body comfort temperature range is 22-26 degrees Celsius). If another IoT operation status training log shows that the light brightness of the intelligent lighting system is at a high level for a long time and the switching frequency is low, its associated energy consumption adjustment task may be to adjust the light brightness according to the actual lighting needs of the classroom to reduce energy consumption.

[0136] 3. Determine the prior state data set and import it into the training sample pool 1) Determine the prior state data set According to each IoT operation state training log and the corresponding state monitoring training data set, the prior state data set of each IoT operation state training log is determined respectively. Taking the IoT operation state training log of the intelligent lighting system as an example, the prior state data set is used to indicate the corresponding selected IoT device monitoring behavior event. For example, if the selected monitoring behavior event is the relationship between the brightness adjustment and energy consumption of the lighting system, then the prior state data set may include data elements such as the initial brightness of the light (for example, 800 lux), the frequency of brightness adjustment (for example, once per hour), the brightness change amount for each adjustment (for example, 100 lux), and the current energy consumption (for example, 0.5 kWh / hour).

[0137] 2) Import training sample pool Import each IoT operation state training log, the corresponding prior state data set, and the corresponding associated energy consumption adjustment task into the training sample pool simultaneously. For example, for a set of data containing air conditioning system information, import the IoT operation state training log (including various operation state data of the air conditioner), the corresponding prior state data set (such as the set temperature of the air conditioner, the indoor and outdoor temperature difference data, etc.) and the associated energy consumption adjustment task (such as reducing the energy consumption of the air conditioner while satisfying the comfort level) into the training sample pool together. Similarly, the relevant data of the intelligent lighting system is also imported in this way.

[0138] In this way, in subsequent operations, the historical IoT operation status log is obtained from several IoT operation status training logs, the second monitoring data set sample is the prior state data set corresponding to the historical IoT operation status log, and the historical energy consumption adjustment task is the associated energy consumption adjustment task configured for the historical IoT operation status log. For example, if a previously imported IoT operation status training log containing an air-conditioning system is selected as the historical IoT operation status log, then the corresponding prior state data set becomes the second monitoring data set sample, and the associated energy consumption adjustment task becomes the historical energy consumption adjustment task.

[0139] In this way, by obtaining the IoT operation status training log and related data sets and determining the log mining results, the operation status of the equipment can be deeply analyzed, such as obtaining a comprehensive operation status through statistical analysis of air conditioning temperature and power data. By obtaining the initial screening energy consumption adjustment tasks and determining the related energy consumption adjustment tasks, the energy consumption adjustment tasks can be accurately matched according to the actual operation status of the equipment, such as determining the appropriate energy consumption adjustment tasks according to the fluctuation of air conditioning operation power. Determine the prior state data set and import it into the training sample pool, so that subsequent operations can obtain complete and relevant data from the training sample pool, which provides an accurate data basis for operations such as energy consumption adjustment analysis, and improves the reliability and effectiveness of the smart campus information data analysis method.

[0140] In the next step, the prior state data sets of each of the IoT operation state training logs and the corresponding state monitoring training data sets are determined respectively, including: inputting each of the state monitoring training data sets into the second energy consumption regulation decision algorithm for energy consumption regulation analysis, and generating optimization mode labels corresponding to each of the IoT operation state training logs; mining IoT device monitoring behavior events based on each of the IoT operation state training logs and the corresponding optimization mode labels, and generating event location window labels corresponding to each of the IoT operation state training logs, wherein the event location window label is used to indicate the corresponding selected IoT device monitoring behavior event; inputting each of the IoT operation state training logs and the corresponding event location window label into the first state data recognition algorithm for processing, and generating prior state data sets corresponding to each of the IoT operation state training logs.

[0141] In addition, after each of the IoT operation status training logs and the corresponding event location window labels are input into the first state data recognition algorithm for processing and a priori state data set corresponding to each of the IoT operation status training logs is generated, the smart campus information data analysis method also includes: obtaining the benchmark energy consumption optimization expected features corresponding to each of the associated energy consumption adjustment tasks, wherein the benchmark energy consumption optimization expected features are used to indicate the energy consumption adjustment trend of the benchmark energy consumption adjustment task, and the feature depth of the benchmark energy consumption adjustment task in the semantic feature space is higher than the feature depth of the associated energy consumption adjustment task in the semantic feature space; each of the IoT operation status training logs and the corresponding benchmark energy consumption optimization expected features are input into the second state data recognition algorithm for processing and a benchmark state data set corresponding to each of the IoT operation status training logs is generated; determining the correlation between each of the priori state data sets and the corresponding benchmark state data set to obtain the target correlation corresponding to each of the priori state data sets; when the target correlation is less than the set correlation, filtering out the IoT operation status training log corresponding to the target correlation.

