AI elevator stopping adjusting system and method

Abstract

The invention provides an AI elevator stopping adjusting system and method. The AI elevator stopping adjusting system comprises a daily taking rule analysis module, a travel information collection and analysis module, an emergency event collection and analysis module, a serious health problem monitoring and analysis module and a stopping adjusting module. The daily taking rule analysis module is used for analyzing the daily elevator taking rules of the users in the building beyond holidays and festivals; the travel information acquisition and analysis module is used for acquiring and analyzing temporary travel information of users in the building; the emergency event acquisition and analysis module is used for acquiring and analyzing emergency event information in the building; the serious health problem monitoring and analyzing module is used for monitoring and analyzing whether the user has a serious health problem; the stopping adjusting module dynamically adjusts elevator stopping floors according to user requirements in a building; the system and the method can help users in the building save more time.

Description

Technical Field

[0001] The present invention relates to the field of software information technology, and in particular to an AI elevator parking adjustment system and method. Background Art

[0002] The traditional way of adjusting the elevator stop is that the user on each floor presses the up or down direction button, and presses the target floor button after entering the elevator. When the elevator is not in use, it will stop at the last stop floor. This simple stop method fails to stop in advance according to the needs of users on each floor. It is necessary to design a more intelligent stop adjustment system and a more reasonable stop adjustment method to help users in the building save more time. Summary of the invention

[0003] In order to help users in a building save more time, the present invention designs an AI elevator stop adjustment system and method.

[0004] The technical solution adopted by the present invention is an AI elevator parking adjustment system: It includes daily riding pattern analysis module, travel information collection and analysis module, emergency event collection and analysis module, serious health problem monitoring and analysis module, and stop adjustment module.

[0005] The daily riding pattern analysis module is used to analyze the daily riding pattern of users in the building outside holidays, specifically in the following ways: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time point when the elevator door starts to open, and the stop time is accurate to seconds.

[0006] A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:00 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; for example, if the recorded elevator stop time is 15:30:28, the conversion method is 15×3600+30×60+28=55828 seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This situation data is marked as P-type data, and the corresponding number of repetitions is R.

[0007] A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions.

[0008] A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change in H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model.

[0009] A5, the final straight line k1×x+k2×y+k3×z+b=0 represents the rules of the daily stop time points of the elevator outside holidays, the corresponding stop floors, and the corresponding personnel flow.

[0010] The travel information collection and analysis module is used to collect and analyze temporary travel information of users in the building, specifically in the following ways: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator.

[0011] The emergency event collection and analysis module is used to collect and analyze the emergency event information in the building, specifically in the following ways: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor sounds an alarm, it is determined that the gas leakage on that floor exceeds the set safety threshold.

[0012] The serious health problem monitoring and analysis module is used to monitor and analyze whether the user has serious health problems, specifically in the following ways: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has serious health problems and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, it is determined that the user is in urgent need of using the elevator.

[0013] The stop adjustment module dynamically adjusts the elevator stop floors according to the needs of users in the building, specifically in the following ways: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

[0014] An AI elevator stop adjustment method: Step 1: Analyze the daily elevator riding patterns of building users outside holidays. The specific method is as follows: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time point when the elevator door starts to open, and the stop time is accurate to seconds.

[0015] A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:00 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; for example, if the recorded elevator stop time is 15:30:28, the conversion method is 15×3600+30×60+28=55828 seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This situation data is marked as P-type data, and the corresponding number of repetitions is R.

[0016] A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions.

[0017] A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change in H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model.

[0018] A5, the final straight line k1×x+k2×y+k3×z+b=0 represents the rules of the daily stop time points of the elevator outside holidays, the corresponding stop floors, and the corresponding personnel flow.

[0019] Step 2: Collect and analyze temporary travel information of users in the building. The specific method is as follows: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator.

[0020] Step 3: Collect and analyze the emergency event information in the building. The specific method is as follows: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor sounds an alarm, it is determined that the gas leakage on that floor exceeds the set safety threshold.

[0021] Step 4: Monitor and analyze whether the user has serious health problems. The specific methods are as follows: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has serious health problems and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, it is determined that the user is in urgent need of using the elevator.

