Elevator car fire early warning device and automatic fire extinguishing device

Abstract

The invention discloses an elevator car fire early warning device and an automatic fire extinguishing device, and belongs to the technical field of fire early warning. The elevator car fire early warning device comprises a detection unit and a first control unit, wherein the detection unit comprises a sensor array; the first control unit comprises a fire factor identification model and a fire identification model, and the fire factor identification model obtains K pieces of fire factor data according to information provided by the sensor array; the fire recognition model estimates the probability of fire occurrence according to K pieces of fire factor quantitative data provided by the fire factor recognition model, and the fire recognition model is formed by training a probability model. The fire recognition model of the elevator car fire early warning device has good generalization ability.

Description

Technical Field

[0001] The invention relates to an elevator car fire early warning device and an automatic fire extinguishing device, belonging to the technical field of fire early warning. Background Art

[0002] For example, the Chinese invention patent application with publication number CN115771823A discloses an elevator fire emergency treatment and alarm system, which is applied in a vertical lift elevator, wherein the elevator moves up and down in a corresponding elevator shaft, and the fire emergency treatment and alarm system includes a smoke alarm, a spray mechanism and a liquid storage mechanism, wherein the smoke alarm is arranged on the upper inner wall of the elevator, the spray mechanism is arranged inside the elevator, the liquid storage mechanism includes a liquid storage tank, and the liquid storage tank stores a fire extinguishing agent for extinguishing fire, and a fire extinguishing pipe for conveying the fire extinguishing agent to the spray mechanism is arranged on one side of the liquid storage tank. The design of the smoke alarm can timely sense the occurrence of a fire inside the elevator by using a sensor, thereby transmitting the information to the control system through the smoke alarm, and making a timely alarm through the control system, and controlling the liquid storage mechanism and the spray mechanism through the control system, so as to timely spray and extinguish the fire in the elevator, thereby realizing emergency treatment of the elevator fire, and being able to extinguish the fire and alarm in time.

[0003] However, there is only a smoke alarm in the elevator, which can sense smoke only when a fire occurs. At this time, if the elevator is between safety exits, the fire will spread rapidly and cause harm to people. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present invention provides an elevator car fire alarm and an automatic fire extinguishing device, which obtains K fire elements according to the information provided by the sensor array through a fire element recognition model; the probability of fire occurrence is estimated through the K fire element quantitative data provided by the fire element recognition model, which can prevent fires before they occur, and the fire recognition model is trained through a probability model. The training method of the present invention can improve the generalization ability of the fire recognition model.

[0005] To achieve the above-mentioned object, the present invention provides an elevator car fire alarm, which includes a detection unit and a first control unit. The detection unit includes a sensor array. The first control unit includes a fire element recognition model and a fire recognition model. The fire element recognition model obtains K fire elements according to information provided by the sensor array. The fire recognition model estimates the probability of fire according to the quantitative data of the K fire elements provided by the fire element recognition model. The fire recognition model is trained by a probability model. The process includes: S1-1: Obtain the quantitative data of the series of fire elements in the elevator car in the past and form a data vector , n is a positive integer greater than or equal to 2; K is a positive integer greater than or equal to 2; S1-2: Construct a probability model p of fire occurrence: , In the formula, , Represents the quantitative data about the kth fire element A scalar function of is a parameter; S1-3: Estimation The best parameters : , In the formula, is the likelihood function, whose gradient is: ; , Express The gradient of Express expectations; Express expectations; S1-4: From the data vector Get a set of positive samples from , m is a positive integer less than or equal to n; a set of negative samples is obtained according to the following formula : , In the formula, To set the coefficient, For the transformation relationship; S1-5: Update by the following formula : , In the formula, is the learning coefficient; S1-6: According to the updated Get the fire identification model: .

[0006] To achieve the above-mentioned purpose of the invention, the present invention also provides an automatic fire extinguishing device, which includes a second control unit, a motor driver, an exhaust fan driver and a second communication unit. The second communication unit is used to communicate with the above-mentioned elevator car fire alarm. When the second control unit receives a detection instruction through the second communication unit, the second control unit provides an instruction to the motor driver, and the motor driver controls the motor to drive the emergency air inlet and smoke exhaust port mechanism to open the emergency air inlet and smoke exhaust port, test run for a set time, and then close the emergency air inlet and smoke exhaust port; the second control unit provides an instruction to the exhaust fan driver, and the exhaust fan driver controls the exhaust fan to run, test run for a set time, and then stop running.