[0142] In the data processing of IoT devices in smart campuses, in order to determine the prior state data set of each IoT operation state training log, each state monitoring training data set is first input into the second energy consumption regulation decision algorithm for energy consumption regulation analysis. For example, for the smart lighting system in the campus, its state monitoring training data set may include light brightness (value range is 0-1000 lux, for example, 500 lux), opening time (in hours, for example, 2 hours), number of switches (for example, 3 times) and other data.

[0143] The second energy consumption regulation decision algorithm can be an algorithm based on machine learning, such as a decision tree algorithm. The decision tree performs analysis based on the input state monitoring training data set. Taking the intelligent lighting system as an example, if the light brightness is high and the on-time is long, the decision tree may generate an optimization mode label of "high energy consumption mode"; if the light brightness is moderate and the number of switches is large, an optimization mode label of "medium energy consumption fluctuation mode" may be generated. For the state monitoring training data set of the air-conditioning system, which contains data such as set temperature (for example, 22 degrees Celsius), operating power (for example, 1000 watts), and operating mode (for example, cooling), the decision tree may generate an optimization mode label of "normal cooling energy consumption mode" based on these data.

[0144] After obtaining the optimization mode labels corresponding to each IoT operation status training log, IoT device monitoring behavior event mining is performed based on each IoT operation status training log and the corresponding optimization mode labels. For example, for the smart lighting system, if the optimization mode label is "high energy consumption mode", events related to high energy consumption are mined in the IoT operation status training log. For example, if it is found in the log that the light continues to maintain high brightness during a certain time period (such as 7-9 pm), this time period can be used as an event positioning window label. For the air-conditioning system, if the optimization mode label is "normal cooling energy consumption mode", events related to normal cooling energy consumption are mined in the IoT operation status training log. For example, if it is found that the air conditioner is stably running near the set temperature during certain time periods during the day (such as 10-12 am), this time period becomes an event positioning window label.

[0145] Further, each IoT operation state training log and the corresponding event location window label are respectively input into the first state data recognition algorithm for processing, and a priori state data set corresponding to each IoT operation state training log is generated. The first state data recognition algorithm can be an algorithm based on a neural network. Taking the intelligent lighting system as an example, the IoT operation state training log containing data such as light brightness, opening time, number of switches, and the corresponding event location window label (7-9 pm) are input into the neural network. The neural network outputs the priori state data set by learning the relationship between these data. For example, data such as average brightness and average energy consumption within this event location window may be output as elements of the priori state data set. For the air-conditioning system, the IoT operation state training log containing data such as set temperature, operating power, and operating mode and the corresponding event location window label (10-12 am) are input, and the neural network may output data such as average operating power and temperature fluctuation range within this period as the priori state data set.

[0146] Next, after determining the prior state data set, the baseline energy consumption optimization expected features corresponding to each associated energy consumption adjustment task are obtained. The baseline energy consumption optimization expected features are used to indicate the energy consumption adjustment trend of the baseline energy consumption adjustment task, and the feature depth of the baseline energy consumption adjustment task in the semantic feature space is higher than the feature depth of the associated energy consumption adjustment task in the semantic feature space. For example, the associated energy consumption adjustment task for the intelligent lighting system may be to reduce energy consumption in a specific classroom, while the baseline energy consumption adjustment task may be to optimize the energy efficiency of the lighting system across the entire campus.

[0147] For the benchmark energy consumption adjustment task, if the goal is to reduce the overall energy consumption of the campus lighting system by 30%, this goal reflects the energy consumption adjustment trend in the expected characteristics of the benchmark energy consumption optimization. Specifically, it may involve comprehensive consideration of the energy consumption of the lighting system in different areas and at different times, such as reducing the use of lighting in areas with sufficient light during the day in the teaching building, and adjusting the lighting brightness in places such as the library according to the flow of people.