[0022] Step 5: Dynamically adjust the elevator stop floors according to the needs of users in the building. The specific method is as follows: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

[0023] The principle of the present invention is as follows: smoke sensors are used to detect smoke particles in the air for fire alarm; temperature sensors are used to monitor temperature changes to help detect fire or overheating; gas leak sensors are used to detect leakage of specific gases, such as natural gas, liquefied gas, etc., to prevent explosion and poisoning.

[0024] Logistic regression is a machine learning algorithm used to predict binary classification problems. Its basic idea is to use linear combination features for classification. The output of logistic regression is a probability value, which is converted by the logistic function Sigmoid.

[0025] The AI ​​elevator parking adjustment system and method of the present invention have the following advantages: (1) Reasonable analysis of the daily elevator riding patterns of building users outside holidays, with novel ideas and significant creativity; (2) Using AI algorithms to monitor and analyze whether users have serious health problems can analyze these complex data faster and more efficiently; (3) The proposed method dynamically adjusts the elevator stops by comprehensively considering the user's physical health, temporary travel needs, elevator stop patterns outside holidays, and emergency situations on each floor. The idea is ingenious.

[0026] Other features and advantages of the present invention will be set forth in the following description or may be learned by practicing the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0028] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0029] The following will further describe in detail an AI elevator stop adjustment system and method of the present invention in conjunction with the accompanying drawings and embodiments.

[0030] The technical solution adopted by the present invention is an AI elevator parking adjustment system: It includes daily riding pattern analysis module, travel information collection and analysis module, emergency event collection and analysis module, serious health problem monitoring and analysis module, and stop adjustment module.

[0031] The daily riding pattern analysis module is used to analyze the daily riding pattern of users in the building outside holidays, specifically in the following ways: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time point when the elevator door starts to open, and the stop time is accurate to seconds.

[0032] A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:00 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; for example, if the recorded elevator stop time is 15:30:28, the conversion method is 15×3600+30×60+28=55828 seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This situation data is marked as P-type data, and the corresponding number of repetitions is R.

[0033] A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions.

[0034] A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change in H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model.

[0035] A5, the final straight line k1×x+k2×y+k3×z+b=0 represents the rules of the daily stop time points of the elevator outside holidays, the corresponding stop floors, and the corresponding personnel flow.

[0036] The travel information collection and analysis module is used to collect and analyze temporary travel information of users in the building, specifically in the following ways: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator.

[0037] The emergency event collection and analysis module is used to collect and analyze the emergency event information in the building, specifically in the following ways: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor sounds an alarm, it is determined that the gas leakage on that floor exceeds the set safety threshold.

[0038] The serious health problem monitoring and analysis module is used to monitor and analyze whether the user has serious health problems, specifically in the following ways: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has serious health problems and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, it is determined that the user is in urgent need of using the elevator.

[0039] The stop adjustment module dynamically adjusts the elevator stop floors according to the needs of users in the building, specifically in the following ways: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

[0040] like Figure 1 As shown, an AI elevator stop adjustment method: Step 1: Analyze the daily elevator riding patterns of building users outside holidays. The specific method is as follows: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time point when the elevator door starts to open, and the stop time is accurate to seconds.

[0041] A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:00 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; for example, if the recorded elevator stop time is 15:30:28, the conversion method is 15×3600+30×60+28=55828 seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This situation data is marked as P-type data, and the corresponding number of repetitions is R.

[0042] A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions.

[0043] A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change in H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model.

[0044] A5, the final straight line k1×x+k2×y+k3×z+b=0 represents the rules of the daily stop time points of the elevator outside holidays, the corresponding stop floors, and the corresponding personnel flow.

[0045] Step 2: Collect and analyze temporary travel information of users in the building. The specific method is as follows: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator.

[0046] Step 3: Collect and analyze the emergency event information in the building. The specific method is as follows: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor sounds an alarm, it is determined that the gas leakage on that floor exceeds the set safety threshold.

[0047] Step 4: Monitor and analyze whether the user has serious health problems. The specific methods are as follows: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has serious health problems and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, it is determined that the user is in urgent need of using the elevator.