[0007] Compared with the prior art, the elevator car fire warning device and automatic fire extinguishing device provided by the present invention have the following beneficial effects:

[0008] The present invention obtains K fire elements according to the information provided by the sensor array through a fire element recognition model; the probability of fire occurrence is estimated through the quantitative data of the K fire elements provided by the fire element recognition model, which can prevent fires before they occur. The fire recognition model is trained through a probability model. The training method of the present invention can improve the generalization ability of the fire recognition model. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 The present invention is a block diagram of the elevator car fire alarm provided by the present invention.

[0010] Figure 2 It is a block diagram of the fire element identification model provided by the present invention.

[0011] Figure 3 It is a block diagram of the composition of the automatic fire extinguishing device provided by the present invention. DETAILED DESCRIPTION

[0012] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0013] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. The meaning of "several" is one or more, unless otherwise clearly and specifically defined.

[0014] Figure 1 1 is a block diagram of the elevator car fire alarm provided by the first embodiment of the present invention. Figure 1 As shown, an elevator car fire alarm provided by the present invention includes a detection unit and a first control unit, wherein the detection unit includes a sensor array; the first control unit includes a fire element recognition model and a fire recognition model, wherein the fire element recognition model obtains K fire elements according to information provided by the sensor array; and the fire recognition model estimates the probability of fire according to the quantitative data of the K fire elements provided by the fire element recognition model. In the present invention, the sensor array includes at least an image sensor.

[0015] In the present invention, the fire identification model is trained by a probability model, and the process includes: S1-1: Obtain the quantitative data of the series of fire elements in the elevator car in the past and form a data vector , n is a positive integer greater than or equal to 2; K is a positive integer greater than or equal to 2; S1-2: Construct a probability model p of fire occurrence: , In the formula, , Represents the quantitative data about the kth fire element A scalar function of is a parameter; S1-3: Estimation The best parameters : , In the formula, is the likelihood function, whose gradient is: ; , Express The gradient of Express expectations; Express expectations; S1-4: From the data vector Get a set of positive samples from , m is a positive integer less than or equal to n; a set of negative samples is obtained according to the following formula : , In the formula, To set the coefficient, For the transformation relationship; S1-5: Update by the following formula : , In the formula, is the learning coefficient; S1-6: According to the updated Get the fire identification model: .

[0016] In the present invention, , In the formula, represents the divergence, express Space for change; Indicates when The distribution of hour The empirical distribution of .

[0017] In the present invention, the elevator car fire alarm also includes an audible and visual alarm, and the first control unit causes the audible and visual alarm to display a light signal corresponding to the probability value of the fire, and to send out a sound signal corresponding to the probability value of the fire.

[0018] In the present invention, the elevator car fire alarm further comprises a first communication unit. When the probability of fire is greater than or equal to a first threshold, the first control unit sends a detection instruction to the automatic fire extinguisher through the first communication unit to detect whether the smoke exhaust module is in good condition; and sends a warning message to the host computer at the same time. The smoke exhaust module comprises a motor driver and an emergency air inlet and smoke exhaust outlet mechanism.

[0019] In the present invention, when the probability of fire is greater than or equal to a second threshold, the first control unit sends fire extinguishing and smoke exhaust instructions to the automatic fire extinguisher, and simultaneously sends elevator car fire extinguishing and smoke exhaust information to the host computer, and the second threshold is greater than the first threshold.

[0020] In the present invention, the elevator car fire alarm also includes a first storage unit, which is used to store control programs and data. The control program includes a program for processing signals obtained by the sensor array and controlling the actions of the sound and light alarm and the first communication unit according to the processing results.