[0148] Then, each IoT operation state training log and the corresponding benchmark energy consumption optimization expected features are input into the second state data recognition algorithm for processing, and a benchmark state data set corresponding to each IoT operation state training log is generated. The second state data recognition algorithm can also be based on a machine learning algorithm, such as a support vector machine algorithm. Taking the intelligent lighting system as an example, the IoT operation state training log (including light brightness, on-time, number of switches and other data) and the benchmark energy consumption optimization expected features (such as a 30% reduction in overall campus lighting energy consumption) are input into the support vector machine. The support vector machine generates a benchmark state data set based on these data, which may include data such as ideal brightness values ​​and maximum on-time in different areas to meet the overall campus energy consumption reduction target.

[0149] Furthermore, the correlation between each prior state data set and the corresponding reference state data set is determined to obtain the target correlation corresponding to each prior state data set. For example, a correlation coefficient calculation method such as the Pearson correlation coefficient can be used. For the intelligent lighting system, the Pearson correlation coefficient between the prior state data set (such as the average brightness and average energy consumption of a specific classroom) and the reference state data set (such as the ideal brightness value and maximum opening time under the overall energy consumption reduction target of the campus) is calculated.

[0150] When the target relevance is less than the set relevance, the IoT operation status training log corresponding to the target relevance is filtered out. For example, if the relevance is set to 0.5, and the Pearson correlation coefficient between the prior state data set of the smart lighting system in a classroom and the overall baseline state data set of the campus is 0.3, then the IoT operation status training log corresponding to this classroom will be filtered out. This helps to remove data with low relevance to the overall energy consumption optimization goal and improve the accuracy and effectiveness of subsequent data analysis.

[0151] It can be seen that by determining the prior state data set through the second energy consumption regulation decision algorithm, the IoT device monitoring behavior event mining and the first state data identification algorithm, the relationship between the equipment operation status and energy consumption can be deeply explored. For example, the intelligent lighting system obtains energy consumption related data for a specific period of time through multi-algorithm processing. Secondly, obtaining the expected characteristics of the baseline energy consumption optimization and generating a baseline state data set will help to examine the energy consumption regulation task from a more macro perspective, such as the ideal state data under the overall lighting energy consumption optimization goal of the campus. Finally, determining the correlation and filtering the logs can remove data with low correlation with the overall goal, improve data quality, provide a more accurate data basis for smart campus information analysis, and improve the accuracy of energy consumption regulation and other analyses.

[0152] In some independent embodiments, before synchronously importing each of the IoT operation status training logs, the corresponding prior state data sets and the corresponding associated energy consumption regulation tasks into the training sample pool, the smart campus information data analysis method also includes: determining the number of state-involved feature blocks in each of the prior state data sets, and obtaining the number of feature blocks corresponding to each of the prior state data sets; when the number of feature blocks is greater than the set number, filtering out the IoT operation status training logs corresponding to the number of feature blocks.

[0153] For the above-mentioned independent embodiments, the detailed implementation steps are as follows.

[0154] 1. Determine the number of feature blocks in the prior state data set 1) The composition of the prior state data set and the concept of feature blocks In the analysis of smart campus information data, the prior state data set is obtained through a series of complex algorithm processing, which contains important information related to the monitoring behavior events of IoT devices. The prior state data set is composed of multiple different types of data elements, which can be divided into different feature blocks according to their inherent logical relationships or data characteristics.

[0155] For example, the prior state data set for the intelligent lighting system in the campus may include light brightness data, light on time data, switching frequency data in different time periods, etc. Among them, the light brightness data can be regarded as a feature block, because it is a relatively independent data set, reflecting an important state feature of the lighting system; the light on time data can also be regarded as a separate feature block, which describes another aspect of the operating state of the lighting equipment; the switching frequency data in different time periods also constitute a feature block, which reflects the time regularity characteristics of the lighting system.