[0048] Step 5: Dynamically adjust the elevator stop floors according to the needs of users in the building. The specific method is as follows: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

[0049] Smoke sensors are used to detect smoke particles in the air for fire alarms; temperature sensors are used to monitor temperature changes to help detect fire or overheating; gas leak sensors are used to detect leaks of specific gases, such as natural gas, liquefied gas, etc., to prevent explosions and poisoning.

[0050] Logistic regression is a machine learning algorithm in AI algorithms. It is used to predict binary classification problems. Its basic idea is to use linear combination features for classification. The output of logistic regression is a probability value, which is converted by the logistic function Sigmoid.

[0051] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. An AI elevator parking adjustment system, characterized in that: It includes daily riding pattern analysis module, travel information collection and analysis module, emergency event collection and analysis module, serious health problem monitoring and analysis module, and stop adjustment module; The daily riding pattern analysis module is used to analyze the daily riding pattern of users in the building outside holidays, specifically in the following ways: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time when the elevator door starts to open, and the stop time is accurate to seconds; A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:0:0 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are all the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This kind of data is marked as P-type data, and the corresponding number of repetitions is R; A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions; A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change of H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model. A5, the resulting straight line k1×x+k2×y+k3×z+b=0 represents the regularity of the elevator’s daily stop time, corresponding stop floors, and corresponding personnel flow except for holidays; The travel information collection and analysis module is used to collect and analyze temporary travel information of users in the building, specifically in the following ways: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator; The emergency event collection and analysis module is used to collect and analyze the emergency event information in the building, specifically in the following ways: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor alarms, it is determined that the gas leakage on the floor exceeds the set safety threshold; The serious health problem monitoring and analysis module is used to monitor and analyze whether the user has serious health problems, specifically in the following ways: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has a serious health problem and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, the user is judged to be in urgent need of using the elevator; The stop adjustment module dynamically adjusts the elevator stop floors according to the needs of users in the building, specifically in the following ways: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