[0021] Figure 2 It is a block diagram of the fire element identification model provided by the present invention, such as Figure 2As shown, the fire element recognition model is trained by a deep convolutional GAN, which includes a discriminator, a generator, a switch and a calculator; during training, the discriminator is connected to the generator through the switch, and the generator generates image training data samples for training the discriminator according to the training data samples, and the discriminator uses the image training data samples generated by the generator for training; during identification, the discriminator is connected to the image sensor through the switch, and outputs the category of combustibles and whether there is a fire source according to the image data provided by the image sensor. The discriminator includes a classifier, which is, for example, a CNN neural network. The classifier can be trained to perform the task of identifying real data samples. The training data for the discriminator is similar to the actual data sample, but can be generated or modified to include the ability to present it as a false sample or an artificial sample. The neural network of the discriminator and the generator can usually be implemented by a multi-layer network, which is composed of multiple processing layers, such as dense processing, batch normalization processing, activation processing, input reshaping processing, Gaussian discarding processing, Gaussian noise processing, two-dimensional convolution and two-dimensional upsampling.

[0022] In the present invention, the generator network can be regarded as a mapping from data space to image space. The discriminator network can be regarded as a mapping from image space to the probability of a real data set. During training, the discriminator tries to get results based on the input. Based on a series of results, both the discriminator and the generator can try to fine-tune parameters to improve their operation. For example, if the discriminator makes a correct prediction, the generator can update its parameters to generate a fake sample and deceive the discriminator. If the discriminator makes an incorrect prediction based on the fake sample, the discriminator can learn from the error and avoid similar errors. Therefore, updating the discriminator and the generator may include a feedback process. The feedback process may be continuous or incremental. The generator and the discriminator may be iteratively executed to optimize data generation and image classification. In the incremental feedback process, the generator may be frozen first and only the discriminator is trained until the discriminator reaches the optimal recognition result. During the period when the generator is frozen, the discriminator may be trained to optimize relative to the state of the generator. Then freeze the discriminator and train the generator. Next, freeze the generator, train the discriminator, and so on.

[0023] In the continuous feedback process, the discriminator can be trained in only one or a few iterations, and the generator can be updated simultaneously with the discriminator.

[0024] If the distribution of the generated simulated dataset can be exactly matched with the distribution of the real dataset, the discriminator is maximally confused and cannot distinguish real samples from fake ones.

[0025] The present invention can determine whether the stopping condition is met by evaluating whether the gradient is large enough. The generator of the present invention updates the network parameters through the back propagation algorithm, and each layer of the generator has one or more gradients.

[0026] Specifically, the training of the deep convolutional neural network GAN includes the following process: The fire element recognition model is trained by a deep convolutional neural network GAN, which includes a generator, a switcher, a discriminator and a calculator. The training process includes: S2-1: The switch is activated so that the generator is connected to the discriminator through the switch; S2-2: enabling the generator to generate correct image training samples including combustibles and fire sources according to the training data samples; S2-3: The discriminator performs identification based on the image training samples including combustibles and fire sources to generate identification results; S2-4: The calculator calculates the loss function based on the current identification result and the training data sample, and determines that if the current loss function has not reached the minimum, step S2-9 is executed, if the current loss function reaches the minimum, step S2-5 is executed; S2-5: Make the generator change its network parameters and generate fake image samples based on the training data samples; S2-6: The discriminator performs identification based on the fake image sample to generate an identification result; S2-7: The calculator calculates the loss function based on the current identification result and the training data sample, and determines that if the current loss function is the largest, step S2-8 is executed, if the current loss function does not reach the maximum, step S2-9 is executed; S2-8: End the training and make the switch act so that the image sensor is connected to the discriminator through the switch; S2-9: The calculator calculates the gradient of the loss function with respect to the current network parameters of the discriminator, uses the gradient to update the current network parameters of the discriminator by gradient boosting, and then returns to step S2-2.

[0027] Figure 3 It is a block diagram of the composition of the automatic fire extinguishing device provided by the present invention, such as Figure 3As shown, the automatic fire extinguishing device provided by the present invention includes a second control unit, a smoke exhaust module and a second communication unit, the smoke exhaust module includes a motor, a motor driver, an exhaust fan driver, an exhaust fan and an emergency air inlet and smoke exhaust port mechanism; the second communication unit is used to communicate with the above-mentioned elevator car fire alarm device, when the second control unit receives the detection instruction through the second communication unit, the second control unit provides an instruction to the motor driver, the motor driver controls the motor to drive the emergency air inlet and smoke exhaust port mechanism to open the emergency air inlet and smoke exhaust port, and then close the emergency air inlet and smoke exhaust port; the second control unit provides an instruction to the exhaust fan driver, the exhaust fan driver controls the exhaust fan to run, and then stops running. Preferably, the fire extinguishing and smoke exhaust instructions are sent to the second control unit, the second control unit makes the ignition unit ignite the potassium ion fire extinguishing agent particles in the fire extinguishing agent cartridge according to the fire extinguishing instruction, sends an instruction according to the smoke exhaust instruction servo mechanism, opens the emergency air inlet and exhaust port of the elevator car, and provides an instruction to the exhaust fan, and the exhaust fan causes the smoke in the elevator car to be discharged from the exhaust port.