[0156] 2) The process of determining the number of feature blocks Determine the number of feature blocks involved in the state in each prior state data set, and obtain the number of feature blocks corresponding to each prior state data set. Taking the intelligent lighting system as an example, if its prior state data set contains the three feature blocks mentioned above (brightness, duration, switching frequency), then the number of feature blocks corresponding to the prior state data set is 3. For the air-conditioning system in the campus, its prior state data set may contain data such as set temperature, operating power, operating mode, and temperature fluctuation range. Among them, the set temperature and temperature fluctuation range can be divided into a feature block according to the data relationship (because they are both related to temperature), the operating power is used as a feature block, and the operating mode is used as a feature block. In this way, the number of feature blocks corresponding to the prior state data set of the air-conditioning system is 3.

[0157] 2. Filtering IoT operation status training logs based on the number of feature blocks 1) Basis for determining the number of settings In this process, there is a set number. This set number is determined based on many factors such as system resource limitations, data processing efficiency requirements, and the applicability of subsequent data analysis algorithms. For example, if the data processing system has limited memory, and the subsequent analysis algorithm will experience performance degradation or a sharp increase in computational complexity when processing data with a large number of feature blocks, then the set number may be set to a relatively small value to ensure the efficiency of data processing and the effectiveness of the algorithm.

[0158] For example, in a smart campus information data analysis system, based on the system's hardware configuration (such as limited memory and processing power) and commonly used data analysis algorithms (such as a matrix-based algorithm whose computational complexity increases exponentially with the number of feature blocks), the set number is determined to be 5. This means that when the number of feature blocks corresponding to a priori state data set exceeds 5, it may have an adverse effect on the system's processing efficiency and the accuracy of the analysis results.

[0159] 2) Filtering operation When the number of feature blocks is greater than the set number, the IoT operation status training log corresponding to the number of feature blocks is filtered out. For example, for a priori state data set of a new smart device containing a large amount of complex sensor data, if the number of feature blocks is 6 after analysis, which is greater than the set number of 5, then the IoT operation status training log related to this new smart device will be filtered out. This is because a priori state data set containing too many feature blocks may lead to problems such as long data processing time, memory overflow, or inaccurate analysis results in the subsequent data analysis process, such as the data analysis process after the IoT operation status training log, the priori state data set, and the associated energy consumption regulation task are synchronously imported into the training sample pool. Through this filtering operation, it can be ensured that the data entering the training sample pool has appropriate complexity and scale, thereby improving the effectiveness and reliability of the entire smart campus information data analysis method.

[0160] Based on the above embodiments, determining the number of feature blocks in the prior state data set helps to gain an in-depth understanding of the data structure, such as distinguishing different state feature blocks in the prior state data sets of the intelligent lighting system and the air-conditioning system. Filtering the IoT operation status training log according to the set number can effectively control the complexity of the data and avoid problems such as low data processing efficiency and memory overflow caused by too many feature blocks, such as filtering out new smart device data with an excessive number of feature blocks in a system with limited resources. The above filtering operation helps to improve the accuracy and reliability of the entire smart campus information data analysis method, ensure the effectiveness of subsequent data analysis, and thus provide a more accurate data basis for tasks such as energy consumption regulation.

[0161] In some other independent embodiments, the first energy consumption regulation optimization strategy for obtaining the historical energy consumption regulation tasks configured by the historical IoT operation status log includes: obtaining the historical energy consumption optimization expected characteristics of the historical energy consumption regulation tasks configured by the historical IoT operation status log, and iteratively downsampling the historical energy consumption optimization expected characteristics to obtain a number of first-stage expected downsampling features; determining the thermal weights of each of the first-stage expected downsampling features in the semantic feature space, and arranging the order of each of the first-stage expected downsampling features according to the ascending rule of each of the thermal weights, and determining the first-stage expected downsampling features that are in the first G as the second-stage expected downsampling features, wherein G is a positive integer; and determining the first energy consumption regulation optimization strategy for the historical energy consumption regulation tasks based on each of the second-stage expected downsampling features.

[0162] In addition, within the scope of smart campus information data analysis, for the historical energy consumption adjustment tasks configured by the historical IoT operation status log, we must first obtain its historical energy consumption optimization expected characteristics. The historical energy consumption optimization expected characteristics are a quantitative representation of the historical energy consumption adjustment task goals, which covers many aspects of information. For example, for the smart lighting system on campus, the historical energy consumption optimization expected characteristics may include reducing the overall energy consumption by a certain percentage (such as 20%) within a specific time period (such as a semester) while ensuring the basic functional requirements of lighting (such as the minimum average brightness of the classroom is 300 lux). For the air-conditioning system, it may be to reduce the operating energy consumption (such as a 30% reduction) on the premise of meeting the indoor comfort level (such as keeping the temperature at 22-26 degrees Celsius).