2. An AI elevator parking adjustment method, characterized in that: Step 1: Analyze the daily elevator riding patterns of building users outside holidays. The specific method is as follows: A1, in 24-hour format, records the elevator stop time, corresponding stop floor, corresponding number of people entering the elevator, and corresponding number of people leaving the elevator in the previous 30 days excluding holidays; The stop time is the time when the elevator door starts to open, and the stop time is accurate to seconds; A2, establish a three-dimensional rectangular coordinate system, where the X-axis is the time point when the elevator stops, the Y-axis is the floor where the elevator stops, and the Z-axis is the sum of the number of people entering the elevator and the number of people leaving the elevator; Taking 0:0:0:0 as the starting time point, convert all recorded elevator stop times in the first 30 days excluding holidays into seconds; There is a situation where the stop time point, stop floor, the sum of the number of people entering the elevator and the number of people leaving the elevator are all the same, that is, the X-axis data, Y-axis data, and Z-axis data are all the same. This kind of data is marked as P-type data, and the corresponding number of repetitions is R; A3, create a straight line k1×x+k2×y+k3×z+b=0 in three-dimensional space, where k1, k2, k3 are slopes and b is a constant; Then the distance from any point (x0, y0, z0) in three-dimensional space to this line is L=|k1×x0+k2×y0+k3×z0+b|÷sqrt(k1 2 +k2 2 +k3 2 ), where sqrt() represents the arithmetic square root; Let H = ΣL i +Σ(L j ×R j ), where L i is the distance from the non-P type data to the created straight line, L j is the distance from the P-type data to the created straight line, R j For L j Number of repetitions; A4, use the stochastic gradient descent method to find the minimum value of H, and the corresponding k1, k2, k3, and b are the parameters of the elevator riding regularity model: ① Initial parameter setting, randomly select the initial values ​​of k1, k2, k3, and b, set the learning rate t=0.01, that is, the update amplitude of k1, k2, k3, and b is 0.01 each time; set the convergence threshold; ② Calculate the initial value of H. According to the initial values ​​of k1, k2, k3, and b, calculate the initial value of the objective function H. For each parameter k1, k2, k3, b, calculate the partial derivatives of H with respect to these parameters, namely ‌grad(k1), grad(k2), grad(k3), grad(b). These partial derivatives measure the impact of small changes in parameters on the objective function H; ③Parameter iteration, The new k1 is equal to the old k1 minus [t×grad(k1)]; The new k2 is equal to the old k2 minus [t×grad(k2)]; The new k3 is equal to the old k3 minus [t×grad(k3)]; The new b is equal to the old b minus [t×grad(b)]; Iterate each parameter continuously until the change of H is less than the convergence threshold and then stop iterating. At this time, k1, k2, k3, and b are the parameters of the elevator riding regularity model. A5, the resulting straight line k1×x+k2×y+k3×z+b=0 represents the regularity of the elevator’s daily stop time, corresponding stop floors, and corresponding personnel flow except for holidays; Step 2: Collect and analyze temporary travel information of users in the building. The specific method is as follows: B1, collect the estimated travel time entered by the user in the mobile phone application, which includes email, memo, social software, navigation software, taxi software, travel software, and itinerary management software; B2, collecting the horizontal coordinates of the elevator and the real-time horizontal coordinates of the user. The horizontal coordinates of the elevator are the horizontal coordinates of all elevator entrances that the elevator passes through; B3, when the estimated travel time is less than or equal to 5 minutes, the distance between the user and the elevator is calculated every 30 seconds. Let the most recently calculated distance between the user and the elevator be S n , the distance between the user and the elevator calculated last time is S n-1 , when S n -S n-1 <0, it is determined that the user needs to travel temporarily and walks to the elevator; Step 3: Collect and analyze the emergency event information in the building. The specific method is as follows: C1, record the floor where each smoke detector, temperature sensor, and gas leak sensor is located; C2, when a smoke detector sounds an alarm, it is determined that the smoke particle concentration on the floor exceeds the set safety threshold; When a temperature sensor alarms, it is determined that the temperature of the floor exceeds the set safety threshold; When a gas leakage sensor alarms, it is determined that the gas leakage on the floor exceeds the set safety threshold; Step 4: Monitor and analyze whether the user has serious health problems. The specific methods are as follows: D1, the user's mobile phone obtains heart rate V1, blood oxygen saturation V2, and body temperature V3 in real time from the health monitoring bracelet worn by the user; D2, construct the feature matrix V=[V1,V2,V3] and the label M, where M is the label corresponding to the sample, M=1 indicates serious health problems, and M=0 indicates no serious health problems; D3, divide the collected data into 80% training set and 20% test set; D4, use the training set to fit the logistic regression model, the model is represented by the Sigmoid function as Sig(V)=[1+e -(U·V) ] -1 ; Where U is the column vector containing the model parameters; Sig() is the logical function Sigmoid function, and the output is the probability value of whether the user has serious health problems, ranging from 0 to 1, excluding 0 and 1; U·V represents the dot product of the parameter column vector U and the feature row vector V, and the result is a scalar; When Sig(V)≥0.5, M=1; when Sig(V)<0.5, M=0; Optimize the parameter U through training so that the output of the Sigmoid function is as close as possible to the corresponding label M; D5, use the test set to evaluate the model, and stop training when the accuracy of the model prediction reaches the set threshold; D6, using the trained model to predict whether the user has serious health problems based on the newly collected user's heart rate, blood oxygen saturation, and body temperature data; D7: When the user has a serious health problem and the distance between the user and the elevator entrance is less than 5 meters for more than 10 seconds, the user is judged to be in urgent need of using the elevator; Step 5: Dynamically adjust the elevator stop floors according to the needs of users in the building. The specific method is as follows: E1, collects the user's floor information; E2: When the situation in C2 exists, that is, when the smoke particle concentration on a certain floor exceeds the set safety threshold, or the temperature exceeds the set safety threshold, or the gas leakage exceeds the set safety threshold, the elevator will not stop at this floor; E3, the priority of elevator stops is as follows: ① The elevator will stop at the floors in D7 where users need to use the elevator urgently; ② When the situation in D7 does not exist, the user adjusts the elevator stop by pressing the direction key outside the elevator and pressing the stop floor in the elevator. On this basis, the elevator stops outside holidays according to the daily elevator riding rules of users in the building outside holidays in A5, and stops at this floor for 5 seconds when the user needs to travel temporarily and walks towards the elevator in B3.

Images