[0028] The automatic fire extinguishing device also includes an ignition unit and a fire extinguishing agent cartridge. When the second control unit receives a fire extinguishing and smoke exhaust instruction through the second communication unit, the second control unit causes the ignition unit to ignite the potassium ion fire extinguishing agent particles in the fire extinguishing agent cartridge according to the fire extinguishing instruction, and provides instructions to the motor driver according to the smoke exhaust instruction. The motor driver controls the motor to drive the emergency air inlet and smoke exhaust port mechanism to open the emergency air inlet and smoke exhaust port until the fire is extinguished; the second control unit provides instructions to the exhaust fan driver, and the exhaust fan driver controls the exhaust fan to operate until the fire is extinguished.

[0029] In the present invention, the automatic fire extinguishing device also includes an input unit, which is used for users to input fire extinguishing and smoke exhaust commands. The input unit is installed on the inner wall of the elevator car to facilitate user operation in an emergency. A protective shell is provided outside the input unit to prevent misoperation.

[0030] In the present invention, the fire extinguishing agent is a potassium ion fire extinguishing agent. The potassium ion fire extinguishing agent formula uses food-grade raw materials and is safe and harmless to the human body. Its fire extinguishing principle is: the ionic potassium oxide released by the combustion of the potassium ion fire extinguishing agent has a universal uniform effect, achieving full flooding coverage of the entire protection area, combining with the flame free radicals when contacting the fire source, repeatedly cutting off the flame free radical reaction chain, forming stable non-combustible products, and eventually causing the flame chain to be interrupted, achieving the purpose of fire extinguishing without consuming the ambient oxygen content. The gas released by the potassium ion fire extinguishing agent can be suspended in the protected room or shell for at least 30 minutes, effectively and continuously suppressing the expansion or rekindling of the fire.

[0031] The safe and non-toxic potassium ion fire extinguisher can deal with five major types of fires. It has a storage temperature of -60~+160℃ and is not afraid of bumps and vibrations. It is stored in a solid state and will only activate when the temperature exceeds 410℃. When ignited, the solid state will be converted into gas and released.

[0032] In the present invention, the automatic fire extinguishing device further includes a second storage unit for storing control programs and data, wherein the control programs include control programs for the motor driver, the exhaust fan driver, the ignition unit, and the second communication unit. The control programs are also used to process instructions input by the input unit, and generate control programs for the motor driver, the exhaust fan driver, the ignition unit, and the second communication unit according to the input instructions.

[0033] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An elevator car fire alarm, characterized in that: The invention comprises a detection unit and a first control unit, wherein the detection unit comprises a sensor array; the first control unit comprises a fire element recognition model and a fire recognition model, wherein the fire element recognition model obtains K fire elements according to information provided by the sensor array; the fire recognition model estimates the probability of fire according to the K fire element quantified data provided by the fire element recognition model, and the fire recognition model is trained by a probability model, and the process comprises: S1-1: Obtain the quantitative data of the series of fire elements in the elevator car in the past and form a data vector , n is a positive integer greater than or equal to 2; K is a positive integer greater than or equal to 2; S1-2: Construct a probability model p of fire occurrence: ， In the formula, , Represents the quantitative data about the kth fire element A scalar function of is a parameter; S1-3: Estimation The best parameters : ， In the formula, is the likelihood function, whose gradient is: ； , Express The gradient of Express expectations; Express expectations; S1-4: From the data vector Get a set of positive samples from , m is a positive integer less than or equal to n; a set of negative samples is obtained according to the following formula : ， In the formula, To set the coefficient, For the transformation relationship; S1-5: Update by the following formula : , In the formula, is the learning coefficient; S1-6: According to the updated Get the fire identification model: 。 2. The elevator car fire alarm according to claim 1, characterized in that: , In the formula, represents the divergence, express Space for change; Indicates when The distribution of hour The empirical distribution of .