[0163] Then, the expected features of historical energy consumption optimization are iteratively downsampled. Downsampling is a data processing technique that simplifies the data structure by reducing the amount of data while retaining key information. For example, for the expected features of historical energy consumption optimization of intelligent lighting systems, if their original representation contains multiple data points (such as energy consumption targets and brightness requirements for different classrooms and different time periods), iterative downsampling may be to gradually reduce these data points according to certain rules. For example, the first downsampling merges the energy consumption target data points for each day of a month into weekly energy consumption target data points, and the second downsampling merges the energy consumption target data points of different classrooms according to regions (such as teaching building A, teaching building B, etc.). After multiple iterative downsampling, several first-stage expected downsampled features are obtained.

[0164] Taking the intelligent lighting system as an example, the original historical energy consumption optimization expected features contain 30 data points (representing the energy consumption and brightness information of each classroom every day within a month). After the first downsampling, the daily data is merged into weekly data, and 4 data points may be obtained (representing information for four weeks). After the second downsampling, the data is merged according to the teaching building area, and 2 data points may be obtained (representing the information of two teaching building areas). These data points are the expected downsampling features of the first stage.

[0165] The following technical solution involves determining the thermal weights of the expected downsampling features in the first stage and sorting out the order of precedence.

[0166] 1) Determine thermal weight Determine the thermal weight of each first-stage expected downsampled feature in the semantic feature space. The semantic feature space is an abstract concept space in which each feature has its specific semantic meaning and relative importance. The thermal weight represents the importance or influence of a feature in this semantic feature space. For the first-stage expected downsampled features of the intelligent lighting system, for example, one of the data points represents the weekly energy consumption target of teaching building A. The determination of its thermal weight may be related to factors such as the frequency of use of the teaching building, the size of the area, and the power of the lighting equipment.

[0167] An algorithm based on the analytic hierarchy process (AHP) can be used to determine the thermal weight. For example, for the first stage expected downsampling feature of the weekly energy consumption target of teaching building A, the analytic hierarchy process is used to consider factors such as the frequency of use of teaching building A (for example, 10 hours per day, with a weight of 0.4), the area size (for example, 5,000 square meters, with a weight of 0.3), and the average power of lighting equipment (for example, 100 watts, with a weight of 0.3), and calculate its thermal weight to be 0.3 (the present invention is a specific value obtained according to the calculation rules of AHP).

[0168] 2) Arrange the order of the relationship based on thermal weight According to the ascending rule of each thermal weight, the first-stage expected downsampling features are arranged in order. For example, there are three first-stage expected downsampling features, and their thermal weights are 0.2, 0.3, and 0.4 respectively. According to the ascending rule, the feature with a thermal weight of 0.2 is processed first, then the feature with a thermal weight of 0.3, and finally the feature with a thermal weight of 0.4. The first-stage expected downsampling features that are in the first G (G is a positive integer) are determined as the second-stage expected downsampling features. For example, G=2, then the two first-stage expected downsampling features with thermal weights of 0.2 and 0.3 will be determined as the second-stage expected downsampling features.

[0169] Finally, the first energy consumption regulation optimization strategy can be determined based on the expected downsampling characteristics of the second stage: the first energy consumption regulation optimization strategy of the historical energy consumption regulation task is determined based on each expected downsampling characteristic of the second stage. Taking the intelligent lighting system as an example, if the expected downsampling characteristics of the second stage include the energy consumption target of teaching building A (determined to be a more important feature after the previous analysis) and the brightness requirement of teaching building A (also a more important feature), then the first energy consumption regulation optimization strategy may be to first adjust the control strategy of the lighting equipment of teaching building A.

[0170] For example, if the energy consumption target of teaching building A is to reduce by 15%, it can be achieved by reducing the brightness during non-teaching hours (such as reducing the brightness during the break from 500 lux to 300 lux), adjusting the switching time of lighting equipment (such as turning off some lighting equipment 10 minutes in advance), and other measures. At the same time, according to the brightness requirements of teaching building A, the brightness during teaching hours is guaranteed to be no less than 300 lux, and intelligent dimming technology is used to optimize lighting energy consumption according to the actual lighting conditions of the classroom (such as the natural light near the window is strong, and the brightness of artificial lighting can be appropriately reduced). For the air-conditioning system, if the second stage expected downsampling features include the temperature range that meets the indoor comfort level and the goal of reducing energy consumption, the first energy consumption regulation optimization strategy may be to optimize the temperature setting strategy of the air conditioner, such as dynamically adjusting the temperature setting according to the indoor and outdoor temperature difference and the density of people, and running the air conditioner in energy-saving mode to achieve the goal of reducing energy consumption.

[0171] In this way, by obtaining the expected features of historical energy consumption optimization and performing iterative downsampling, complex energy consumption regulation target data can be simplified, such as simplifying the energy consumption data of each classroom every day in the intelligent lighting system to the energy consumption data of the teaching building area, reducing the amount of data while retaining key information. Determine the thermal weights of the expected downsampling features in the first stage and sort out the order of precedence. The processing order can be reasonably arranged according to the importance of each feature, such as determining the order of importance of energy consumption target features according to various factors of the teaching building. Determine the first energy consumption regulation optimization strategy based on the expected downsampling features in the second stage, and you can formulate a more targeted energy consumption regulation strategy that better meets actual needs, thereby improving the energy consumption management efficiency of the smart campus.

[0172] Further, Figure 2 FIG. 2 is a schematic diagram of a data analysis system 200 provided by the present invention. Figure 2 The data analysis system 200 shown includes a processor 210, which can call and execute a computer program from a memory to implement the method of the present invention.

[0173] Alternatively, if Figure 2 As shown, the data analysis system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method of the present invention.

[0174] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0175] Alternatively, if Figure 2As shown, the data analysis system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0176] Optionally, the data analysis system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with the storage engine in each method of the present invention, which will not be described in detail here for the sake of brevity.

[0177] It should be understood that the processor of the present invention may be an integrated circuit chip having signal processing capabilities.

[0178] It is understood that the memory in the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memory. It should be noted that the memory of the systems and methods described herein is intended to include but is not limited to suitable types of memory.

[0179] Based on the above, a readable storage medium is provided, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the above method are implemented.

[0180] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present invention is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0181] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0182] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the purpose of the present invention and the scope of protection of the present invention, all of which are within the protection of the present invention.

Claims

1. A smart campus information data analysis method, characterized in that: The execution subject of the smart campus information data analysis method is a data analysis system, and the smart campus information data analysis method includes: Obtain the device status monitoring data set corresponding to each initially screened IoT device monitoring behavior event in the campus IoT operation status log, and determine a target status monitoring data set from a number of the device status monitoring data sets, wherein the target status monitoring data set is used to indicate a target IoT device monitoring behavior event with the highest priority among a number of the initially screened IoT device monitoring behavior events; The campus IoT operation status log is respectively subjected to state vector mining based on the target state monitoring data set and each of the device state monitoring data sets to obtain a number of IoT operation state vectors, and the state monitoring attention vectors corresponding to the target state monitoring data set and each of the device state monitoring data sets are extracted; each of the IoT operation state vectors is respectively combined with the corresponding state monitoring attention vector to obtain a number of global state monitoring vectors; Perform operation energy consumption identification on the campus IoT operation status log to obtain a first operation energy consumption quantification vector, combine the first operation energy consumption quantification vector and several of the global status monitoring vectors to obtain energy consumption regulation auxiliary information; perform energy consumption regulation analysis based on the energy consumption regulation auxiliary information, and generate a target energy consumption regulation optimization strategy for the target IoT device monitoring behavior event.

2. The smart campus information data analysis method according to claim 1, characterized in that: The combining of the first operation energy consumption quantization vector and a plurality of the global state monitoring vectors to obtain energy consumption regulation auxiliary information includes: respectively determining data involvement weights of the equipment state monitoring data sets, and combining the global state monitoring vectors corresponding to the equipment state monitoring data sets using the data involvement weights to obtain a first state monitoring combination vector; The first operating energy consumption quantization vector, the first state monitoring combination vector, and the global state monitoring vector corresponding to the target state monitoring data set are combined to obtain energy consumption regulation auxiliary information.