3. The elevator car fire alarm according to claim 1, characterized in that: The fire element recognition model is trained by a deep convolutional neural network GAN, which includes a generator, a switcher, a discriminator and a calculator. The training process includes: S2-1: The switch is activated so that the generator is connected to the discriminator through the switch; S2-2: enabling the generator to generate correct image training samples including combustibles and fire sources according to the training data samples; S2-3: The discriminator performs identification based on the image training samples including combustibles and fire sources to generate identification results; S2-4: The calculator calculates the loss function based on the current identification result and the training data sample, and determines that if the current loss function has not reached the minimum, step S2-9 is executed, if the current loss function reaches the minimum, step S2-5 is executed; S2-5: Make the generator change its network parameters and generate fake image samples based on the training data samples; S2-6: The discriminator performs identification based on the fake image sample to generate an identification result; S2-7: The calculator calculates the loss function based on the current identification result and the training data sample, and determines that if the current loss function is the minimum, step S2-8 is executed, if the current loss function does not reach the minimum, step S2-9 is executed; S2-8: End the training and make the switch act so that the image sensor is connected to the discriminator through the switch; S2-9: The calculator calculates the gradient of the loss function with respect to the current network parameters of the discriminator, uses the gradient to update the current network parameters of the discriminator by gradient boosting, and then returns to step S2-2.

4. The elevator car fire alarm according to claim 3, characterized in that: It also includes an audible and visual alarm. The first control unit causes the audible and visual alarm to display a light signal corresponding to the probability value of a fire, and to send out a sound signal corresponding to the probability value of a fire.

5. The elevator car fire alarm according to any one of claims 1 to 4, characterized in that: It also includes a first communication unit. When the probability of fire is greater than or equal to a first threshold, the first control unit sends a detection instruction to the automatic fire extinguisher through the first communication unit to detect whether the smoke exhaust module is in good condition; and sends an early warning message to the host computer at the same time.

6. The elevator car fire alarm according to claim 5, characterized in that: When the probability of fire is greater than or equal to a second threshold, the first control unit sends fire extinguishing and smoke exhaust instructions to the automatic fire extinguisher, and sends elevator car fire extinguishing and smoke exhaust information to the host computer, and the second threshold is greater than the first threshold.

7. An automatic fire extinguishing device, characterized in that: It includes a second control unit, a smoke exhaust module and a second communication unit. The smoke exhaust module includes a motor, a motor driver, an exhaust fan driver, an exhaust fan and an emergency air inlet and smoke exhaust port mechanism. The second communication unit is used to communicate with the elevator car fire alarm according to claim 6. When the second control unit receives a detection instruction through the second communication unit, the second control unit provides instructions to the motor driver, and the motor driver controls the motor to drive the emergency air inlet and smoke exhaust port mechanism to open the emergency air inlet and smoke exhaust port, test run the set time, and then close the emergency air inlet and smoke exhaust port. The second control unit provides instructions to the exhaust fan driver, and the exhaust fan driver controls the exhaust fan to run, test run the set time, and then stop running.

8. The automatic fire extinguishing device according to claim 7, characterized in that: It also includes an ignition unit and a fire extinguishing agent cartridge. When the second control unit receives a fire extinguishing and smoke exhaust instruction through the second communication unit, the second control makes the ignition unit ignite the potassium ion fire extinguishing agent particles in the fire extinguishing agent cartridge according to the fire extinguishing instruction, and provides instructions to the motor driver according to the smoke exhaust instruction. The motor driver controls the motor to drive the emergency air inlet and smoke exhaust port mechanism to open the emergency air inlet and smoke exhaust port; the second control unit provides instructions to the exhaust fan driver, and the exhaust fan driver controls the operation of the exhaust fan.

9. The automatic fire extinguishing device according to claim 8, characterized in that: It also includes an input unit, which is used for a user to input fire extinguishing and smoke exhausting instructions.

10. The automatic fire extinguishing device according to claim 9, characterized in that: The fire extinguishing agent is potassium ion fire extinguishing agent.

Images