3. The smart campus information data analysis method according to claim 2, characterized in that: The target energy consumption regulation optimization strategy is generated by a first energy consumption regulation decision algorithm, and the first operating energy consumption quantization vector, the first state monitoring combination vector, and the global state monitoring vector corresponding to the target state monitoring data set are combined to obtain energy consumption regulation auxiliary information, including: Acquiring regulation demand information for guiding the first energy consumption regulation decision algorithm to generate an energy consumption regulation optimization strategy; Extract the regulation demand vector of the regulation demand information, combine the first operating energy consumption quantization vector, the regulation demand vector, the first state monitoring combination vector and the global state monitoring vector corresponding to the target state monitoring data set to obtain energy consumption regulation auxiliary information.

4. The smart campus information data analysis method according to claim 1, characterized in that: The campus IoT operation status log is respectively subjected to state vector mining based on the target status monitoring data set and each of the device status monitoring data sets to obtain a plurality of IoT operation status vectors, including: Performing pyramid state vector mining on the campus IoT operation status log to obtain a pyramid operation state vector of the campus IoT operation status log; According to the target state monitoring data set and each of the device state monitoring data sets, feature downsampling is performed on the pyramid operation state vector to obtain a plurality of pyramid downsampling vectors; Each of the pyramid downsampling vectors is interacted with respectively to obtain a plurality of IoT operation state vectors.

5. The smart campus information data analysis method according to claim 4, characterized in that: The step of interacting with each of the pyramid downsampling vectors to obtain a plurality of IoT operation state vectors includes: For any of the pyramid downsampling vectors, respectively adjust the local state vectors of each feature depth in the pyramid downsampling vector to obtain a plurality of implicit state vectors of the same size, and interact the implicit state vectors to obtain an interactive state vector; Perform feature full connection on each of the interaction state vectors to obtain a plurality of IoT operation state vectors.

6. The smart campus information data analysis method according to claim 1, characterized in that: The step of obtaining a device status monitoring data set corresponding to each initially screened IoT device monitoring behavior event in the campus IoT operation status log, and determining a target status monitoring data set from a plurality of the device status monitoring data sets, includes: Obtaining a campus IoT operation status log and index information of the campus IoT operation status log, wherein the campus IoT operation status log includes a plurality of pre-screened IoT device monitoring behavior events, and the index information is used to indicate a target IoT device monitoring behavior event with the highest priority among the plurality of pre-screened IoT device monitoring behavior events; The campus IoT operation status log is disassembled to obtain the device status monitoring data set corresponding to each of the initially screened IoT device monitoring behavior events, and the target status monitoring data set is determined in each of the device status monitoring data sets based on the index information.

7. The smart campus information data analysis method according to claim 6, characterized in that: Determining a target state monitoring data set from each of the device state monitoring data sets according to the index information includes: When the index information is a first type index, determining a target state monitoring data set in each of the device state monitoring data sets according to a relative distribution feature between the first type index and each of the device state monitoring data sets; Alternatively, when the index information is a second type of index, a target state monitoring data set is determined in each of the device state monitoring data sets according to the association between the second type of index and the key value feature of each of the device state monitoring data sets; Alternatively, when the index information is a third type of index, the target state monitoring data set is determined in each of the device state monitoring data sets according to the association between the third type of index and each of the device state monitoring data sets.

8. The smart campus information data analysis method according to claim 1, characterized in that: The target energy consumption regulation optimization strategy is generated by a first energy consumption regulation decision algorithm, and the first energy consumption regulation decision algorithm is debugged by the following steps: Obtain a historical IoT operation status log and a second monitoring data set sample corresponding to a historical IoT device monitoring behavior event in the historical IoT operation status log, and disassemble the historical IoT operation status log to obtain a first monitoring data set sample corresponding to each frequent IoT device behavior event in the historical IoT operation status log, wherein the historical IoT device monitoring behavior event is one IoT device monitoring behavior event among several frequent IoT device behavior events; Extracting the second operation energy consumption quantization vector of the historical IoT operation status log, mining the state vectors of the historical IoT operation status log according to the second monitoring data set sample and each of the first monitoring data set samples, obtaining a plurality of historical operation state vectors, and extracting the historical monitoring attention vectors corresponding to the second monitoring data set sample and each of the first monitoring data set samples; The second operation energy consumption quantization vector, the historical operation state vector, and the historical monitoring attention vector are combined and input into the first energy consumption regulation decision algorithm for energy consumption regulation analysis to generate an energy consumption regulation heat map, wherein the energy consumption regulation heat map is used to determine the energy consumption regulation optimization strategy of the historical IoT device monitoring behavior event; Obtain a first energy consumption regulation optimization strategy for the historical energy consumption regulation task configured by the historical IoT operation status log, determine the algorithm training error based on the energy consumption regulation heat map and the first energy consumption regulation optimization strategy, and debug the first energy consumption regulation decision algorithm based on the algorithm training error, wherein the historical energy consumption regulation task is used to indicate the historical IoT device monitoring behavior events.

9. The smart campus information data analysis method according to claim 8, characterized in that: The historical IoT operation status log, the second monitoring data set sample, and the first energy consumption regulation optimization strategy are all obtained from a training sample pool. Before obtaining the historical IoT operation status log and the second monitoring data set sample corresponding to the historical IoT device monitoring behavior event in the historical IoT operation status log, the smart campus information data analysis method further includes: Obtain a plurality of IoT operation status training logs and a status monitoring training data set corresponding to each IoT operation status training log, and determine the log mining results corresponding to each IoT operation status training log according to each IoT operation status training log and the corresponding status monitoring training data set, wherein the status monitoring training data set is used to characterize the selected IoT device monitoring behavior events obtained by mining the corresponding IoT operation status training log; Acquire a number of preliminary screening energy consumption adjustment tasks, and determine the associated energy consumption adjustment tasks corresponding to each of the IoT operation status training logs in the number of preliminary screening energy consumption adjustment tasks according to the log mining results; Determine, according to each of the IoT operation state training logs and the corresponding state monitoring training data set, a priori state data set of each of the IoT operation state training logs, wherein the priori state data set is used to indicate the corresponding selected IoT device monitoring behavior event; Synchronously importing each of the IoT operation status training logs, the corresponding prior state data set, and the corresponding associated energy consumption adjustment task into the training sample pool, wherein the historical IoT operation status log is obtained from a number of IoT operation status training logs, the second monitoring data set sample is the prior state data set corresponding to the historical IoT operation status log, and the historical energy consumption adjustment task is the associated energy consumption adjustment task configured by the historical IoT operation status log; The determining of the prior state data sets of each IoT operation state training log respectively according to each IoT operation state training log and the corresponding state monitoring training data set includes: Inputting each of the state monitoring training data sets into the second energy consumption regulation decision algorithm to perform energy consumption regulation analysis, and generating an optimization mode label corresponding to each of the IoT operation state training logs; Performing IoT device monitoring behavior event mining according to each IoT operation state training log and the corresponding optimization mode label, generating an event location window label corresponding to each IoT operation state training log, wherein the event location window label is used to indicate the corresponding selected IoT device monitoring behavior event; Inputting each of the IoT operation state training logs and the corresponding event location window labels into a first state data recognition algorithm for processing, and generating a priori state data sets corresponding to each of the IoT operation state training logs; After inputting each of the IoT operation state training logs and the corresponding event location window labels into the first state data recognition algorithm for processing and generating a priori state data set corresponding to each of the IoT operation state training logs, the smart campus information data analysis method further includes: Obtaining the benchmark energy consumption optimization expected features corresponding to each of the associated energy consumption adjustment tasks, wherein the benchmark energy consumption optimization expected features are used to indicate the energy consumption adjustment trend of the benchmark energy consumption adjustment task, and the feature depth of the benchmark energy consumption adjustment task in the semantic feature space is higher than the feature depth of the associated energy consumption adjustment task in the semantic feature space; Inputting each of the IoT operation state training logs and the corresponding benchmark energy consumption optimization expected features into a second state data recognition algorithm for processing, and generating a benchmark state data set corresponding to each of the IoT operation state training logs; Determine the correlation between each of the prior state data sets and the corresponding reference state data set to obtain the target correlation corresponding to each of the prior state data sets; When the target correlation is less than the set correlation, the IoT operation status training log corresponding to the target correlation is filtered out.

10. A data analysis system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 9.

Images